Misbehaviour in Pony Club horses: Incidence and risk factors

bs_bs_banner General Articles Misbehaviour in Pony Club horses: Incidence and risk factors P. BUCKLEY*, J. M. MORTON 1, D. J. BUCKLEY 2 and G. T. COLEMAN School of Agricultural and Veterinary Sciences, Charles Sturt University, Wagga Wagga, New South Wales, Australia School of Veterinary Science, The University of Queensland, St Lucia, Brisbane, Queensland, Australia. Equine Veterinary Journal ISSN 0425-1644 DOI: 10.1111/j.2042-3306.2011.00541.x *Correspondence email: pbuckley@csu.edu.au; Received: 24.05.11; Accepted: 04.12.11 Present addresses: 1 PO Box 2277, Geelong, Victoria 3220, Australia; 2 356 Mitchell Road, Wagga Wagga, New South Wales 2650, Australia. Summary Reasons for performing study: Horse misbehaviour is an important cause of poor performance in Pony Club horses, is associated with horse-related rider injuries and has been implicated as a nonspecific presenting sign for musculoskeletal pain. Despite this, little is known about the incidence of and risk factors for misbehaviour in Pony Club horses. Objective: This study aimed to describe the incidence and types of misbehaviour in a cohort of Pony Club horses and to identify risk factors for misbehaviour during riding. Methods: A prospective longitudinal study was conducted with 84 Pony Club horses from 41 families belonging to 7 Pony Clubs in one inland region of Australia. Owners recorded misbehaviour events and kept daily records of horse housing, exercise, nutrition, healthcare and disease status. Horses were subjected to a monthly veterinary examination. Descriptive statistics were calculated to describe the incidence of misbehaviour, and multivariable logistic regression was used to assess putative risk factors. Results: Misbehaviour during riding occurred on 3% of days when horses were ridden. On 52% of days with misbehaviour, the misbehaviour was classified as dangerous. Risk of misbehaviour was independently increased on exercise days when the horse was competing, and in months when the horse was fat or obese, fed supplementary feed daily, grazed on paddocks with >50% of ground cover as green grass, exercised on 5 days per month or less, and ridden for a total of 12 h or more in the month. No significant relationship was detected between misbehaviour and back pain. Conclusions: In populations such as the study population the risk of misbehaviour is higher in fatter horses, in horses with access to pastures with greater green grass cover, in those fed daily supplements, in horses receiving exercise less frequently, and during competition. Potential relevance: These results highlight the importance of considering horse body condition, nutrition and exercise in any investigation of horse misbehaviour. In addition, recommendations based on these results could be used by veterinarians assisting horse owners to prevent misbehaviour. From the perspective of recreational horse owners, behaviour is a key determinant of horse performance. Keywords: horse; Pony Club; misbehaviour; longitudinal study; poor performance; risk factor Abbreviations DE: Digestible energy ROC: Receiver operating characteristic Introduction Misbehaviour, defined as unwelcome behaviour exhibited by a horse when it is being handled or ridden [1], is an important but poorly understood problem [2]. Misbehaviour is a threat to rider safety and may render horses unacceptable for equestrian pursuits and so contribute to horse wastage (loss of horses from activity due to sale, retirement or euthanasia). Misbehaviour not endangering horse or rider may also contribute to horse wastage due to reduced rider enjoyment. A recent survey of owners of horses used for Pony Club in regional Australia highlighted the importance of horse misbehaviour, which owners considered synonymous with poor performance [3]. Horse behaviour is related to horse temperament, and parents with children who ride at Pony Club rated temperament more highly than soundness when choosing a suitable horse for their children [3]. The importance of horse behaviour to horse owners has also been recognised by Swiss researchers [2]. Horse misbehaviour has the potential to result in serious injury to riders. In Australia, one horse-related injury occurs per 1000 riding hours [4], and approximately two-thirds of horse-related injuries have been attributed to horse misbehaviour [5,6]. Furthermore, children s injuries are more serious than their adult counterparts, and the risk of head injuries is greater in younger, less experienced riders [6]. This is of particular concern to recreational equine