• 1.

    Hind KBurrows M. Weight-bearing exercise and bone mineral accrual in children and adolescents: a review of controlled trials. Bone 2007; 40:1427.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2.

    Gunter KBaxter-Jones ADGMirwald RL, et al. Impact exercise increases BMC during growth: an 8-year longitudinal study. J Bone Miner Res 2008; 23:986993.

    • Search Google Scholar
    • Export Citation
  • 3.

    Firth EC. The response of bone, articular cartilage and tendon to exercise in the horse. J Anat 2006; 208:513526.

  • 4.

    Firth ECRogers CW. Musculoskeletal responses of 2-year-old Thoroughbred horses to early training. Conclusions. N Z Vet J 2005; 53:377383.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5.

    Parkin TDH. Epidemiology of racetrack injuries in racehorses. Vet Clin North Am Equine Pract 2008; 24:119.

  • 6.

    Ely ERPrice JSSmith RK, et al. The effect of exercise regimens on racing performance in National Hunt racehorses. Equine Vet J 2010; 42:624629.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7.

    Verheyen KLPPrice JSWood JLN. Exercise during training is associated with racing performance in Thoroughbreds. Vet J 2009; 181:4347.

  • 8.

    Barneveld Avan Weeren PR. Conclusions regarding the influence of exercise on the development of the equine musculoskeletal system with special reference to osteochondrosis. Equine Vet J Suppl 1999;(31):112119.

    • Search Google Scholar
    • Export Citation
  • 9.

    Rogers CWFirth ECMcIlwraith CW, et al. Evaluation of a new strategy to modulate skeletal development in Thoroughbred performance horses by imposing track-based exercise during growth. Equine Vet J 2008; 40:111118.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10.

    Meredith KBolwell CFRogers C, et al. The use of allied health therapies on competition horses in the North Island of New Zealand. N Z Vet J 2011; 59:123127.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11.

    Rogers CWFirth ECMcIlwraith CW, et al. Evaluation of a new strategy to modulate skeletal development in racehorses by imposing track-based exercise during growth: the effects on 2 and 3 year old racing careers. Equine Vet J 2008; 40:119127.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12.

    Rogers CWKidd LFirth EC. Linear and temporal changes in the trot of 2-year-old Thoroughbred racehorses in relation to early exercise and race training. Comp Exerc Physiol 2010; 7:6571.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13.

    Bolwell CFRogers CWFirth EC, et al. Management and exercise of Thoroughbred yearlings during preparation for yearling sales in North Island of New Zealand, in Proceedings. 70th N Z Soc Anim Prod Conf 2010;157161.

    • Search Google Scholar
    • Export Citation
  • 14.

    Gibbs PGCohen ND. Early management of race-bred weanlings and yearlings on farms. J Equine Vet Sci 2001; 21:279283.

  • 15.

    Bolwell CFRogers CWFrench NP, et al. Exercise in Thoroughbred yearlings during sales preparation: a cohort study. Equine Vet J 2012; 44:2024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16.

    More SJ. A longitudinal study of racing Thoroughbreds: performance during the first years of racing. Aust Vet J 1999; 77:105112.

  • 17.

    Bailey CJReid SWHodgson DR, et al. Factors associated with time until first race and career duration for Thoroughbred racehorses. Am J Vet Res 1999; 60:11961200.

    • Search Google Scholar
    • Export Citation
  • 18.

    Tanner JCRogers CWFirth EC. The relationship of training milestones with racing success in a population of Standardbred horses in New Zealand. N Z Vet J 2011; 59:323327.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Bolwell CFRogers CWFrench N, et al. Risk factors for interruptions occurring before the first trial start of 2-year-old Thoroughbred racehorses in training. N Z Vet J 2012; 60:241246.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20.

    Perkins NRReid SWJMorris RS. Profiling the New Zealand Thoroughbred racing industry. 1. Training, racing and general health patterns. N Z Vet J 2004; 53:5968.

    • Search Google Scholar
    • Export Citation
  • 21.

    Dohoo IMartin WStryhn H. Modelling survival data. In: Dohoo IMartin WStryhn H, eds. Veterinary epidemiologic research. Charlottetown, PE, Canada: AVC Inc, 2003;439.

    • Search Google Scholar
    • Export Citation
  • 22.

