Identification of risk factors for traumatic injury in the general horse population of north-west England, Midlands and north Wales

Authors


  • [Correction added after online publication 23 September, 2011: ≥ on page 5 of the paper was incorrect. It has been changed to ≤ as requested by the authors]

Summary

Reasons for performing study: Traumatic injuries are a major cause of morbidity and mortality in the horse and consequently pose a serious threat to horses' wellbeing. To date, there have been no published studies assessing the frequency of injuries in the general horse population of the UK.

Objectives: To obtain information regarding husbandry management strategies and injury prevalence in horses aged ≤15 years, with the aim of identifying predisposing risk factors for injury.

Methods: A postal questionnaire was distributed to a randomly selected sample of horse owners across north-west England, Midlands and north Wales. Factors associated with injury were assessed using univariable and multivariable logistic regression analysis performed with the binary outcome variable defined as whether or not the horse had sustained an injury within the previous 12 months.

Results: A usable questionnaire response rate of 68% (652/953) was achieved. Forty percent of horses had sustained a traumatic injury within the past year, of which 62% occurred in the field and 13% during ridden exercise. Factors identified as being associated with an increased risk of traumatic injury included the following: breed other than cob or pony (P = 0.001), shorter duration of ownership (P = 0.002), being turned out with an increasing number of horses (P = 0.001), being used for competitive (P = 0.001) or Parelli (P = 0.006) purposes. Stabling at all times during the spring (P = 0.005), the use of wood fencing in paddocks (P = 0.05) and being prone to becoming distressed if left alone in a field (P = 0.04) were also found to be associated with an increased risk of injury. Stabling at all times during winter was associated with a decreased risk of injury (P = 0.006).

Conclusions and potential relevance: Risk factors for sustaining injuries have been identified in association with management practices. This information may be used to educate owners regarding management of their horse(s) in order to prevent injury.

Introduction

Traumatic injury can be described as damage or harm caused to the structure or function of the body by an external or internal force which may be physical or chemical in nature. In horses, traumatic injuries are a frequent occurrence in both general purpose and competition horses. Injury in the performance horse has been a focus in recent literature, with particular attention being paid to the identification of risk factors that predispose to injuries in racehorses (Williams et al. 2001; Parkin et al. 2004; Pinchbeck et al. 2004; Boden et al. 2007), event horses (Singer et al. 2008) and dressage horses (Murray et al. 2010). Improvements in fence construction and enforced legislation on the use of riding whips on the racetrack have subsequently been employed in an attempt to reduce the frequency of injury.

Traumatic injuries are also common in the general equine population. In a practice-based survey on the health of horses in northern Britain, each horse received on average 0.88 veterinary visits per year and traumatic injuries were found to be the principal reason for veterinary treatment for nonroutine purposes (Mellor et al. 2001).

In the National Animal Health Monitoring System Equine 98 study in the US (Anon 1998a), it was reported that injury, wounds and trauma affected 13.4% of foals aged <6 months and 6.6% of horses aged >6 months (Anon 1998b). In addition, 10.5% of deaths in horses aged >30 days were attributed to injury, wounds or trauma with only colic and ‘old age’ accounting for greater percentages of mortality. Despite these figures and the significant impact that traumatic injuries have on equine welfare, there has been little effort to identify risk factors for these types of injuries in horses in the UK or USA.

Epidemiological studies on the prevalence of bite and kick injuries in Swiss horse populations identified an overall injury prevalence of 8% with 21.6% of these attributed to a bite or kick. Maintenance of a stable group hierarchy and provision of adequate space have been found to be important factors in the prevention of injury (Furst et al. 2006; Knubben et al. 2008). Christensen et al. (2002) looked at the effect of stabling on social behaviour in stallions and concluded that those stabled alone demonstrated increased aggressive behaviour towards other horses.

The following study was designed using a postal questionnaire to obtain information regarding the prevalence and type of traumatic injuries sustained by horses in north-west England, Midlands and north Wales and to identify risk factors that may predispose horses to injury. Prior hypotheses were that the types of fencing, the use of the horse, the size of fields, group composition during periods of turnout and stereotypical behaviour would be associated with the risk of injury.

Materials and methods

Selection of study population

A comprehensive description of recruitment of the initial study population is outlined by Ireland et al. (2011a). Horses registered with veterinary practices in the north-west and Midlands areas of England and north Wales, were eligible for inclusion in the study. From the database of referring veterinary practices held at the Philip Leverhulme Equine Hospital (PLEH), University of Liverpool, 117 practices with equine clients in these areas were invited to participate in PLEH studies.

