Musculoskeletal disease and injury, now and in the future. Part 1: Fractures and fatalities


  • P. D. CLEGG

    Corresponding author
    1. Department of Musculoskeletal Biology, Faculty of Health and Life Sciences, University of Liverpool, Leahurst Campus, Neston, Cheshire, UK
    Search for more papers by this author



Orthopaedic injury is common in the racing Thoroughbred and is a cause of both economic loss to the industry, as well as being an obvious welfare issue. Fatal injuries sustained during racing are an obvious major concern and it is important that all is done to reduce their frequency. As a consequence of the importance of these causes of horse morbidity and mortality, the Horserace Betting Levy Board in the UK has invested considerable resources into research in recent years in this area in an attempt to decrease the frequency of such injuries. This review summarises current knowledge relating to epidemiological investigations relating to fractures, falling and fatalities in the racing Thoroughbred. Studies identify the importance of track surface conditions, the importance of pre-existing orthopaedic pathology, as well as issues relating to the horse's past training and racing experience as important determinants of both injury and death. Such findings can now be used to develop interventions to reduce racehorse injury and death for the benefit of both the industry and the horse.

Historical background

Major musculoskeletal injury and, in particular fracture, have always been an occupational hazard for the Thoroughbred racehorse and are a significant cause of both morbidity and mortality within the racing industry. Notwithstanding the economic loss such injuries lead to in the industry, such injuries are often emotional and unpleasant to all concerned with the horse and are an undoubted welfare concern. Initiatives to address such injuries occurring in the racecourse were first proposed in the late 1960s, where a proposal at a Horserace Betting Levy Board (HBLB) conference led to the establishment of the Racecourse Equine Fatal Accident Scheme in 1970. Data from this scheme identified that 63% of fatalities on UK racecourses were due to fractures in the initial first two and a half years of the scheme, although it was recognised that within the scheme there was considerable under reporting of injuries (Vaughan and Mason 1975). A number of key studies of Thoroughbred racehorse wastage performed in the 1980s further identified that musculoskeletal disorders were a major reason for horses not training or racing (Jeffcott et al. 1982; Rossdale et al. 1985). McKee (1995) reported on data on catastrophic injuries that occurred during racing and were reported to the UK Jockey Club between 1987 and 1993 and again identified limb fractures to be the most common cause for euthanasia on the racecourse (McKee 1995). Data from 3 years of Jockey Club racecourse injury and fatality surveillance data was analysed for 3 seasons (1996–1998) and detailed morbidity and mortality data from 222,993 racing starts. The majority of the clinical events identified (1937/2358) related to the musculoskeletal system. Euthanasia or death occurred in 657 horses (0.29% of starts). The risk of injury during racing was found to increase with an increasing age of horse and softer racing surfaces were found to be associated with fewer injuries or fatalities compared with racing on firmer surfaces (Williams et al. 2001). As a consequence of the frequency and severity of such injuries, the HBLB has for many years taken a strong lead in funding research studies, which have the potential to lead to a greater understanding of the aetiology and risks of racing injuries. The ultimate aim of such research is to identify interventions that may reduce the frequency of such events ultimately for the benefit of all involved in racing.

Overseas studies

Internationally, away from the UK, there was a growing interest in investigating aspects of racing morbidity and mortality through the last decade of the last century. In particular, there was increasing use of analytical epidemiological techniques to determine risk factors for specific injuries or death in order to move away from descriptional studies, ultimately to develop preventative measure to reduce the risk of such injuries. Whilst these studies do provide valuable insights into the frequency and causes of racing injuries, there is often conflicting conclusions between differing studies to what are the specific risk factors for injury (Mohammed et al. 1991; Johnson et al. 1994; Peloso et al. 1994; Estberg et al. 1995, 1996a,b, 1998a,b; Kane et al. 1996, 1998; Bailey et al. 1997, 1998; Cohen et al. 1997, 1999a,b, 2000; Carrier et al. 1998; Boden et al. 2006, 2007a,b; Anthenill et al. 2007). Many of these studies demonstrate conflicting data, which highlights the uncertainty associated with this area of research (Parkin 2002). For example, Estberg and coworkers demonstrating that accumulation of greater distances of high speed exercise over a 1–2 month period were associated with a higher risk of fatal skeletal injury during racing (Estberg et al. 1995, 1996a). In contrast, Cohen and coworkers reported that injured horses had undergone significantly less cumulative high speed exercise than control horses during a 1–2 month period prior to the race in which the injury happened (Cohen et al. 2000). Such conflict may be due to differences in studies populations under different racing jurisdictions or could be as a result of differences in case definitions and study design (Cohen et al. 2000; Parkin 2002).