    Collett D. Model checking in the Cox regression model: assessment of model fit. In: Collett D, ed. Modelling survival data in medical research. 2nd ed. Boca Raton, Fla: Chapman and Hall/CRC Press, 2003;122.

    • Search Google Scholar
    • Export Citation
  • 23.

    Dohoo IMartin WStryhn H. Modelling survival data. In: Dohoo IMartin WStryhn H, eds. Veterinary epidemiologic research. Charlottetown, PE, Canada: AVC Inc, 2003;438.

    • Search Google Scholar
    • Export Citation
  • 24.

    Bolwell CFRogers CWFrench N, et al. The effect of interruptions during training on the time to the first trial and race start in Thoroughbred racehorses. Prev Vet Med 2012; 105:in press.

    • Search Google Scholar
    • Export Citation
  • 25.

    Murray RCWalters JMSnart H, et al. Identification of risk factors for lameness in dressage horses. Vet J 2010; 184:2736.

  • 26.

    Hobbs SJLicka TPolman R. The difference in kinematics of horses walking, trotting and cantering on a flat and banked 10 m circle. Equine Vet J 2011; 43:686694.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 27.

    Davies HMSMerritt JS. Surface strains around the midshaft of the third metacarpal bone during turning. Equine Vet J 2004; 36:689692.

    • Search Google Scholar
    • Export Citation
  • 28.

    Verheyen KLHenley WEPrice JS, et al. Training-related factors associated with dorsometacarpal disease in young Thoroughbred racehorses in the UK. Equine Vet J 2005; 37:442448.

    • Search Google Scholar
    • Export Citation
  • 29.

    Reed SRJackson BFMcIlwraith CW, et al. Descriptive epidemiology of joint injuries in Thoroughbred racehorses in training. Equine Vet J 2012; 44:1319.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 30.

    Ely ERAvella CSPrice JS, et al. Descriptive epidemiology of fracture, tendon and suspensory ligament injuries in National Hunt racehorses in training. Equine Vet J 2009; 41:372378.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 31.

    AVMA Equine injury reporting system debuts at racetracks. J Am Vet Med Assoc 2007; 231:190.

Advertisement

Associations between yearling exercise and interruptions during race training in Thoroughbred racehorses

Charlotte F. BolwellInstitute of Veterinary, Animal and Biomedical Sciences, College of Sciences, Massey University, Palmerston North 4474, New Zealand.

Search for other papers by Charlotte F. Bolwell in
Current site
Google Scholar
PubMed
Close
 PhD
,
Christopher W. RogersInstitute of Veterinary, Animal and Biomedical Sciences, College of Sciences, Massey University, Palmerston North 4474, New Zealand.

Search for other papers by Christopher W. Rogers in
Current site
Google Scholar
PubMed
Close
 PhD
,
Nigel P. FrenchInstitute of Veterinary, Animal and Biomedical Sciences, College of Sciences, Massey University, Palmerston North 4474, New Zealand.

Search for other papers by Nigel P. French in
Current site
Google Scholar
PubMed
Close
 PhD
, and
Elwyn C. FirthNational Research Centre for Growth and Development, College of Sciences, Massey University, Palmerston North 4474, New Zealand.

Search for other papers by Elwyn C. Firth in
Current site
Google Scholar
PubMed
Close
 DSc

Abstract

Objective—To investigate the effect of exercise during yearling sales preparation on the risk of interruptions during training in Thoroughbred racehorses.

Animals—114 Thoroughbred racehorses.

Procedures—Information regarding the daily exercise of yearlings during sales preparation was obtained prospectively from a convenience sample of stud farms. Yearlings were followed to entry into race training, and subsequently, daily training information was recorded until the end of the racing season. Competing-risks survival analysis was used to model time from entry into race training to voluntary training interruption (no known condition or disease identified) and time from entry into race training to involuntary training interruption (due to presence of a condition or disease) occurring before the first trial (practice race for education). Total hand walking time and mechanical walker time accumulated during sales preparation were the main exposures of interest.

Results—82 of 114 (71.9%) horses had an interruption before the first trial; 65 (79%) interruptions were voluntary, and 17 (21 %) interruptions were involuntary. Increased total hand walking time was significantly associated with decreased risk of voluntary interruptions, whereas longer cumulative distances at a canter were significantly associated with decreased risk of involuntary interruptions.