A sample size calculation was performed, which allowed an estimation of the prevalence of traumatic injury of 10–20% with a precision of 3%. Assuming a response rate of approximately 70%, with a usable response rate of 60%, our sample size for mailing questionnaires was 900–1000 individual horse owners.

Horse-owning clients (n = 7350) were randomly selected from all client lists using randomly generated numbers. Owners were mailed postcards requesting information on the number and age ranges of the horses and ponies they owned. Horses aged ≥15 years were used in a cross-sectional study of geriatric horses (Ireland et al. 2011a) and horses aged ≤15 years were included in this injury study. Owners expressing an interest in participating in PLEH studies were incorporated into a database and all 953 owners with horses aged ≤15 years were subsequently mailed questionnaires.

Questionnaire design

The postal questionnaire was designed using high accuracy, high throughput data capture software (Teleform version 9.1)1. An 11 page A4 booklet was designed and precoded with the front cover detailing instructions on how to complete the questionnaire (the questionnaire is available online - Supporting item 1). The participant was asked to answer the questionnaire based on a horse or pony that they owned or loaned aged <15 years at the time of questioning. If the owner had more than one horse in this category they were asked to complete the questionnaire based on the animal whose name came first alphabetically.

The questionnaire comprised 7 sections relating to the horse's background, daily management, behaviour and any injuries incurred over the previous 12 months. Owners were asked to report ‘all injuries whether severe (e.g. fracture) or mild (minor wound or graze)’. In the event that the horse had sustained >2 injuries during this period, the owner was requested to complete the questions based on the 2 most recent injuries. The owners were asked to classify the severity of injury under the following headings: mild, moderate and severe, guidelines for classification were as follows; ‘e.g. graze = mild, fracture = severe’. The majority of questions were close-ended style with tick boxes for the respondents' answers and capture boxes for any further written information.

A pilot study was conducted with 20 individuals and the questionnaire design was reviewed in accordance with their comments. The tailored design method (Dillman 2007) of questionnaire distribution was utilised to optimise the response rate. The self-administered questionnaire was distributed to 953 horse owners. An information letter outlining the study, and assuring confidentiality of the information received was mailed along with a prepaid return envelope. The information letter was purposefully designed to lack focus on the ‘injury theme’ of the study, in order to avoid participant bias. Postcard reminders were sent to all owners who had not responded within 3 weeks of the initial postal date. Those who had not responded within 3 weeks of the postcard reminder were mailed a second copy of the original questionnaire. All completed questionnaires satisfying the inclusion criteria, returned within 28 weeks from the original postal date were included in the study. The study design was approved by the Ethics Committee at the University of Liverpool.

Data analysis

Questionnaire and postcard returns were scanned and manually verified by an operator (K.R.O.) into an Excel database (Microsoft Corporation)2 using Teleform software. Statistical analyses were performed using commercial statistical software (PASW Statistics version 17)3. Data are described as mean or median for continuous data dependent upon data distribution and as proportions (%) for categorical data. Descriptive statistics, Chi-squared tests and Kruskal-Wallis tests were used to compare injury types, treatment and outcome, and to compare the ‘kept at livery’ and ‘kept at home’ horse populations.

Sustaining at least one injury within the previous 12 months was defined as the binary outcome. Chi-squared analysis testing was performed on the categorical variables to test for associations with outcome. Screening of all variables and their association with injury was performed using univariable logistic regression analysis. Variables identified with a P value <0.25 were considered for inclusion in a multivariable logistic regression model which was built using a step-wise, backward elimination procedure. Variables remained in the model if they significantly improved the fit (P≤0.05), or if removal resulted in substantial change to the effect of other variables. Biologically plausible interaction terms were assessed. The goodness-of-fit of the model was assessed using the Hosmer-Lemeshow test statistic.

Results

Responses

From the initial postcard mailing requesting age information and study participants, 2097 completed postcards were returned providing age information for a total of 7088 horses and ponies. Overall, 1974 owners (94.1%) indicated they were willing to participate in further PLEH studies (Ireland et al. 2011a).