groups, such as Pony Club, that cater for young and often inexperienced riders. Despite the importance of misbehaviour, no studies describing the incidence of misbehaviour in Pony Club horses and assessing putative risk Equine Veterinary Journal 45 (2013) 9 14 2012 EVJ Ltd factors for horse misbehaviour have been published. Pony Club horses were of interest because of their important contribution to the Australian pleasure horse sector and as a favourite pastime for Australian children. Despite this, there is a dearth of information about horse misbehaviour in this population. The aim of the present study was to describe the incidence of misbehaviour in Pony Club horses and to investigate potentially modifiable management- and health-related risk factors for misbehaviour during riding in Pony Club horses. Materials and methods Study design A prospective longitudinal study involving 84 Pony Club horses in regional Australia was conducted from June 2000 to July 2001. The target population was all Pony Club horses in that organisation s Zone 12 region of New South Wales (a semi-arid, predominantly winter rainfall region in the Wagga Wagga region, in the south east of the state). Multistage sampling was used, with horses selected within families (riders and one or both of their parents). Families were selected randomly from membership lists for each of the 7 Pony Clubs in the region using computer-generated random numbers; the same proportion of families was selected from each club. A horse was eligible if it was cared for and/or regularly ridden by someone within the study family and was used regularly for Pony Club purposes by a child in the family. Additional horses purchased during the study period that met the eligibility criteria were also enrolled. Approval for the study was granted from the Human Ethics in Research Committees, Charles Sturt University. 9

Misbehaviour in Pony Club horses P. Buckley et al. Data collection Information about the horses, riders and riders parents was collected once, at the commencement of the study, during the first visit to horse and farm. Visits by experienced equine veterinary practitioners occurred approximately every 28 days. At each monthly visit, a veterinary examination of the horse was conducted. This included visual assessment at rest and at a trot in a straight line on even ground, palpation of back and girth area, bodyweight measurement (with portable live weight scales; Ruddweigh, KM3 Basic indicator with G3 split live weight platform for horses) a and estimation of body condition score using a previously described technique [7] in which horses are scored from 1 (emaciated) to 5 (obese) and a score of 3 is considered ideal. Back pain was diagnosed when an abnormal response was detected following gentle pressure exerted by manual palpation of the dorsal spinous processes and epaxial soft tissues in the thoracic, lumbar and sacral regions. Horse responses were considered abnormal if they were repeatable and either remained constant or increased in intensity with repeated periods of pressure, and could include localised muscle fasciculations, back lordosis and aggression towards the examiner. The presence of clinical diseases (including lameness and sore back) was recorded. Ancillary aids were not used. Samples and data on pasture quantity and composition were collected at each visit, using representative pasture cuts and visual observations, respectively. The type and weight of supplements fed were measured at each visit. The amount of digestible energy (DE) fed as roughage, grains and concentrates as a proportion of the horses estimated daily DE requirement was calculated. Details can be found elsewhere [8]. Family members maintained daily diary records of horse nutrition, exercise, housing, routine healthcare and observed disease occurrences. Owners were instructed to record days on which they considered that the horse misbehaved during ridden exercise or while being handled, and to briefly note the nature of the misbehaviour. During each monthly visit, the diaries containing the previous month s data were collected and any ambiguities clarified by the investigator. This included clarification of the misbehaviour reported by the family to ensure that the behaviour was, in fact, misbehaviour. Data analysis Based on owner descriptions, misbehaviour events were subsequently categorised as dangerous (substantial risk of injury to rider, any other people handling the horse and other horses in the vicinity) or unwelcome (no substantial risk of injury; Table 1). For each horse, days with one or more misbehaviour