Conclusions and Clinical Relevance—Results identified an association between early exercise during sales preparation and decreased risk of voluntary interruption and increased risk of involuntary interruption during training of 2-year-old Thoroughbred racehorses. Further investigation into the effects of early exercise on racing performance is needed, but results have indicated that there may be an opportunity to modify early exercise programs.

Abstract

Objective—To investigate the effect of exercise during yearling sales preparation on the risk of interruptions during training in Thoroughbred racehorses.

Animals—114 Thoroughbred racehorses.

Procedures—Information regarding the daily exercise of yearlings during sales preparation was obtained prospectively from a convenience sample of stud farms. Yearlings were followed to entry into race training, and subsequently, daily training information was recorded until the end of the racing season. Competing-risks survival analysis was used to model time from entry into race training to voluntary training interruption (no known condition or disease identified) and time from entry into race training to involuntary training interruption (due to presence of a condition or disease) occurring before the first trial (practice race for education). Total hand walking time and mechanical walker time accumulated during sales preparation were the main exposures of interest.

Results—82 of 114 (71.9%) horses had an interruption before the first trial; 65 (79%) interruptions were voluntary, and 17 (21 %) interruptions were involuntary. Increased total hand walking time was significantly associated with decreased risk of voluntary interruptions, whereas longer cumulative distances at a canter were significantly associated with decreased risk of involuntary interruptions.

Conclusions and Clinical Relevance—Results identified an association between early exercise during sales preparation and decreased risk of voluntary interruption and increased risk of involuntary interruption during training of 2-year-old Thoroughbred racehorses. Further investigation into the effects of early exercise on racing performance is needed, but results have indicated that there may be an opportunity to modify early exercise programs.

In young children, exercise may result in positive skeletal effects that remain in early adulthood.1,2 Within many fields of sport, such as soccer, hockey, and gymnastics, coaches have recognized the benefit of developing the appropriate skills at an early age, during growth and maturation. The mechanisms by which the musculoskeletal system responds to early training exercise in horses have been demonstrated and discussed.3,4 Various frequencies, intensities, and types of exercise during training of horses have been associated with reductions in the rate of musculoskeletal injury5 and with measures of improved performance during racing.6,7

For horses, the effects of exercise at an early age, before the start of race training, on the development of the musculoskeletal system have been reported.8 Early conditioning exercise of foals in addition to free exercise at pasture does not result in any harmful effects and has a stimulatory effect on components of the musculoskeletal system.9,10 In 2 studies,11,12 previously conditioned horses were shown to have an early training advantage, such as being ready for racing earlier, compared with unconditioned horses. However, the sample sizes in those studies11,12 were too small to enable identification of significant training-associated advantages, suggesting further investigation with a larger number of horses is required.

Within the racing industry, exercising young horses prior to race training occurs during the preparation of yearlings for sale.13,14 The median duration of sale preparations is 12 weeks (IQR, 12 to 16 weeks), and stud masters use methods of exercise such as hand walking and walking or trotting on mechanical horse walkers to train yearlings and to stimulate muscle development.13 The differences observed in the type and amount of exercise during sales preparation13,15 suggest that because the exposure to exercise varies among young horses, these differences in early exercise could impact future athletic performance.

Performance during the early years of training has been associated with better racing success and longevity among Thoroughbred racehorses.16,17 Results of a recent study18 in Standardbred racehorses indicate the importance of having horses in race training for the first time as 2-year-olds rather than at a later age, suggesting that the identification of training factors (eg, exercise and age) that can advance horses' capabilities to train and race is beneficial.

Given the body of evidence to support the benefits of early conditioning exercise, the purpose of the study reported here was to investigate the effect of exercise during sales preparation on the risk of interruptions occurring before the first recognized industry milestone (trial) during training for a cohort of Thoroughbred racehorses. Competing-risks analysis was used to examine the effect of hand walking and walker exercise in yearlings on the risk of voluntary interruptions (no known condition or disease identified) and involuntary interruptions (due to the presence of a condition or disease) to training before the first trial. We hypothesized that the cumulative durations of hand walking and mechanical horse-walker exercise provided to horses during sales preparation would be associated with interruptions during their training as 2-year-olds.