The questionnaires were sent to 953 respondents that owned a horse or pony aged ≤15 years. Seventy-one percent (677/953) of questionnaires were returned and 68% (652/953) satisfied the inclusion criteria. Excluded responses included: horses and ponies aged >15 years, incomplete questionnaires and one donkey.

Case details and management

Sixty-three percent of the study population were geldings, 36% mares and 1% stallions. Descriptive statistics are shown in Table 1 and Supporting item 2 (available online). The majority of horses were classed as general purpose riding horses; however, many horses were also used for other purposes, e.g. nonprofessional showjumping and dressage. Insurance policies were held by 81.4% of owners questioned, 85% of which were full comprehensive cover inclusive of veterinary fees, 14% of which held additional loss of use insurance cover.

Table 1. Descriptive statistics and univariable logistic regression analyses of all continuous variables investigated in the study for their association with sustaining at least one traumatic injury in the previous 12 months (Odds ratios expressed per increase of one unit. P values indicated in bold type were included in the multivariable logistic regression model due to the P value being <0.25)
VariableNo. of participantsMedianRange (mean)Coefficients.e.Odds ratio95% CIWald
P value
General horse and management details        
Continuous variables        
 Age (years)641100.25–15 (9.6)-0.0010.0021.01.0–1.00.5
Height (cm)64415781–182 (156)0.020.0061.01.0–1.0 0.06
Length of ownership (months)634481–180 (55.2)-0.0090.0021.00.98–0.99 <0.001
Length of time at current premises (months)632281–158 (38)-0.0050.0031.00.99–1.00 0.06
 No. people responsible for horse63721–10 (1.9)-0.080.080.90.8–1.10.4
No. horses on premises642100–160 (16.1)0.0090.0041.01.0–1.02 0.03
No. horses in field64120–50 (4.07)0.030.011.01.0–1.1 0.07
 Size of field/paddock (acres)48030.1–80 (5.03)0.0080.021.00.98–1.040.6
 No. water troughs63210–8 (1.4)-0.080.10.90.8–1.10.4
No. feed areas in field/forage8820–10 (2.15)-0.30.20.70.5–1.01 0.06
No. feed areas in field/concentrates7410–6 (1.55)-0.70.30.50.3–0.9 0.01
No. times ridden per week57251–10 (4.43)0.20.051.31.1–1.4 <0.001
Intensity of work (0–10)63160–10 (5.41)0.10.041.11.1–1.2 0.002
No. veterinary visits in past 12 months63510–20 (1.9)0.40.061.51.3–1.6 <0.001

Horses stabled on livery yards comprised 45% of the study population with 70% of these kept on a do-it-yourself livery basis. Most horses were housed in rows of purpose-built stables, or buildings with stables inside. Feed was provided for horses inside a stable in 79% of cases, 9% were fed in the field and 4% fed on a yard. The remaining 8% were fed in a combination of these locations. Daily management of the horses was categorised according to season.

Injuries

Forty percent (261/652) of horse owners reported that their horse had sustained at least one injury within the previous 12 months. The number of traumatic incidents per horse ranged from 1–8. Forty-three percent of injured horses suffered a single injury, 30% 2 injuries and 13% 3 injuries. In total, 432 injuries occurred within the sample population over the previous year giving a period prevalence of 66.3% (95% CI 62.6–69.9). Details of 370 of these injuries were reported in the questionnaire responses. Fifty-four percent reported the injury sustained to be mild, 36% moderate and 10% severe. Injury details are summarised in Table 2.

Table 2. A summary of the type, location and reported cause of 370 injuries sustained by 261 horses within a 12 month period
Injury variableNo. horses (%)
Type of injury 
 Single wound174 (45)
 Multiple wounds35 (9)
 Bruising57 (15)
 Ligament/tendon injury36 (9)
 Single fracture10 (3)
 Multiple fractures1 (0.3)
 Synovial sepsis5 (1.3)
 Foot penetration16 (4)
 Eye penetration3 (1)
 Other e.g. fractured tooth47 (12)
Part of body injured 
 Distal limb169 (46)
 Proximal limb62 (17)
 Foot45 (12)
 Back34 (9)
 Head28 (7)
 Neck15 (4)
 Hindquarters12 (3)
 Eye9 (2)
 Trunk3 (1)
Cause of injury 
 Unknown74 (19)
 Kick68 (18)
 Bite26 (7)
 Slip/fall51 (14)
 Fencing34 (9)
 Penetration31 (8)
 Other e.g. over-reach injury94 (24)
Incident location 
 Field229 (62)
 Ridden48 (13)
 Stable40 (11)
 Yard7 (2)
 At a competition17 (4)
 School/arena11 (3)
 On a road10 (3)
 During transportation7 (2)

Wounds were the most common type of injury reported. Sixty-two percent of injuries were sustained during periods of turn out to grass. Thirteen percent of injuries were reported to have occurred during ridden exercise, with 37.5% of these taking place during hacking and 32.5% during cross-country activities. Incidents in the stable accounted for 11% of injuries, with a positive association between injuries in the stable and the head or eye being affected (P = 0.005).

