Thoroughbred horse jump racing is popular in Great Britain (GB). Unfortunately it is associated with inherent risk of injury to the horses involved and it has been shown that the risk is significantly higher in jump than in flat racing. As a result, jump racing has been made a priority in racehorse injury investigation by the racing authorities in GB and is the focus of this thesis. Data about injuries and fatalities collected by veterinary surgeons, from all official race meetings between 2000 and 2009 was made available by the British Horseracing Authority (BHA). Following initial examination of the data, review of the literature and discussion with the BHA, a list of outcomes (injuries and fatality) was defined for further investigation. Multivariable logistic regression modelling was employed to investigate associations between potential risk factors and the outcomes. Model validation techniques were then used for outcomes with the greatest frequencies. In addition, post-mortem (PM) findings from a subset of the available data provided the opportunity to evaluate the accuracy of the information provided. Outcomes selected for further investigation were: fatality, tendon strain, epistaxis, hind limb fracture, pelvic fracture, and proximal forelimb fracture. Multiple risk factors were identified as being significantly associated with each outcome which can be used to guide legislation or further investigation. Risk factors common to many of the outcomes were: season, surface firmness (going), race distance and previous racing history (especially previous flat start history). Notably in some instances the relationships between these common risk factors and the outcomes varied, such that a risk factor might be associated with increased likelihood of one outcome but a decreased likelihood of another. Attempts to validate the models with the most frequent outcomes (fatality, superficial digital flexor tendinopathy and epistaxis) against a novel data set (from the year 2010), demonstrated variable calibration and discrimination and relatively poor predictive ability for all of the models. This was thought to be related to the low outcome frequencies and potentially related to risk factors unaccounted for in the models. Evaluation of the accuracy of the recording system for fatal distal limb fractures using PM findings demonstrated good identification of fracture presence, but relatively poor definition of all affected bones. Frustratingly it was concluded that making policy decisions based on the risk factor models will not be straightforward. Few risk factors had strong associations with all outcomes, not all risk factors are readily modifiable and many potential modifications (such as stopping horses from racing) would have major long term deleterious implications for horses. However, new risk factors for injury were identified providing some additional information about injury aetiology; previously recognised associations (such as firm ground and injury) are supported by the work; and sensible recommendations can be made to the industry, such as: closer monitoring of horses based on their previous racing careers or previous injuries. In addition, further training of racecourse veterinarians and/or provision of diagnostic aids (such as radiography) can be recommended to help with diagnoses made at the racecourses.

Musculoskeletal conditions and exercise-induced pulmonary haemorrhage are commonly diagnosed in Thoroughbred racehorses worldwide, and have serious consequences for racehorse welfare and the racing economy. Despite increasing interest in the study of genetic susceptibility to disease from the veterinary research community as a whole over past decades, the Thoroughbred has been largely ignored as a study group. The availability of software capable of complex genetic analyses using large, unbalanced pedigrees has made the study of genetic susceptibility to disease a realistic prospect for veterinary researchers. This study aimed to complete preliminary analyses of the genetics of a number of important musculoskeletal conditions, and of exercise-induced pulmonary haemorrhage, in two different Thoroughbred populations. Multivariable regression analyses were performed to identify important environmental risk factors for each condition in each population, and heritability analyses were conducted. Genetic correlations between disease conditions were also investigated. Fracture, tendon injury, suspensory ligament injury, osteoarthritis and EIPH/epistaxis were found to be heritable traits in the Hong Kong population. Distal limb fracture, SDFT injury and epistaxis were also found to be heritable in the UK Thoroughbred population. Most heritability estimates were small or moderate in magnitude. Selective breeding strategies that identify those animals with low genetic risk could play a part in future efforts to reduce the incidence of these conditions, in conjunction with favourable environmental manipulations based on research evidence. Due to low heritability, most of the conditions studied here would reduce in incidence slowly if selective breeding were implemented, thus strategic environmental manipulations would be warranted alongside such longer-term efforts to provide effective incidence reductions. A number of conditions were found to be positively genetically correlated, suggesting that risk reduction through breeding could reduce the risk of multiple diseases concurrently. For example, fracture and osteoarthritis were found to be positively genetically correlated (0.85 – 0.89) in the Hong Kong racehorse population. However, using the Hong Kong Thoroughbred population dataset, EIPH/epistaxis and tendon injury were negatively genetically correlated, which suggests that reduction in genetic risk of one of these may lead to increased genetic risk of the other. iii Measures of the durability and performance of racehorses were investigated to assess whether they were heritable traits in the UK and Hong Kong racehorse populations, and to assess their relationship to the disease conditions studied. Selection based on more holistic measures of horse health and longevity such as ‘career length’ could be a more attractive prospect for stakeholders, as this could forego the need to select for many different traits individually. Career length, number of starts over the career, and the level of earnings were all heritable traits in both populations. These holistic traits were found to have variable relationships with the disease conditions studied in each population. These analyses are the first to assess the genetic contribution to risk for many important diseases in the Thoroughbred. They provide a starting point from which further investigations into the applicability of genetic manipulations could yield realistic and achievable tools for racing stakeholders to use to ‘improve’ the breed in future.