Materials and Methods

Data were collected as part of a prospective investigation of the effect of early exercise on future training and racing performance of Thoroughbreds born in the 2007 foal crop in New Zealand. The investigation was conducted in 2 parts: data were collected during yearling sales preparation, which have been reported,15 and during training as 2-year-old horses. Briefly, information on the daily exercise of yearlings during sales preparation was obtained from a convenience sample of 18 Thoroughbred stud farms, from September 29, 2008, until January 25, 2009. Stud masters recorded the type and duration of exercise and the number of hours spent in pasture. An additional 6 stud farms at which horses were not exercised as part of preparation provided the sire and dam names of their sales yearlings so those horses could be tracked to trainers. The stud masters also provided a list of nonsale horses (classified as stores), but these horses were not included in any analysis in the present study.

The yearlings enrolled in the first part of the investigation were then tracked to determine whether they were registered with a trainer (by New Zealand Thoroughbred Racing), and several methods were used to determine the location of the horses. The sire and dam names of the enrolled horses were checked against export lists of named and unnamed horses published by New Zealand Thoroughbred Racing. Additionally, bloodstock agencies and purchasers who had bought yearlings and stud farms that were vendors of the yearlings at the sales were contacted to identify the current location of the yearlings. Lastly, online searches of the New Zealand Thoroughbred Racing database provided information to establish whether a horse was currently registered with a trainer. These methods were used until all the enrolled horses were located or until the study ended on July 31, 2010. Horses that were exported overseas or registered with trainers in the South Island of New Zealand were not tracked further.

After a horse enrolled in the first part of the investigation was identified as being registered with a trainer, the trainer was contacted and invited to participate in the next part of the project. Some trainers were already recording data for horses enrolled in an ongoing study,19 and other trainers were contacted by phone or email. Data regarding daily training activity were recorded from the first day the horse entered the trainer's yard until July 31, 2010 (end of the 2010 racing season), or until the horse was lost to follow-up. Horses that were sold to overseas buyers or reassigned to trainers not already enrolled in the investigation were considered lost to follow-up. The type of data recorded and the methods of data collection were described as part of a prospective study investigating risk factors for race training interruptions in a different cohort of horses.19 Briefly, the daily training activity, including the distances exercised at canter (> 13.3 m/s [ie, > 15 seconds to travel 200 m]) and fast speeds (13.3, 15.4, and 16.7 m/s [ie, 15, 13, and 12 seconds to travel 200 m, respectively]), and reasons for not exercising were recorded on standardized forms or were provided directly through copies of the trainers' records. A voluntary interruption was a break in training attributed to no known condition or disease; an involuntary interruption was a break in training attributed to the presence of a condition or disease. Data were collected in paper format, except for those from 1 trainer who provided copies of an electronic spreadsheeta and a second trainer who provided information from records by telephone on a weekly basis. Information on reasons for horses not being exercised was provided by trainers and not veterinarians.

Statistical analysis—Data were entered into a customized database,b and random subsamples of the data set were checked for data entry errors. Once all the data were entered, a customized checklist was followed to look for outliers and inconsistencies within the data. Data were changed only when the entry in the database deviated from the paper records.

Competing-risks survival analysis was used to model the time from entering training to the first voluntary interruption (no known condition or disease present) that occurred before the first trial and time from entering training to the first involuntary interruption (due to the presence of a condition or disease) that occurred before the first trial. Horses that were away from the training yard or not exercised for > 7 consecutive days were considered not at risk, and these periods were classed as an interruption (any type) to the training program.20

Data were arranged as multiple records per horse, each representing a training day, and in each model, the other outcome event was censored on the day that the event took place. Only the first interruption that occurred was considered for inclusion in the analysis. The main exposures of interest were the total time spent undergoing hand walking exercise and total time spent undergoing mechanical horse-walker exercise during yearling sales preparation, which were calculated for each horse. The number of hours on pasture during yearling sales preparation could not be used in the models because this information was not provided by the stud farms that were not exercising horses. The training exposure variables consisted of the exercise distances (at canter [> 13.3 m/s], at 13.3 m/s, and at 15.4 m/s), number of exercise events (high-speed events [at 16.7 m/s], swimming [horse swam in equine pool], or use of water walker), number of jump outs (practice leaving the starting gates at speed), and the number of days off (out of training) accumulated since entering training. The units of the yearling and training exercise variables were converted to hours and furlongs (1 furlong = 201.168 m) to allow better interpretation of the hazard ratios. The other exposure variables investigated were age (months) at the start of training and sex. Stud farm was modeled as a fixed effect, and trainer was modeled as a shared frailty term to adjust for potential clustering of horse within trainers.

