Recent data recorded by British Horseracing Authority veterinary officers reported an overall incidence of fatality in flat racing of 0.76 per 1000 starts between 2000 and 2013 [4]. It also estimated that in flat racing the incidence of fracture (and fatal fracture) of the third metacarpal or metatarsal bone (Mc/Mt3) was 0.30 (0.19) per 1000 starts, the pelvis 0.26 (0.08) per 1000 starts, the proximal phalanx 0.21 (0.12) per 1000 starts and the proximal sesamoid bones 0.07 (0.05) per 1000 starts. These figures include all flat races on turf and all‐weather surfaces, and it was noted that the overall risk of bone injury (of which 94% were described as fractures or possible fractures) on all‐weather surfaces was 1.5 per 1000 starts compared to 1.24 per 1000 starts on turf. In a follow‐up study, the same authors calculated that the risk of fatal or non‐fatal distal limb fracture in all‐weather flat racing was 0.95 per 1000 starts [5]. In national hunt flat races between 2000 and 2013, the incidence of fracture (and fatal fracture) that included Mc/Mt3 was 1.03 (0.88) per 1000 starts and the pelvis was 0.42 (0.22) per 1000 starts [3].
Thoroughbred Racing in Other Countries
Recent work from Australia estimated the risk of fatality in flat racing in New South Wales and the Australian Capital Territory [6]. Overall, the incidence of musculoskeletal fatality was 0.52 per 1000 starts with a fracture risk of 0.35 per 1000 starts. Euthanasia following proximal sesamoid bone fracture had an incidence of 0.07 per 1000 starts with fetlock as the reported site in 0.06 per 1000 starts. Obviously, utilizing the term fetlock does not identify which bone was fractured, and it is likely that some of ‘fetlock’ fractures were also fractures of the proximal sesamoid bones. A study in Victoria between 1989 and 2004 reported the risk of fatality as 0.44 per 1000 flat race starts and 8.3 per 1000 jump race starts [13]. This study further described the risk of ‘catastrophic limb injury’ as 0.32 per 1000 starts in flat racing and 5.7 per 1000 starts in jump racing.
A recent review of many years of work by the Japan Racing Association demonstrates the value of monitoring data over a prolonged period and also highlights significant differences in the risk of fatal fracture in Thoroughbred racing around the world [7]. In 1980, the incidence of fracture (fatal and non‐fatal) during racing was approximately 23 per 1000 starts, whereas the latest figures presented show a significant decrease to 14 per 1000 starts in 2001. The risk of fracture while racing on turf was reported as a 14‐year (1987–2000) average of 17.7 per 1000 starts and on dirt as 19 per 1000 starts. A 10‐year average of fatal fracture (euthanasia following a fracture) during racing (1985–1994) was reported at 3.2 per 1000 starts.
Work from Hong Kong focussed on proximal limb, and pelvic fractures (only) reported no significant difference in risk between dirt (0.27 per 1000 starts) and turf (0.31 per 1000 starts) tracks [8]. Fractures during racing that resulted in euthanasia had an incidence of 0.08 per 1000 starts; all of which occurred on turf.
Quarter Horse Racing
In comparison to Thoroughbreds, little epidemiological research has been conducted in Quarter Horse racing [21–23]. All of the information comes from the work conducted through the CHRB Post‐mortem Program. Even though full post‐mortems were conducted, the predominant fracture site was described as the fetlock (0.76 per 1000 starts) followed by the carpus (0.48 per 1000 starts), vertebra (0.14 per 1000 starts) and scapula (0.13 per 1000 starts) [21]. The work demonstrates some clear differences in fractures resulting in euthanasia from Thoroughbreds: fractures of the Mc/Mt3 condyles and humerus are more common in Thoroughbreds, whereas carpal, vertebral (particularly lumbar) and scapula fractures are more common in Quarter Horses [22, 23]. The authors hypothesized that differences may be due to inherent breed characteristics such as conformation or limb geometry affecting locomotor biomechanics and, as they compete in races of very different distances, there are also consequent differences in speed.
Endurance Riding
There is very little epidemiological information relating to fractures sustained during endurance rides. In a small series of cases from a single region, in the 2007–2008 season, there were 4.2 fractures per 1000 starts [24]. One horse was subject to euthanasia at the ride following a fracture resulting in an estimate of 0.35 fatal fractures per 1000 starts. However, given the fact that this was a single fatality, this estimate should be treated with caution. Interestingly, the anatomical locations predominantly affected closely mirror those of the Thoroughbred racing on turf with the lateral condyles of Mc3 (37%: 14 of 38) and the proximal phalanx (21%: 8 of 38) being the two most common sites of fracture [24].
Eventing
There is limited information available about the risk of fracture and types of fracture sustained by eventing horses [25, 26]. Two reports indicate that fractures are relatively rare, whether in competition or during training, particularly in comparison to soft tissue injuries. Neither study attempts to quantify risk, in terms of numbers of fractures per 1000 starts during, in particular, the cross‐country phase.
Incidence of Fractures Sustained During Training
Thoroughbred Training
Far fewer studies have been conducted to investigate the risk of, or risk factors for, fracture during training compared to racing. This is due to the ready availability of data from racing in contrast to the need to design robust studies and data collection protocols and to recruit trainers who are willing to participate.
Additionally, studies of horses in training lack an obvious denominator, as with ‘per 1000 starts’ in racing which also complicates attempts to examine risk of fracture away from the racecourse. It might be optimal to quantify risk per training events at different speeds, but that level of detail is rare and it is more common to simply report fracture rates by the number of horse months. This also has the advantage that one can directly compare different trainers with different numbers of horses in training for different periods of time. An inconsistency that does arise concerns the definition of a ‘day at risk’. A horse on box rest is generally regarded as not being at risk of a training‐related fracture. But, is a horse at risk when it is walking and trotting, cantering or only when doing galloping speed exercise? Similarly, at how much greater risk is a horse that is galloping compared to when cantering? Such questions complicate markedly the investigation of training fracture risk.
The majority of training‐related work has been conducted in the UK [27–29]. The first study included details of 1178 horses providing almost 13 000 horse months in flat race training. Using total months in training as the denominator, the incidence of non‐traumatic fractures was estimated at 0.94 per 100 horse months [27]. It is important to note that this estimate excludes 22% of fractures in the same population that occurred during racing. The most common sites of fracture were Mc3 (20%), ilium (16%) and tibia (14%). The respective estimates of incidence rate were 0.22 per 100 horse