The distal limb was affected in 46% cases. A positive association was identified between injuries that occurred during ridden exercise and the distal limb or foot being involved (P = 0.005).

Injury treatment

Owners sought veterinary treatment in 47% (175) of cases. Of those injuries not treated by a veterinary surgeon, 33% were reported to have required no treatment, with 67% treated by the owner or a friend. There was an association between the severity of injury sustained and whether veterinary treatment was sought, with more severe injuries being more likely to be treated by a veterinary surgeon (P<0.001). Analgesia (67%) and antibiotics (43%) were the most commonly administered medications. Veterinary surgeons referred horses for further investigation in 17% of cases.

Burden of disease in terms of cost and time

Twenty-one (6%) horses underwent a period of hospitalisation, ranging from <1 day to 8 weeks with a median of 3 days. Horses were confined to a stable in 37% (138/370) of cases, with the period of box rest ranging from one day to 50 weeks (median 14 days). The overall duration of time taken to recover ranged from <1 week to 12 months, with 5% (17) horses failing to return to their previous level of activity. Of those horses that had at least a 7 day recovery period (n = 265, 72%), the mean recovery period was 5 weeks. When asked the financial cost of the injury to the owner in relation to veterinary fees, the majority (84%) of injuries cost <£500.

Risk of injury

Length of ownership was found to have a significant association with the risk of injury, with those horses owned for longer periods of time having a reduced risk of injury. As the number of horses kept on a single premises increased, the risk of injury increased. Increasing the number of separate areas of concentrate feeding in the field was associated with a reduced risk of sustaining an injury. Horses ridden with increased frequency and those performing a higher intensity of work had a significantly higher risk of sustaining an injury (Table 1).

A table summarising the categorical variables can be found online (Supporting item 2). Cob breeds had a significantly decreased risk of injury when compared with all other breeds. Horses used for dressage, showjumping and/or eventing purposes and horses trained using Parelli methods were found to have an increased risk of injury. Overall, horses used competitively were found to be at an increased risk of injury.

Horses stabled at livery yards were found to be at increased risk of injury when compared with horses kept at the owner's home (P = 0.04). Horses stabled 24 h/day during the spring and winter had a significantly increased chance of sustaining an injury when compared with horses turned out at all times. Horses turned out in the day and stabled at night during winter and autumn months were similarly found to be at increased risk of injury. Wearing a rug during the colder months in both the stable and the field was also identified as a significant risk factor for injury (Supporting item 2).

The final multivariable model is shown in Table 3. The risk of injury decreased as the length of ownership increased and the risk increased as the number of horses grouped in a field together increased. Cob types and ponies had a reduced likelihood of injury compared to other breeds.

Table 3. Multivariable logistic regression model of risk factors for sustaining an injury within the previous 12 months in 261 horses
PredictorLRS P valueOdds ratio95% CI
  1. WBL: Warmblood; TB: Thoroughbred; ID: Irish Draught; ISH: Irish sports horse; TO: Turnout.

Length of ownership (months)0.0020.990.99–1
No. horses in field0.0011.11.0–1.1
Breeds0.001  
 Reference Cob 1.0 
 WBL 1.91–3.7
 TB 2.11.1–4
 ID 3.51.7–7.1
 ISH 3.11.1–8.3
 Arab 2.71.1–6.4
 Pony 0.80.4–1.6
 Unknown 2.21.2–4.1
Parelli   
 Reference   
  No 1.0 
  Yes0.0068.11.5–43.5
Compete   
 Reference   
  No 1.0 
  Yes0.0012.01.3–2.9
Management spring0.005  
Reference TO all time 1.0 
 TO day/stable night 1.00.6–1.7
 Stable day/TO night 0.90.3–2.2
 Stable all time 15.52.9–81.4
Management winter0.006  
Reference TO day/stable night 1.0 
 Stable day/TO night 3.10.2–58.7
 Stable all time 0.30.2–0.7
 TO all time 0.60.3–1.2
Wood fencing   
 Reference   
  No 1.0 
  Yes0.052.01–3.8
Electric fencing   
 Reference   
  No 1.0 
  Yes0.061.41–2.1
Distressed in field alone   
 Reference   
  No 1.0 
  Yes0.041.51–2.3