Exposure variables were considered for inclusion in multivariable models at a cutoff of P < 0.20. As hand walking and mechanical horse-walker exercise times were the main exposures of interest, these variables were forced into the multivariable models if the Wald test P value was greater than the cutoff. A backward selection procedure was followed, retaining exposure variables in the model on the basis of a combination of Wald test and LRS P values < 0.05. The functional form of continuous variables was investigated by visually assessing the martingale residuals as described by Dohoo et al.21 The proportional hazards assumption was tested for each of the final models, and the model fits were visually assessed through plots of Cox-Snell residuals.22,23 A sensitivity analysis was performed to investigate informative censoring whereby all censored observations were recoded and assumed to have experienced the event on the day of censoring. All analyses were conducted with statistical software,c and values of P < 0.05 were considered significant.

Results

Of the 490 yearlings (on 24 stud farms) that were officially catalogued for the sales in the first part of the investigation, 11 were misclassified as stores at 1 farm and no exercise information was recorded during the sales preparation. These 11 horses were excluded from further analysis, so that 479 horses were included in the population of sales horses. Of these sales horses, 258 horses were officially recorded by New Zealand Thoroughbred Racing as 2 year olds that were registered with a trainer; an additional 25 horses were reported by the trainer as being in training. Thirty-four trainers agreed to take part, recording information for 135 of 283 horses that were registered with trainers. On the first visit to 3 trainers, data had not been accurately recorded on a daily basis, so information from these trainers (10 horses) was excluded from further analysis. Another 4 horses that were with trainers enrolled in the study were excluded because the trainers did not want to record information. One horse that was registered with a trainer died as a result of a paddock accident before entering training, and 6 horses did not enter training before the study ended.

Overall, data were recorded by 26 trainers for 114 horses enrolled in both parts of the investigation, which provided 6,788 days at risk. Of the 114 horses, 74 (65%) were males and 67 (59%) were exercised as part of the yearling sales preparation. The median age at the start of training was 21 months (IQR, 19 to 23 months). Twenty-four horses had short (median, 12 days; IQR, 11 to 15 days) pretraining preparations before entering training; the first start date for these horses was when they entered the training stable for the second time after their pretraining preparation.

Eighty-two horses had an interruption in race training before the first trial, of which 65 (79%) were voluntary and 17 (21%) were involuntary. Most (33/65 [51%]) voluntary interruptions did not have a specific reason cited; rather, they were breaks between exercise preparations considered as part of the program. Of the specific reasons cited by trainers for 32 voluntary interruptions, 14 (44%) were for growth and strengthening of the horses. Nine of the 17 involuntary interruptions were cited as being due to shin soreness. Median time from entering training to a voluntary interruption was 60 days (95% CI, 52 to 77 days); the median time from entering training to an involuntary interruption could not be calculated because > 50% of the cohort did not have an involuntary interruption. Twenty-eight horses reached their first trial without an interruption, 3 horses were lost to follow-up before their first trial, and 1 horse was censored at the end of the study. Descriptive statistics for the continuous exercise variables considered in the competing-risks analysis were summarized (Table 1).

Table 1—

Descriptive statistics for total durations of mechanical horse-walker and hand walking yearling exercise and race training exercise accumulated since entering training among a cohort of one hundred fourteen 2-year-old Thoroughbred racehorses (26 trainers) in New Zealand.

VariableMedianIQR*MinimumMaximum
Total time on mechanical horse-walker (h)00–0019
Total time hand walking (h)0.80–10024
Cumulative distance at canter (> 13.3 m/s [furlongs])321134–69002,400
Cumulative distance at 13.3 m/s (furlongs)00–3026
Cumulative distance at 15.4 m/s (furlongs)00–0018
Cumulative high-speed events at 16.7 m/s (furlongs)00–0026
Cumulative days off (out of training)52–11037
Cumulative swim events00–009
Cumulative jump outs§00–003
Cumulative water walker events00–0026

Data regarding daily training activity were recorded from the first day the horse entered the trainer's yard until July 31, 2010 (end of the 2010 racing season), or until the horse was lost to follow-up.