events recorded were considered misbehaviour days. On rare days where the same horse had more than one misbehaviour event, we recorded only the most dangerous misbehaviour event for that day. One horse recorded as misbehaving because it was chewing tree bark was excluded from all misbehaviour data analyses. Data were analysed using Stata Statistical Software, release 11 b. Confidence intervals for daily incidence risks of misbehaviour were calculated using robust standard errors adjusted for clustering of exercise day within horses, using Stata s -svy- and -proportion- commands. Each day that a study horse was ridden constituted one exercise day. Putative risk factors for misbehaviour during riding were analysed with the exercise day as the unit of analysis; only days when horses were ridden were used for these analyses. Odds of misbehaviour occurring while the horse was being ridden on an exercise day were modelled using random effects logistic regression with Stata s -xtlogit- command, with horse fitted as a random effect. This approach was chosen over more complex 3-level multilevel logistic models that also specifically accounted for horses within family, as the mean number of horses per family was only 2.05 (84 horses from 41 families), so the associated variance inflation factor (i.e. the design effect) at that level would have been relatively small. Variables with univariable P values < 0.25 based on likelihood ratio tests were all fitted simultaneously and eliminated sequentially using a backwards elimination approach based on P value. Some putative risk factors were averaged (pasture and supplement data) or summed (monthly exercise frequency and duration) for days between visits (the horse month ). Fit of the fixed part of the final logistic model to the observed data was assessed using the Hosmer Lemeshow goodness-of-fit table and test, and TABLE 1: Owner terminology for horse misbehaviours on 219 misbehaviour days recorded in 84* Pony Club horses in the Wagga Wagga district of New South Wales, Australia, monitored from June 2000 to July 2001, and classifications as either dangerous or unwelcome by the first-named author Dangerous misbehaviour the predictive ability of the model was assessed using area under the receiver operating characteristic (ROC) curve and a 2-graph ROC curve [9]. Clustering of misbehaviour on exercise days among horses was assessed using an intraclass correlation coefficient, calculated using the latent variable approach. Results Unwelcome behaviour Bucking Snitchy Pulling like steam train Rearing Shying Reluctant to walk Pig rooting Resenting foot trimming Fussing with new bit Biting when girthing up Misbehaving in float Tends to rush Biting and aggressive to Not responsive to bit Not listening other horse Kicking other horse Bit hot Head tossing Bolting Refusing jumps Backing up Kicking child in paddock Throwing head Excited Attacking person Resisting bit Headstrong Biting when grooming Not responsive Trouble catching Cranky Playing up Frisky Hard to catch Toey Sour Fresh Sluggish Escape from paddock Reacting to noise *These 219 misbehaviour days occurred in 50 of the 84 horses. A further 32 misbehaviour days occurred, but details of the misbehaviour exhibited were not recorded. Horse behaviour relating to unobliging responses to rider aids, tending to aggressive. Horse behaviour relating to increased reactiveness to external stimuli, as well as increased limb movement in excess of that asked by the rider. Description of study population Of the 88 randomly selected families, 41 were enrolled in the study [8]. These families had 84 eligible horses, and all were enrolled. Study horses were exercised on a total of 4839 days and ridden most frequently during spring (September November), with a third of all exercise days and hours occurring in this season, followed by autumn (March May) and then summer (December February). Horses were generally exercised infrequently and for short time periods; total daily exercise duration ranged from 15 min to 10 h (mean 2 h per exercise day); the shorter durations generally occurred during riding for leisure and schooling, with the longer durations generally occurring on competition days. Description and incidence of misbehaviour Fifty of the 84 horses (59%) exhibited owner-reported misbehaviour (including misbehaviour during handling) once or more during the study period; these misbehaviour events occurred on a total of 251 days. Details of misbehaviour were not recorded on 32 of these misbehaviour days. Of the remaining 219 days, the misbehaviour was classified as dangerous on 60% (132 of 219) of misbehaviour days and unwelcome on 40% (87 of 219; Table 1). From the owners diary entries, the type of exercise could be classified into 3 categories: leisure, schooling and competition. Of the 84 study horses, 7 had no exercise days recorded and a further 2 had an unusually high daily risk of misbehaviour during riding in particular months; both were excluded from incidence risk calculations and risk factor analyses. For the remaining 75 horses, the daily incidence risk for misbehaviour on riding 10 Equine Veterinary Journal 45 (2013) 9 14 2012 EVJ Ltd