Horses trained using Parelli techniques were found to be at a significantly increased risk of sustaining an injury than horses that were not used for this purpose; however, the confidence intervals were wide. Competition was confirmed as a significant risk factor, with horses being used competitively having approximately double the risk compared to horses used for pleasure purposes only. When all injuries sustained during riding were omitted from the model and the model re-run, competition remained a significant risk factor for injury (P = 0.002, OR 1.9, 95% CI 1.3–2.9).

Stabling and grazing management during winter significantly improved the fit of the model and so this variable remained in the final model. Stabling horses at all times during the spring posed a significant increased risk (P = 0.005), with these horses being 15 times more likely to sustain an injury when compared with horses turned out all times. In contrast, stabling horses at all times during the winter was found to be protective against injury (P = 0.006) compared to those horses turned out in the day and stabled at night. Use of wood fencing was associated with increased risk of injury. Use of electric fencing remained in the model as it had an effect on the coefficient for wood fencing, increasing the risk of injury. Horses described by their owners as having a tendency to become distressed if left alone during periods of turnout were also found to be at increased risk of injury. The Hosmer-Lemeshow statistic for this model was 6.28 (P = 0.62). There were no significant interaction terms present.

‘Kept at home’ and ‘kept at livery’ populations

Horses kept on livery yards were less likely to be turned out at all times than horses kept at home premises (P = 0.01). The median number of horses/field was 5.6 at livery yards in comparison to 2.8 at home (P<0.001). Horses at home were found to be less likely to show signs of distress if turned out alone in the field compared with those kept at livery (P = 0.03), and electric fencing was found to be used with increased frequency at livery yards (P = 0.03).

Discussion

This study provides a detailed description of the prevalence and type of injuries sustained by a sample of horses in a subpopulation of the UK. The analysis of management practices in association with the occurrence of injury has identified multiple risk factors for traumatic injury in this equine population.

A high prevalence of injury was encountered in this study, with 40% of horses sustaining an injury over a 12 month period, confirming the general perception that traumatic injuries are common in horses. The prevalence figure is substantially higher than the 7.7% (Knubben et al. 2008) and 7% (Mellor et al. 2001) reported in previous studies. However, the study reported here recorded mild injuries that may not have been attended by a veterinary surgeon, whereas the previous studies only included horses requiring veterinary attention. In a study using identical sampling techniques on horses aged ≥15 years in an identical area of the UK, 10.2% of the population had suffered a wound or accidental injury in the previous 12 months (Ireland et al. 2011b); however, this study did not specifically ask for mild injuries. In the horses aged ≥15 years, 70.2% of reported injuries were attended by a veterinary surgeon compared to 47% in this study, suggesting that owners were more likely to report severe injuries. In addition, the older horses may have a lower rate of injury due to lower competition levels, and possible changes in behaviour as animals age.

Many horses in the study were reported to show signs of distress when left alone in a field, which was identified as a risk factor for injury. Attempts should be made to turnout horses in groups or as a minimum, adjacent to other horses that are separated with fencing. With the rising cost of land and the human requirement to maintain easy access to their horse, horses are frequently stabled individually with limited access to turnout.

Horses that were stabled at all times during the winter were less likely to sustain a traumatic injury when compared with those horses that were stabled during the night and turned out during the day. Although the difference in numbers between these groups was large, this finding correlates with the observation that the majority of traumatic injuries occur during periods of turnout. In contrast, stabling at all times during the spring was found to be a significant risk factor for injury when compared with horses turned out all times. Horses socially confined followed by release are predisposed to demonstrating abnormal behaviours such as redirected sexual behaviour or weaving (Cooper et al. 2000). Horses also frequently demonstrate an increased intensity of normal behaviour, previously termed the ‘rebound’ effect (Henderson 2007; Jorgensen et al. 2009), upon release from confinement. These abnormal or exaggerated normal behaviours may be likely to increase the risk of injury to horses within the group during initial turnout following a prolonged period of stabling.