Falling and racing injuries

Falling during National Hunt racing has always been considered a significant cause of both injury and mortality and a series of studies have determined risk factors associated with injury and, most specifically, falling within National Hunt racing. A prospective cohort study investigated aspects of racing practice and injuries in 2879 starts by 2216 horses on 6 UK racecourses. This identified 83 injuries or medical events (28.8/1000 starts) of which 18 were fatal (6.3/1000 starts). The most common injuries were tendon/ligament injuries and lacerations/wounds. Risk factors for traumatic injuries included longer races, faster races and horses having a long toe/low heel conformation (Pinchbeck et al. 2004d). Specifically relating to falling, analysis of horse falls on all UK racecourses in 1999 identified a falling risk of 6.0/100 starts in steeplechasing, 2.1/100 starts in hurdle races and 3.7% of steeplechase fallers died compared with 7.1% of hurdle fallers. In all races, 38% of fatalities in National Hunt racing occurred subsequent to a fall. In hurdling, the frequency of falls was 1/447 jumping efforts and in steeplechasing the frequency was 1/254 jumping efforts (Pinchbeck et al. 2002).

In steeplechasing the risk of falling increased with increasing race distance and lack of previous racing experience by the horse. Furthermore, the greater number of times a steeplechaser had run on a particular course, the lower the risk of falling. In hurdling, falling was more likely in horses that started hurdling at an older age or were ridden by less experienced jockeys. The importance of race experience was evident in hurdling, where horses competing in their first hurdle race were at almost 5 times the risk of falling compared with those that had hurdled before. This finding relating to horses racing experience was shown to be significant in analysis of reasons why horses may not complete the Grand National at Aintree, which demonstrated that horses that had never raced previously in any race over the National fences were twice as likely (hazard ratio, 2.0; 95% CI, 1.52–2.63) not to finish the race as horses that had previous experience of the course without a fall (Pinchbeck et al. 2003, 2004a,b; Proudman et al. 2004). Using video analysis of fallers during racing, it was determined that most falls were due to mistakes at fences and not due to identifiable injury prior to jumping the fence. Horses that were being whipped and progressing through the race were at greater than 7 times the risk of falling compared with horses not being whipped and which had no change in position or lost position through the field (Pinchbeck et al. 2004c).

A study investigating Thoroughbred fatality in jump starts in Victoria, Australia confirmed differing fatality rates between hurdle and steeplechase races and identified an adverse effect of prolonged prior flat racing careers on the risk of fatality in jump starts (Boden et al. 2007b). Interestingly, in Australia, epidemiological studies have been performed to specifically detail risks associated with race riding by jockeys (Hitchens et al. 2009, 2010, 2011).

Fractures and racing

Two large consecutive epidemiological studies were performed at the University of Liverpool between 1998–2003 investigating risk factors for fatal distal limb fractures during racing. The methodologies of these studies involved identifying all fatal distal limb fractures that occurred distal to the tarsus and carpus on UK racecourses, collecting and performing a post mortem on the limb (and the contralateral [opposite] limb) and then using classical epidemiological (case–control methods) identifying the differences between either the horses that sustained fractures and other horses not affected by fractures (horse-level risk factors), as well as defining differences in races and racecourses at which fractures occurred in comparison to those which did not have fractures (race- and course-level risk factors).