P. Buckley et al. Misbehaviour in Pony Club horses days was 2.7% (133 of 4839; 95% confidence interval (CI) 1.7 3.7). Among these 75 horses, the daily incidence risk of misbehaviour during riding was highest on days when horses competed (5.4% or 21 of 389; 95% CI 2.5 8.3) and lower during leisure riding (2.3% or 53 of 2313; 95% CI 1.0 3.5) and on days when horses were schooled (2.8% or 59 of 2137; 95% CI 1.5 4.0). Risk factors for misbehaviour during riding Descriptive statistics and univariable odds ratios and P values for 14 exposure variables that were eligible to enter the final regression model as potential risk factors for misbehaviour for the 75 horses are listed in Table 2. An additional 11 variables were assessed but deemed not eligible to enter the final model, based on high univariable P values (>0.2). These were horse breed, sex and height, duration of horse ownership by family, age and sex of rider, presence of back pain at the start of the month, horse bodyweight at the start of the month, feeding concentrates during the month in which the exercise days occurred (fed on either any days or none), and duration and frequency of stabling on which the exercise days occurred. Of the 4839 exercise days by the 75 horses, 4092 (84.6%) from 73 horses were included in the final model. The mean number of exercise days during the study period for these 73 horses was 56.1 (range 2 208). These horses were from 39 study families. Seven diet- and exerciserelated exposure variables were significantly associated with misbehaviour in the final model (Table 3). Risk of misbehaviour on an exercise day was increased in months with fewer than 5 exercise days, in months with more than 12 h of riding time, in months when the horse was fed supplements 7 days per week, in months when pasture cover was predominantly green grass, in months when body condition score was >3.5 (i.e. more obese) and on exercise days when the horse was competing. Risk of misbehaviour was less in months when the pasture cover of legumes was intermediate (proportion of ground area covered by legumes >0% to <10%). As risk of misbehaviour on an exercise day was increased in months with few exercise days and, independently, in months with more hours of riding time, effects of average time exercised on exercise days were further explored by fitting this variable separately with each of exercise frequency and total riding time (adjusted for other variables in the final model); average time exercised on exercise days was not associated with misbehaviour in either model (P = 0.68 and 0.37, respectively). The Hosmer Lemeshow goodness-of-fit P value was <0.001, and there were more exercise days with misbehaviour than expected based on predicted probabilities in 9 of the 10 predicted probability deciles. The highest predicted probability for any exercise day was only 0.28. The final model had only modest predictive ability; area under the ROC curve was 0.78 (95% CI 0.74 0.83), sensitivity was high only at very low probability cut-points (<0.02), whereas specificity was high only at probability cut-points >0.1. The intraclass correlation coefficient for misbehaviour on an exercise day from the final model was 0.39 (95% CI 0.24 0.58). Discussion This study is, to the authors knowledge, the first to quantify the incidence of misbehaviour in a population of horses. The study formed part of a larger longitudinal study investigating horse health and performance in Australian Pony Club horses [8], where misbehaviour emerged as an important cause of poor performance in an exploratory qualitative survey [3]. Pony Club horses were of interest because they are the grass roots of Australian equestrian sporting pursuits. In addition, Pony Club horses represent other diverse populations of horses kept for recreational purposes. Over half of the misbehaviour events were regarded as dangerous, and it is clear that misbehaviour is an important issue in this sector of the horse industry, particularly given the young age (and hence vulnerability