The cause of most injuries was not determined as owners were not present at the time of injury. In this study, kicks and bites were known to be responsible for 18 and 7% of injuries, respectively. This figure is higher than the overall bite and kick prevalence rate of 21.6% reported in the Swiss population of horses (Knubben et al. 2008). Social confinement can also lead to increased aggression when a previously unsocialised horse is mixed with other horses (Christensen et al. 2002). Horses have been found to be at a lower risk of injury if kept in stable hierarchical groups (Knubben et al. 2008). The introduction of new horses can lead to increased aggression and excitement during periods of adjustment. Horses that had been owned for a short period of time were found to be at increased risk of injury in this study. New horses introduced to a yard with already-established hierarchical groups are therefore at a perceived increased risk of injury. The lower injury rate in older horses found in the study by Ireland et al. (2011b) may be explained by these horses being more likely to stay with the same owner for long periods of time without the need to re-enter new social groups with the same frequency as younger horses. Pre-exposure of new horses by adjacent stabling and gradual introduction of new herd mates may help to reduce the incidence of aggression levels when subsequently turned out into a field (Hartmann et al. 2009). Fitting horses with protective boots or removing their hind shoes until hierarchies have been established may also be advisable during the transition period (Henderson 2007).

Horses turned out with larger groups of horses were also at increased risk of injury, probably due to the necessity to enter larger hierarchy with multiple horses competing for dominance. Invasion of personal space is a common cause of aggression within horses (Heitor et al. 2006), therefore adequate space should be provided for horses to facilitate escape from aggressive behaviour (Knubben et al. 2008; Jorgensen et al. 2009). Livery yards generally contained larger numbers of horses than ‘at home’ premises, with higher numbers of horses turned out together, which was shown to be a risk factor for injury. In addition, a livery yard may be more susceptible to regular changes in turnout group composition.

The number of concentrate feeding areas in a field was found to have a significant effect on injury in the univariable analysis; however, this variable was not included in the multivariable analysis due to the small number of participants that fed horses in the field. Similarly, other studies have found feeding and spacing to be important factors that influence the incidence of aggression between horses (Jorgensen et al. 2009). It is therefore advisable to provide at least one additional feeding zone in addition to the total number of horses in a field, in order to reduce competition for feed and subsequent injury.

The distal limb was the site most commonly affected, with wounds being the predominant injury type. The lower limb of the horse has minimal soft tissue present, leaving the area, poorly protected from injury in comparison to the upper limb and trunk. Wire injuries are frequently blamed for distal limb injuries (Mespoulhes-Riviere et al. 2008); however, wire fencing was not demonstrated to be a significant risk factor for injury in the study. There was a trend for the presence of wood fencing to increase the risk of injury, when electric fencing was also included in the multivariable model. Electric fencing is frequently used to fence off small areas of grazing, particularly on livery yards. It is possible that horses fenced off in this manner may be grazed individually and therefore lack social interaction, or if fenced within a group they do not have adequate space to either escape aggressive behaviour or display normal behaviour. The effect of fencing types and the risk of injury is an area that requires further investigation.

Horses ridden regularly and those used for competitive purposes were at a greater risk of injury. Although very few injuries were sustained during competition in this study, injuries sustained in competition horses have been well documented (Murray et al. 2006; Singer et al. 2008). The greater athletic demands and exposure to high risk activities placed upon horses during competition may be responsible for these horses being at an increased risk of injury. To investigate this association further, injuries sustained during ridden exercise were omitted from the multivariable model and competition remained a significant risk factor. This finding suggests that the variable ‘competing’ may be a proxy for how horses used for competition purposes are managed. Horses used competitively are sometimes stabled individually at all times during the competition season with lack of social contact with other horses due to the perceived risk of injury (Henderson 2007). Providing 1 h of exercise/day, either on a horse walker, turn out or being ridden, has been found to have a positive effect on horse welfare, training ability and owner safety (Freire et al. 2009).

Horses trained using Parelli techniques were found to be 8 times more likely to sustain an injury. Parelli techniques are often advertised as alternative training methods for poorly behaved horses, which may explain why these horses are at an increased risk of injury. However, numbers in this category were low so this finding should be interpreted with caution.