This study identified that the risk of fracture varied depending on the type of racing. Flat turf racing was the safest (0.4 fatal fractures/1000 starts) whilst National Hunt flat racing was associated with the highest risk (2.2 fatal fractures/1000 starts). The type of fracture sustained did vary with the type of racing. Lateral condylar fractures of the third metacarpus/metatarsus (Mc/MtIII) (cannon bone) (Figs 1, 2) were the most common fracture overall and were the most common type in all forms of National Hunt racing. In flat turf racing proximal phalanx fractures were the most common diagnosis in fatal fractures (Fig 3) (Parkin et al. 2004a). However, in all-weather flat racing, biaxial sesamoid fractures predominated. In the case of proximal sesamoid bone fractures (Figs 4, 5), the overall incidence of such fractures in all racing was 0.63/10,000 starts, which increased to 1.63/10,000 starts in all-weather flat racing. This equated to a risk ratio of 4.4 for these fractures occurring in all-weather flat racing compared with flat turf racing (Kristoffersen et al. 2010). Overall, it was concluded that the incidence of fatal fractures in racing in the UK had not changed since a previous survey a decade earlier (McKee 1995). Furthermore, the risk of a particular fracture occurring varied with the type of racing indicating that the reasons why particular fractures occur are different (Parkin et al. 2004a).

Figure 1.

A lateral condylar fracture of the third metacarpal bone in a 6-year-old National Hunt hurdler. Such fractures are the commonest form of fatal fracture sustained during National Hunt racing.

Figure 2.

A medial condylar fracture of the third metatarsal bone in a 5-year-old National Hunt hurdler. Note how the fracture spirals proximally up the bone and does not exit the cortex as is commonly seen with lateral condylar fractures.

Figure 3.

A mildly displaced sagittal proximal phalanx fracture. Such fractures are the most common fatal distal limb fracture occurring during flat racing.

Figure 4.

A dorsolateral-palmaromedial oblique radiograph of the medial sesamoid bone. There is a nondisplaced mid-body fracture of the proximal sesamoid bone. Such fractures are most frequently diagnosed in horses racing on all-weather surfaces.

Figure 5.

A flexed lateromedial radiograph. There is a large apical fracture of one of the sesamoid bones.

When the reasons why horses are at altered risk of fracture during racing were assessed, it was discovered that horses that had either done no galloping work during training and those in their first year of racing were at an increased risk of sustaining a fatal distal limb fracture when they raced. Furthermore, horses trained on a sand gallop were at an increased risk of fracture. When the analysis was extended to just include reasons why horses sustain the most common fracture, lateral condylar fracture of Mc/MtIII, it was discovered that affected horses again were horses that had either done no galloping work during training and those in their first year of racing were at an increased risk of this particular fracture. Also, horses that sustained a lateral condylar fracture of Mc/MtIII were more likely to have commenced their racing career as 3- or 4-year-olds rather than as 2-year-olds (Parkin et al. 2004b,d).

When the reasons why particular races may be at risk were examined, it was determined that longer races that contained more runners were at increased risk. Both firmer going and fewer days since the last race on the same course, were also associated with an increased risk of a horse sustaining a fatal distal limb fracture. When the analysis was extended to just include reasons why races may cause an altered risk of sustaining the most common fracture, lateral condylar fracture of Mc/MtIII, it was found that such fractures were more common in races with a larger number of runners, races in which professional jockeys were not permitted to ride and races in which the going was described as good or firm (Parkin et al. 2004c,d).

Using videos to analyse risks and occurrences of such fatal distal limb fractures, it was found that approximately 75% of all fractures occurred spontaneously and were not associated with a fall or interference from another horse. In National Hunt racing, most fractures (74%) occurred in the second half of the race, whilst in flat racing fractures occurred equally at any time through the race. The majority of horses (66%) that sustained a forelimb fracture did so when the affected leg was the lead leg at the time of fracture (Parkin et al. 2006a).

The pathological configurations of lateral condylar fracture of Mc/MtIII were examined and level of pre-existing pathology determined in the limbs. This study concluded that there was a much higher degree of bone comminution (both affecting the joint and further up the bone) in comparison with previous reports of such fractures, as well as a greater presence of other concurrent fractures. Many limbs had evidence of pre-existing pathology (Fig 6), particularly in the medial and lateral parasagittal grooves of the distal articular surface of Mc/MtIII, although the presence of such pathology was not associated with horse age, length of career and number of race starts during their career (Parkin et al. 2006b; Barr et al. 2009).

Figure 6.

A post mortem view of the distal condyle of the third metacarpus from a 4-year-old flat racing Thoroughbred. The condyle shows extensive condylar pathology with the presence of wear lines, transvere ridge arthrosis and an obvious fissure affecting the parasagittal groove of the lateral condyle. Such pre-existing pathology is frequently observed in racing Thoroughbreds.