to falling and injury) of Pony Club riders. However, the daily incidence risk of misbehaviour during riding was low, and this may reflect the importance that Pony Club participants place on horse behaviour [3]; in other words, horses that behave poorly are selected out of the Pony Club group. It is worth noting that this low daily incidence risk may be an underestimate because of failure of some owners to complete diary entries each day and the probable occurrence of misclassification of certain misbehaviours as normal. Equine Veterinary Journal 45 (2013) 9 14 2012 EVJ Ltd Exercise Misbehaviour was more likely when horses were competing, a period when they are subjected to greater mental and physical challenges than during leisure riding. Mental challenges for the horse while competing may be associated with factors such as horse rider mismatch [10], rider tension [11] and horse temperament [10], and could adversely affect horse behaviour. Furthermore, unrealistic expectations during competition may lower the rider s threshold for frustration, giving rise to horse rider conflict [12] and, consequently, to misbehaviour. It is also possible that the threshold for rider perception of misbehaviour during stressful situations, such as competition, is also lower. Potential explanations for the observed increased risk of misbehaviour with 12 or more hours of exercise per month are difficult to identify, particularly in light of the observed benefit of more frequent riding days, which is consistent with evidence that an accumulation of positive interactions reinforces the development of a positive relationship [10]. However, this latter association may also be due to reverse causation, with children being less likely to choose to ride horses that have previously misbehaved. Nutrition Odds of misbehaviour on exercise days were more than twice as high when supplements (roughage, concentrates and/or grain) were fed daily compared with feeding on 2 days per week or less. On average, supplemental feed accounted for 49% of the horses daily DE requirements, and horses were fed supplements containing DE in excess of daily DE requirements in 10% of horse months [8] in addition to daily access to pasture. Despite the observations and perceptions of horse owners, any association between the nonstructural carbohydrate content of ration and horse behaviour remains largely speculative [13,14]. In our study, roughage was fed in the largest amounts of any supplements, up to 10.6 kg per day and contributing a mean of 41.5% to horses daily DE requirements [8]. However, the percentage contribution to daily DE requirement of horses of roughage, concentrates and grain feeding was not associated with increased risk of misbehaviour on univariable analysis, but the mean amount of grain fed on any day was small (0.3 kg) and possibly insufficient to affect horse behaviour adversely. In contrast, access to pasture with greater ground coverage of green grass and moderate amounts of legumes was associated with increased risk of misbehaviour, and this effect was independent of any supplements fed. Whether this reflects an effect of pasture content or quantity remains unclear. Interestingly, Spanish researchers demonstrated a reduction in the intensity of horses startle responses by increasing the fat content of an otherwise isocaloric diet in a mixed cohort of 28 dressage horses [15]. We did not measure dietary fat content in the present study. Daily digestible energy intake above maintenance requirements unmatched by daily exercise results in surplus energy. Based on the increased risk of misbehaviour with daily feeding of supplements, and in fat to obese body condition scores (see Body condition score below), it was hypothesised that this surplus energy is expended in nonessential physical activities, including misbehaviour. It may also result in horses becoming more reactive to external stimuli, with consequent misbehaviour. Body condition score Body condition scores of study horses varied seasonally in association with pasture availability, with marked increases from winter to spring, followed by a gradual decline to the next winter [8]. This pattern suggests that owners decisions to feed horses supplements were not based on pasture growth curves or on ideal body condition scores. There is little scientific information about the effects of supplements on horses body condition scores. In one study, Thoroughbred geldings fed high-fibre and high-fat diets accumulated higher fat scores at the neck and withers than geldings fed a diet high in nonstructural carbohydrates, but this did not change their overall body condition score [16]. Although high body condition score has been related to increased risk of diseases, such as laminitis [17 20], it has not previously been identified as a risk of horse misbehaviour. Risk of misbehaviour was higher in horses that were fat and obese. These results suggest that a link exists between nutrition, body condition scores and misbehaviour, where higher body condition scores reflect dietary intake exceeding requirements and infrequent exercise, and fatter horses are 11