Over half of the injuries sustained were not treated by a veterinary surgeon, which was probably due to the ‘mild’ nature of a high number of the injuries reported. Classification of injuries by owners is likely to be subjective and would undoubtedly differ between individual respondents. Some basic categorisation guidelines were given in the questionnaire in an attempt to minimise inaccurate descriptions of injury severity.

The significant association between injuries occurring in the stable and the head or eye being affected may be related to the number of hazardous objects within a stable, such as hay racks, automatic water and feed troughs. Providing hay from the floor instead of in a hay net or rack, ensuring there are no protruding objects (nails, hooks) in the stable and placing a protective barrier along sharp edges may also help protect from injury.

Cobs and ponies were less likely to suffer an injury than many other breeds. Cobs and ponies are generally thought of as being more placid and less ‘flighty’ in nature compared with Thoroughbred breeds. They also have a lower expected level of athleticism when compared with Thoroughbred breeds, which may explain why they are less prone to injury. Injuries in Cobs may also be under-reported due to their generally longer hair coats, which can lead to minor injuries being missed by the owner.

Injuries in the horse undoubtedly have major financial implications for owners, both in terms of duration of rehabilitation and cost. The average cost of a mild injury was <£100, moderate injury from £100–500 and severe injury from £1000–2000. In conjunction with the financial implications, horses having time off from work or those failing to return to their previous level of activity contributes to the overall burden of injury.

The limitations of this study were similar to those encountered in all postal questionnaire studies. The data collected relies heavily on owner compliance for the gathering of accurate information. This method is also challenged by human error, whereby questionnaires are incompletely filled out or completed incorrectly. The time between the injury and questionnaire completion may affect the accuracy of data provided, as identified in a previous study looking at the accuracy of dressage scores by riders following an event (Murray et al. 2004). The frequency of data entry errors in this study was reduced by manually verifying the Teleform capture data before exporting to a database; however, it is likely that some errors may have gone unchecked. There may have been an element of bias with owners wanting to complete the questionnaire based on more serious injuries or those that occurred >12 months previously. Efforts to limit bias included specific instructions to record all severities of injuries and by asking owners to provide a date as to when the injury occurred.

Only horses aged ≤15 years were included in the study, as those aged ≥15 years were being used in another study (Ireland et al. 2010a). Despite requesting owners to complete the questionnaire based on horses aged ≤15 years, 20 respondents owned horses aged exactly 15 years, which were included in the study; however, it is possible that data on some 15-year-old horses were not received. This is unlikely to have had a significant impact on the results as age was not identified as a risk factor for injury in this study. The age category classification of the studies was designed to avoid concomitant sampling of the same owner group with 2 questionnaires, since there was a risk that the response rate for each study may have been reduced. The overall results could have been affected by not including those horses in older age categories, which, as discussed previously, may have a lower prevalence of injury.

Equine management and husbandry practices vary widely across the equine population of the UK, particularly in accordance with the weather. It was difficult to categorise this information accurately, e.g. whether the horse was predominantly stabled or turned out during specific times of the year. In addition, there may be variation in management practices in different regions; therefore these data may not be directly extrapolated to all horses in the UK.

Although a number of horses had sustained >2 injuries over the previous year, detailed information was only collected for the 2 most recent. Consequently, analysis was based on an outcome of sustaining at least one injury and horses that sustain multiple injuries may have different risk factors, which would not be captured in this study.

In conclusion, traumatic injury is frequent in the general horse population with the majority occurring during turn out. Simple amendments can be made to the management of horses in the UK to help prevent the occurrence of injury, which can lead to considerable cost and periods of time absent from work. The provision of adequate feeding space, allowing regular social interaction with other familiar horses, and the gradual introduction of new horses to yards may help to reduce the risk of injury in the horse. Further observational studies to investigate how injuries are occurring during periods of turnout at grass are warranted.

Authors' declaration of interests

No conflicts of interest have been declared.

Source of funding

We would like to express our thanks to The Horse Trust for generously funding this study and K.R.O.'s Clinical Scholarship.

Acknowledgements

We would like to acknowledge the Veterinary practices and horse owners who frequently participate in PLEH research studies.

Manufacturers' addresses

1 Verity Inc., Sunnyvale, California, USA.

2 Microsoft Corporation, Redmond, Washington, USA.

3 PASW Statistics, Chicago, Illinois, USA.

Author contributions All authors contributed to the initiation, conception, planning, execution and writing of this study. Statistics by K.R.O. and G.L.P.

Ancillary