Fractures during training

There have been a number of studies performed to describe the risk factors for severe musculosketelal injury in horses during training and, whilst such studies inevitably involve other injuries, by virtue of their frequency fractures are often the most common injury described.

An epidemiological study recruited a cohort of 13 flat racehorse trainers and followed 1178 horses over a 2 year period to give a total of 12,893 months at risk during training. This produced a figure of nontraumatic fracture incidence of 1.15/100 horse months, with 78% of the fractures occurring during training. This would translate to a trainer with 100 horses in their year having just over one horse sustain a fracture each month in their yard. They found a wide variety of bones were involved in fracture, with tibial and pelvic stress fractures making up 28% of the total number of fractures (Verheyen and Wood 2004).

A subsequent publication from this study analysed in detail the reasons why horses sustain tibial and pelvic stress fractures, the 2 commonest causes of fracture in this population. Results from this study identified some influences in the training of the horse on the risk of sustaining such fractures. Looking at the 30 day period prior to the horse sustaining a fracture, the risk of sustaining a tibial and pelvic stress fracture increased with increasing distance cantered by the horse, reaching a peak at 50 km, before declining. However, if the period of 60 days prior to fracture was examined, no effect on distance cantered and the risk of fracture was identified. There was some association with horses sustaining such fractures and the use of a particular sand-based all-weather training surface, although this finding was recommended to be treated with caution as there may be undetermined influences from the construction and maintenance of the surface. There was no effect of gender or age of horse on the risk of fracture, but there was an identifiable trainer effect. Whilst the authors concluded that there was evidence that such injuries could be potentially modified by scientifically based training regimes, it was premature to make any firm specific recommendations on training regimes (Verheyen et al. 2006a).

A further study using this same cohort of flat racehorses in training was performed to identify specific training regimes that could influence the risk of bone injury. Results from this study showed a strong interactive effect of exercise distances at different speeds on fracture risk. Horses that exceeded 44 km at canter (< or = 14 m/s) and 6 km at gallop (>14 m/s) in a 30 day period were at a particularly increased risk of fracture. These distances equate to approximately 7700 bone loading cycles at canter and 880 loading cycles at gallop. Fifty-six fractures occurred in the subset of study horses that were followed since entering training as yearlings, when skeletally immature (n = 335). Cohort analysis of this data set showed that, in previously untrained bones, accumulation of canter exercise increased the risk of fracture whereas accumulation of high-speed gallop exercise had a protective effect (P<0.01). However, increasing distances at canter and gallop in short time periods (up to one month) were associated with an increasing fracture risk. Results from this study provide further data on the effects of physical exercise on bone adaptation and injury risk (Verheyen et al. 2006b).

The same investigators using the same cohort of trainers and horses described epidemiological factors associated with dorsal metacarpal disease ([DMD] sore shins). They determined that in their cohort of horses in training DMD had an incidence of 1.87/100 horse months. Thus a trainer with 100 horses in training would expect to have just less than 2 cases each month. The DMD was associated with horses that had increasing exercise distances at canter and increased high speed exercise over a short period of time (<1 month). However, the incidence was decreased in horses that had greater cumulative exercise distances since entering training. Overall, it was concluded that the risk of DMD in young Thoroughbreds decreases with accumulation of distances exercised at canter and high speed, probably reflecting adaptive response of the third metacarpal bone to the loads placed upon it. However, increasing exercise distances in short periods (up to 1 month) increases the risk of DMD, probably as a consequence of microdamage and associated remodelling response and should therefore be avoided. For the first time, investigators gave specific recommendations on training and recommended that early but gradual introduction of small amounts of high-speed exercise may be beneficial. Canter exercise should be kept to a minimum on high-speed work days and large amounts of both cantering and high-speed work per week, 2 weeks, or per month should be avoided, particularly during the early stages of training, in order to minimise the risk of DMD (Jackson et al. 2005).