Misbehaviour in Pony Club horses P. Buckley et al. TABLE 2: Descriptive statistics and univariable odds ratios for putative risk factors for misbehaviour during riding with univariable P values <0.2, for 75 Pony Club horses in the Wagga Wagga district of New South Wales, Australia, from June 2000 to July 2001 Variable Number of exercise days Number of days with misbehaviour during riding Percentage of exercise days with misbehaviour during riding Odds ratio 95% confidence interval P value* Exercise frequency (days in horse month) <0.01 >0 and 5 563 31 5.5% Reference category >5 and 10 1006 47 4.7% 0.6 (0.4 1.1) 0.09 >10 and 15 899 30 3.3% 0.4 (0.2 0.7) <0.01 >15 2371 25 1.1% 0.2 (0.1 0.4) <0.01 Exercise type on exercise day <0.01 Leisure 2313 21 0.9% Reference category Schooling 2137 59 2.8% 3.5 (1.9 6.3) <0.01 Competition 389 53 13.6% 1.9 (1.2 2.9) 0.01 Total riding time (h in horse month) <0.01 <4 744 28 3.8% Reference category 4 and <7 1404 20 1.4% 0.5 (0.2 0.8) 0.01 7 and <12 1041 22 2.1% 0.5 (0.3 1.0) 0.06 12 1643 63 3.8% 1.4 (0.8 2.6) 0.2 Not recorded 7 0 Supplementary feeding frequency (days/week) <0.01 2 847 19 2.2% Reference category 3 6 547 12 2.2% 0.6 (0.2 1.4) 0.24 7 3303 101 3.1% 2.5 (1.2 5) 0.01 Not recorded 142 1 Average time exercised on exercise days (h/exercise day) <0.01 1 1820 45 2.5% Reference category >1 and 2 1786 40 2.2% 0.6 (0.4 1.0) 0.05 >2 and 3 726 24 3.3% 0.8 (0.5 1.6) 0.60 >3 501 24 4.8% 1.6 (0.9 2.8) 0.12 Not recorded 6 0 Roughage fed (% of daily digestible energy requirement) 0.02 <20 953 22 2.3% Reference category 20 and <40 1286 38 3.0% 1.4 (0.7 2.5) 0.32 40 and <60 1098 28 2.6% 1.5 (0.7 2.9) 0.27 60 1138 40 3.5% 2.9 (1.4 5.6) <0.01 Not recorded 364 5 Available pasture (kg dry matter/hectare) 0.03 <700 1255 37 2.9% Reference category 700 and <1400 1142 26 2.3% 0.9 (0.5 1.7) 0.83 1400 and <2100 679 34 4.8% 2.0 (1.1 3.7) 0.03 2100 1305 27 2.1% 0.9 (0.5 1.7) 0.80 Not recorded 458 9 Legume availability (% of ground area covered by legumes) 0.03 0 1623 46 2.8% Reference category >0 and <10 1582 28 1.8% 0.7 (0.4 1.2) 0.22 10 1180 48 4.1% 1.5 (0.9 2.5) 0.10 Not recorded 454 11 Green grass availability (% of ground area covered by green grass) 0.04 <10 1084 34 3.1% Reference category 10 and <25 1259 25 2.0% 0.6 (0.3 1.1) 0.09 25 and <50 1373 41 3.0% 1.3 (0.7 2.2) 0.40 50 659 22 3.3% 1.6 (0.8 3.2) 0.17 Not recorded 464 11 Grain fed during horse month 0.04 0 3139 108 3.4% Reference category >0 1336 20 1.5% 0.5 (0.2 1.0) 0.04 Not recorded 364 5 Lameness 0.05 No 3120 92 2.9% Reference category Yes 1439 32 2.2% 1.1 (0.4 2.8) 0.05 Not recorded 280 9 Body condition score 0.08 3 1929 43 2.2% Reference category >3 and 3.5 792 17 2.1% 0.5 (0.3 1.1) 0.08 >3.5 and 4 1160 39 3.4% 1.3 (0.7 2.5) 0.47 >4 665 25 3.8% 1.4 (0.6 2.9) 0.44 Not recorded 293 9 Age of horse (years) 0.09 <6 753 23 3.1% Reference category 6 and <11 1094 37 3.4% 2.2 (0.6 8.5) 0.26 11 and <16 1739 55 3.2% 1.2 (0.3 4.5) 0.76 16 1194 15 1.3% 0.5 (0.1 2.0) 0.29 Not recorded 59 3 Horse experience of rider s most experienced parent 0.11 None 106 6 5.7% 0.3 (0.0 1.8) 0.17 Pony Club committee member 1690 49 2.9% 0.4 (0.1 2.6) 0.34 Pony Club instructor 2257 71 3.1% 0.1 (0.0 0.8) 0.03 Experienced # 786 7 0.9% Reference category *Bold values are overall likelihood ratio test P values; nonbold (non-hyphen) values are Wald P values for individual coefficients. Average or total for horse month in which exercise days occurred. At end of horse month in which exercise days occurred. At start of horse month in which exercise days occurred. # Parent actively and regularly riding and/or handling horses. 12 Equine Veterinary Journal 45 (2013) 9 14 2012 EVJ Ltd

