A large study from New Zealand followed 1571 horses during 3333 training preparations over 392 290 training days [30]. A total of 55 fractures were recorded, and the authors estimated that the incidence rate for first occurrence fracture‐related lameness was 0.14 (95% confidence interval 0.1–0.18) per 1000 training days, and for second occurrence it was 0.16 (95% confidence interval 0.08–0.3) per 1000 training days. The first incidence rate equates to 0.43 fractures per 100 horse months, suggesting that there are significant differences in the rate of fracture in Thoroughbreds being trained in different parts of the world. It is important to remember that such differences could be due to different gene pools as well as differences in how horses are trained and raced.
Quantifying the number of days lost from training has been the focus of a few studies, providing valuable information about the medium‐ to longer‐term impact of fractures. In one study data from seven UK training yards reported incidence rates for two‐ and three‐year‐old horses separately [31]. Stress fracture incidence rates were similar for two‐ and three‐year olds at 1.48 and 1.43 per 100 horse months, respectively, but the incidence of fatal fracture was almost twice as high in three‐year olds compared with two‐year olds (0.3 and 0.17 per 100 horse months, respectively). The most common fracture site seen in both two‐ and three‐year olds was the pelvis with cumulative incidences of 3% and 5% in two‐ and three‐year olds, respectively. Overall, of 52 601 days available for training in two‐year olds and 29 369 days available for three‐year olds, 27% of two‐year‐old days (14 091 days) and 22% of three‐year‐old days (6324 days) were lost from training. Of the total days lost from training, fractures accounted for 18% in two‐year olds and 25% in three‐year olds. With a mean of 95 days lost per two‐year‐old fracture case and 115 days lost per three‐year‐old fracture case.
Further evidence of the impact of fracture on training is provided in a retrospective study of veterinary records from three training yards in Newmarket [32]. Over the period of study, an average of 332 horses were in training, and 50 tibial stress fractures, 35 proximal phalangeal fractures and 27 carpal fractures were recorded. Average annual injury rates (musculoskeletal injuries in general) were similar between the three yards (between 23 and 26%). However, there were significant differences in the types of fracture seen in different yards with proximal phalangeal fractures being up to three times more common in one yard compared with the other two, and tibial stress fractures being more than twice as common in one of the other yards.
These studies demonstrate how important it is to accurately record detailed information about the occurrence and impact of fractures during training as well as racing in order to clearly identify the level of risk to which horses are exposed. Without this, it is impossible to assess the impact of intervention. The fact that there were significant yard‐level differences in both studies also show how important it is, where possible, to conduct studies at the individual trainer level. There are almost certainly unique trainer characteristics that increase or decrease the risk of fracture or injury more generally. If data are collected and investigated as a whole from a number of trainers and not interrogated for individuals, subtle important differences will be lost and interventions will be less effective. That said, there also has to be a consideration of statistical power and, for some less frequent outcomes, it is often an unavoidable necessity to collect data from multiple trainers.
Showjumping Training
One significant international study used the concept of days lost to training to describe problems associated with elite showjumping horses [33]. The authors note that only 6% of available training days were lost – far fewer than comparative estimates from Thoroughbreds. It perhaps says something in itself, about the prevalence of fracture, that the word is not used in the paper. Clearly, injuries as a whole and in particular fractures are far less common in the elite showjumper. The best approximation of the impact of fractures comes from the estimate that 22% of the 2357 (from a total of 39 028 horse days at risk) days lost were due to an acute orthopaedic injury. This equates to only 1.3% of all available training days.
Measures of Fracture Incidence in Other Horses
Reports of fracture incidence rates in other breeds or non‐sports horses are few and far between. There is some work that describes differences in the type and configuration of carpal bone fractures in Thoroughbreds compared with Standardbreds (and Quarter Horses). However, this work is based on veterinary records of horses admitted to a particular referral hospital in the USA and does not provide any denominator data from which it may be possible to produce estimates of incidence [34]. A more recent study based on 356 Standardbred racehorses, providing 8961 horse months at risk, does provide some estimates of incidence risk for a number of different types of fracture [35]. The authors calculated that the most common fracture types in this population were those affecting the proximal sesamoid bones (0.32 per 100 horse months), followed by proximal phalangeal fractures (0.28 per 100 horse months) and both pelvic and Mc/Mt3 fractures (0.16 per 100 horse months). In comparison to Thoroughbreds in training, it appears that the incidence of pelvic fracture is similar; proximal sesamoid bone and proximal phalangeal fractures are more common in Standardbreds; and fractures of Mc/Mt3 are marginally more common in Thoroughbreds.
Away from sports horses, some work has focussed on the geriatric horse [36–38]. This indicates that lameness is a significant problem in the older horse, and is the primary reason for euthanasia. However, fractures are not of particular concern, and the authors suggest that this is, at least in part, due to changes in management and reduced exercise levels.
Risk Factors, Predisposing Factors and Evidence
All but a very few epidemiological studies that have sought to identify risk factors for fracture have been conducted in Thoroughbreds, and from these a large number of different risk factors have been shown to be associated with various different fracture outcomes. Some risk factors have been identified in one or two studies only, and others are clearly not modifiable. For this reason, the primary focus of this section is on the more commonly identified risk factors and those that have the potential to be altered by way of carefully designed interventions.
Gender is a good example of a risk factor for fracture that has been commonly identified as being important [18, 19,39–41]. However, it is obviously unrealistic to expect male or female horses to be prevented from racing, or indeed entire males to be gelded purely to reduce the risk of fracture when racing. However, such findings do have value in that they may provide insight into the pathogenesis of a particular injury type. From an analytical point of view, it is also important to include such risk factors in multivariable models to account for the potential confounding effect that they may have on other risk factors within the model, so they will continue to be reported, even if their impact and usefulness are limited.
Risk Factors Associated with Training Regimens
A major focus of work to identify, in particular modifiable, risk factors has been the association between training regimens and the risk of fracture in either racing or training. Early studies in the USA reported that the total distance accumulated during a two‐month period was associated with the risk of catastrophic musculoskeletal injury [39], and that risk was greatest within a 30‐day period of above average high‐intensity exercise [42]. In these studies, a period of high‐intensity exercise was defined as a 60‐day period, where the average daily high‐speed distance accumulated was in the top 25th percentile of daily