A further study was performed to estimate the incidence of both fractures and tendon and suspensory ligament injuries in National Hunt racehorses in training and determine particular risk factors for these injuries. This study used the same methodology as the earlier training studies with data collected on horses from 14 UK National Hunt training yards for 2 racing seasons. Data was available from 1223 horses that spent 9466 months at risk of injury. The fracture incidence rate was 1.1/100 horse months (very similar to that seen in flat horses) and varied significantly by trainer (P<0.001) but not by gender, age or background (whether a National Hunt store or previous flat-trained horse). The fracture incidence rate in racing was 9.2 per 1000 starts. The pelvis and third metacarpal bone (McIII) were the most commonly fractured bone, with McIII fractures occurring more commonly during racing whilst pelvic fractures occurring predominantly during training (Ely et al. 2009).

The Verheyen group determined the effects of dam age and parity on the rate of fracture in offspring of Thoroughbreds in training for flat racing. Using data from 8 trainers on 335 horses, monitored since the start of their training as yearlings, it was shown that first foals had a significantly (approximately one-third the risk) lower fracture rate than subsequent ones and rate of fracture decreased with increasing dam age (Risk ratio = 0.91 per year increase in dam age). This study identified that the rate of equine injury may be influenced by factors that affect skeletal development and, whilst no mechanism or reason for the phenomenon was proposed, it was stated that further research on intra-uterine and perinatal determinants of injury risk in later life in horses was needed (Verheyen et al. 2007).

Conclusions and future directions

There is now a considerable body of baseline data available on the rate and type of fracture that occurs in training and racing, which can provide the basis for future intervention studies to determine the success of strategies to reduce the incidence of fracture and fatality in racing, particular in the UK. In studies of racing, a number of key variables have been identified which can be modified to reduce the incidence of such fractures. Relating to racecourses, the ground conditions have consistently been shown to be important with firm/harder going being an important determinant of increased risk. There is good evidence that track conditions are important as fractures are more common in races where the ground has been raced over most recently, almost certainly decreasing the time allowed for ground maintenance. Undoubtedly a key area of future research is the development of an understanding of identifying, measuring and maintaining an optimal racing surface. Furthermore, such risks may be modifiable by identifying how the hoof may interact with particular surfaces, and how such interactions can be modified by both surface, hoof and shoe design.

Pre-existing pathology is common in affected limbs and is likely to be a major determinant of fracture risk. Techniques to image or otherwise identify such pathology in vivo is vital, and in particular, development of knowledge relating to which particular pre-existing pathologies are the major risk determinants for potentially catastrophic injury.

The past training experience of the horse appears to be a key determinant of fracture risk. Horses that are in their first year of racing, those that have not undertaken galloping work during training, and those that did not start racing until 3–4 years of age are at an increased risk. There is an increasing understanding of how specific training regimens and schedules may be particular risky or protective, although currently few investigators have so far made firm recommendations on exercise protocols (except loosely in the case of DMD). It is key that studies are developed in the future to really test particular training protocols and determine optimal training schedules to both produce winning success and allow for injury reduction, Such studies will be hard to design and perform, particularly in the area of racehorse training where there is a strong traditional ethos relating to the ‘art’ of the discipline, whilst allowing for relevant scientific development. Such investigations will require excellent field and epidemiological skills, as equine experimental studies are often too expensive to practically achieve sufficient experimental power to identify an effect. Other aspects relating to training are important, illustrated by studies relating to falling, which identified that a horse's past experience is strongly implicated in the risk of falling.

All these studies have the potential to have an impact on animal welfare, racing practice and the high levels of morbidity frequently seen in the racing industry. A key challenge in the future is the development of strategies that have the potential to reduce morbidity and mortality and are acceptable to the racing industry, where performance is key.

Author's declaration of interests

HBLB's Veterinary Advisory Committee commissioned and sponsored this article as part of a series summarising progress made in areas relating to their priorities for research funding. The author and other workers at his university hold current and previous research grants funded by the HBLB.

EVJ is delighted to publish HBLB's Advances in Equine Veterinary Science and Practice Review Series in recognition of the major contribution that HBLB research and educational funding has made to the health and welfare of the Thoroughbred.

Source of funding

The author would like to thank the HBLB and Aintree and Cheltenham racecourses for generously funding some of the research cited in this review.


The author would like to thank all colleagues, veterinary surgeons, scientists and trainers who contributed to the cited studies.