P. Buckley et al. Misbehaviour in Pony Club horses TABLE 3: Adjusted odds ratios and 95% confidence intervals and P values from a multivariable model of risk factors for misbehaviour during riding in 75 Pony Club horses in the Wagga Wagga district of NSW, Australia, from June 2000 to July 2001 Variable Adjusted odds ratio 95% confidence interval P value* Exercise frequency (days in horse month) <0.01 >0 and 5 Reference category >5 and 10 0.5 (0.2 1.1) 0.07 >10 and 15 0.2 (0.1 0.4) <0.01 >15 0.0 # (0.0 0.1) <0.01 Exercise type (on exercise day) 0.03 Leisure Reference category Schooling 1.6 (1.0 2.6) 0.07 Competition 2.4 (1.2 4.8) 0.01 Total riding time (h in horse month) <0.01 <4 Reference category 4 and <7 0.9 (0.4 2.2) 0.83 7 and <12 1.6 (0.6 4.2) 0.38 12 8.6 (3.2 23.0) <0.01 Supplementary feeding frequency (days/week) <0.01 2 Reference category 3 6 0.5 (0.2 1.4) 0.19 7 2.4 (1.1 5.3) 0.03 Legume availability (% of ground area covered by legumes) 0.01 0 1.1 (0.6 2.0) 0.77 >0 and <10 0.5 (0.2 0.8) 0.01 10 Reference category Green grass availability (% of ground area covered by green grass) 0.01 <10 Reference category 10 and <25 0.6 (0.3 1.2) 0.18 25 and <50 1.7 (0.8 3.3) 0.14 50 2.6 (1.2 5.9) 0.02 Body condition score 0.02 3 1.6 (0.8 3.3) 0.18 >3 and 3.5 Reference category >3.5 and 4 2.8 (1.3 6.1) 0.01 >4 3.1 (1.3 7.5) 0.01 *Bold values are likelihood ratio test P values; nonbold values are Wald P values for individual coefficients. Average or total for horse month in which exercise days occurred. At end of horse month in which exercise days occurred. At start of horse month in which exercise days occurred. # Actual odds ratio was 0.040 (95% CI 0.014 0.116), but with rounding to one decimal place this appears as 0.0 (95% CI 0.0 0.1). more likely to misbehave. If so, body condition score could be used as a measure of adequacy of diet and exercise regimen, and high scores may be a warning of increased misbehaviour risk in Pony Club horses. Back pain and lameness Misbehaviour can be a nonspecific clinical sign of back pain in horses [21]. Long-term exposure of horses to aggressive or poor riding techniques can predispose horses to back problems [11], with related pain that may manifest as misbehaviour [1]. The difficulties in diagnosing back pain have been noted [22]. The present misbehaviour study used only veterinarydiagnosed back pain and lameness events in the analyses; ownerdiagnosed events were excluded. To improve the specificity of veterinary diagnosis of back pain and lameness diagnoses further, only obvious cases that were detected without ancillary aids following the routine outlined in the Materials and Methods were included. In this study, no association was demonstratedbetweeneitherbackpainorlamenessandmisbehaviour, but it is possible that misclassification of exposure occurred; if nondifferential, Equine Veterinary Journal 45 (2013) 9 14 2012 EVJ Ltd odds ratios for any true association would have been biased towards 1, i.e. towards no observed effect. Such misclassification could occur due to both imperfect sensitivity and specificity of veterinary diagnoses, and because instances of back pain and lameness may have commenced and resolved between monthly veterinary visits. Non-included risk factors Misbehaviour remained moderately clustered within horse, after accounting for the exposure variables in the final model. The model fit and predictive ability of the final model are consistent with the study population being exposed to additional risk factors for misbehaviour that were not included in the final model. The level of horse experience within each study family and the duration of horse ownership were assessed, but were not significant risk factors in the present study. Further insights into horse misbehaviour may be gained by investigating the complex relationship between horse and rider. This may include horse temperament, level of horse education and fitness for the task, and rider ability and experience. Widths of confidence intervals Many of the confidence intervals for odds ratios in the present study were quite wide. One consequence of unduly wide confidence intervals is failure to detect true effects: only strongly associated factors are likely to be detected where confidence intervals are wide. Accordingly, we have made no conclusions about the absence of effect for any risk factors assessed in the present study. In contrast, there is no reason to disregard factors that are clearly associated with misbehaviour solely because the odds ratio confidence intervals are wide. Conclusions In conclusion, horse nutrition- and exercise-related factors are associated with increased risk of misbehaviour in horses during riding. In populations such as the study population the risk of misbehaviour is higher in fatter horses, in horses with access to pastures with greater green grass cover, in those fed supplements daily, in horses receiving exercise less frequently, and during competition. Our results highlight the importance of considering horse body condition, nutrition and exercise in any investigation of horse misbehaviour. Recommendations, including exercising horses at least 3 times a week and maintaining an ideal body condition score of 3 by more closely matching pasture and supplemental feed intake to horses energy requirements, could be used by veterinarians to assist owners of Pony Club horses to prevent misbehaviour. Authors declaration of interests No conflicts of interest have been declared. Source of funding This project was funded in part by Charles Sturt University, New South Wales, Australia and the Rural Industries Research and Development Corporation. Acknowledgements The authors would like to thank participating Pony Club members (Zone 12 Pony Clubs, New South Wales, Australia), Garry Anderson, Charles Sturt University and the Rural Industries Research and Development Corporation (Project Number UCS 24A). Manufacturers addresses a Ruddweigh International Scale Company Pty Ltd, 270 Falconer Rd, Guyra, New South Wales 2365, Australia. b StataCorp LP, 4905 Lakeway Drive, College Station, Texas 77845, USA. References 1. McGreevy, P.D. (2004) Equine Behavior: A Guide for Veterinarians and Equine Scientists, Saunders, Sydney. 13

Misbehaviour in Pony Club horses P. Buckley et al. 2. Burger, D., Bachmann, I. and Poncet, P.A. (2008) Applied research on the behaviour of horses. Agrarforschung 15, 172-177. 3. Buckley, P., Dunn, T. and More, S.J. (2004) Owners perceptions of the health and performance of Pony Club horses in Australia. Prev. Vet. Med. 63, 121-133. 4. Cripps, R.A. (2000) Horse-related injury in Australia. Australian Injury Prevention Bulletin: National Injury Surveillance Unit, 1-20. http://nisu.flinders.edu.au/pubs/bulletin24/bulletin24.pdf 5. Christey, G.L., Nelson, D.L., Rivara, F.P., Smith, S.M. and Condie, C. (1994) Horseback riding injuries among children and young adults. J. Fam. Pract. 39, 148-152. 6. Williams, F. and Ashby, K. (1995) Horse related injuries. Victorian Injury surveillance system: Hazard 23, 1-13. http://monashuniversity.mobi/miri/ research/research-areas/home-sport-and-leisure-safety/visu/hazard/haz23.pdf 7. Huntington, P.J. (1991) Field estimation of body condition and weight. Equine Nutr. Proc. Comm. Vet. Sci. 181, 15-23. 8. Buckley, P. (2009) Epidemiological Studies of Health and Performance of Pony Club Horses. PhD Thesis, University of Queensland, Brisbane. 9. Hosmer, D.W. and Lemeshow, S. (2000) Applied Logistic Regression, John Wiley and Sons, New York. pp 147-200. 10. Hausberger, M., Roche, H., Henry, S. and Visser, E.K. (2008) A review of the human-horse relationship. Appl. Anim. Behav. Sci. 109, 1-24. 11. Denoix, J.-M. and Pailloux, J.-P. (1997) Physical Therapy and Massage for the Horse, Manson Publishing, London. 12. McGreevy, P.D. (2007) The advent of equitation science. Vet. J. 174, 492-500. 13. Harris, P. (2005) Nutrition, behaviour and the role of supplements for calming horses: the veterinarian s dilemma. Vet. J. 170, 10-11. 14. Ralston, S.L. (2007) Evidence-based equine nutrition. Vet. Clin. N. Am.: Equine Pract. 23, 365-384. 15. Redondo, A.J., Carranza, J. and Trigo, P. (2009) Fat diet reduces stress and intensity of startle reaction in horses. Appl. Anim. Behav. Sci. 118, 69-75. 16. Suagee, J.K., Burk, A.O., Quinn, R.W., Petersen, E.D., Hartsock, T.G. and Douglass, L.W. (2008) Effects of diet and weight gain on body condition scoring in thoroughbred geldings. J. Equine Vet. Sci. 28, 156-166. 17. Buckley, P., Morton, J. and Coleman, G.T. (2007) A longitudinal study of naturally occurring laminitis in Pony Club horses in regional Australia. Proceedings of the Australian College of Veterinary Scientists Scientific Meeting (Equine Chapter), Gold Coast, Australia. 18. King, C. and Mansmann, R.A. (2004) Preventing laminitis in horses: dietary strategies for horse owners. Clin. Tech. Equine Pract. 3, 96-102. 19. Hoffman, R.M., Boston, R.C., Stefanovski, D., Kronfeld, D.S. and Harris, P.A. (2003) Obesity and diet affect glucose dynamics and insulin sensitivity in Thoroughbred geldings. J. Anim. Sci. 81, 2333-2342. 20. Carter, R., Treiber, K., Cubitt, T., Staniar, B., Harris, P. and Geor, R. (2007) Regional fat distribution in equines is associated with metabolic variables and laminitis predisposition. FASEB J. 21, A1058. 21. Martin, B.B. (1999) Physical examination of horses with back pain. Vet. Clin. N. Am.: Equine Pract. 15, 61-70. 22. Haussler, K.K. and Erb, H.N. (2006) Mechanical nociceptive thresholds in the axial skeleton of horses. Equine Vet. J. 38, 70-75. 14 Equine Veterinary Journal 45 (2013) 9 14 2012 EVJ Ltd