Includes 25th to 75th percentile.

1 furlong = 201.168 m.

Swim events occurred in an equine pool.

Jump outs involved practice out of the starting gates at speed.

Voluntary interruptions—The type of yearling exercise, sex, age at the start of training, number of days in which the water walker was used, and cumulative high-speed events (at 16.7 m/s) were not associated with voluntary interruptions in the univariable analysis. The exposure variables associated with voluntary interruptions that were significant at the cutoff of P < 0.20 were summarized (Table 2). Each additional hour of hand walking or time spent on the mechanical horse-walker significantly decreased the risk of voluntary interruptions. After adjusting for the cumulative distances at canter, number of days off, stud farm, and trainer in the final multivariable model, total mechanical horse-walker time was no longer associated with the chance of voluntary interruptions. The variables included in the final multivariable model for voluntary interruptions were summarized (Table 3). After adjusting for the other variables in the model, total hand walking time was significantly associated with decreased risk of voluntary interruptions. The risk of voluntary interruptions decreased as the cumulative distance at canter increased but increased significantly with each additional day off. Stud farm was significantly associated with voluntary interruptions, and there was significant clustering at the trainer level.

Table 2—

Associations of variables with voluntary and involuntary interruptions during training of the cohort of 2-year-old Thoroughbred racehorses in Table 1 as determined by use of univariable competing-risks analysis.

Type of interruptionVariableCoefficientHazard ratio95% CIWald P value*LRS P value*
VoluntaryTotal time hand walking (h)−0.0500.940.91 to 0.990.010.008
Total time on mechanical horse-walker (h)−0.0800.910.82 to 1.010.080.04
Cumulative days off (out of training)0.1001.111.04 to 1.190.0020.002
Cumulative distance at canter (furlongs)−0.0030.990.99 to 0.990.0040.002
Cumulative distance at 13.3 m/s (furlongs)0.0501.051.00 to 1.110.050.06
Cumulative swim events0.1101.120.97 to 1.300.110.14
Cumulative jump outs§0.4301.540.98 to 2.420.050.08
Stud farm (19 df)0.01
InvoluntaryAge at start of training (mo)0.0901.090.96 to 1.250.160.17
Total time hand walking (h)−0.0200.970.90 to 1.050.520.51
Total time on mechanical horse-walker (h)0.0501.050.95 to 1.160.310.35
Cumulative distance at canter (furlongs)−0.0050.990.99 to 0.990.010.01
Cumulative distance at 13.3 m/s (furlongs)0.1101.111.02 to 1.220.010.03
Stud farm (19 df)0.01

A voluntary interruption was a break in training attributed to no known condition or disease. An involuntary interruption was a break in training attributed to the presence of a condition or disease. Interruptions occurred before the first trial (practice race for education); only the first interruption that occurred was included in the analysis.

The critical level for assigning significance was P < 0.20.

— = Not applicable. df = Degrees of freedom.

See Table 1 for remainder of key.

Table 3—

Associations of variables with voluntary and involuntary interruptions during training of the cohort of 2-year-old Thoroughbred racehorses in Table 1 as determined by use of multivariable competing-risks analysis.

Type of interruptionVariableCoefficientHazard ratio95% CIWald P value*LRS P value*
VoluntaryTotal time hand walking (h)−0.5900.550.42 to 0.73< 0.0010.003
Cumulative days off (out of training)0.3101.371.18 to 1.58< 0.001< 0.001
Cumulative distance at canter (furlongs)−0.0080.990.98 to 0.99< 0.001< 0.001
Stud farm (19 df)0.003
Trainer< 0.001
InvoluntaryTotal time on mechanical horse-walker (h)0.1201.131.01 to 1.270.0400.06
Cumulative distance at canter (furlongs)−0.0070.990.98 to 0.990.0080.005
Trainer0.006

Trainer was modeled as frailty term variance (2.63 SE, 0.94 and 2.37 SE, 1.72 for voluntary and involuntary interruptions, respectively).

The critical level for assigning significance was P < 0.05.

See Tables 1 and 2 for remainder of key.

Involuntary interruptions—The exposure variables associated with involuntary interruptions that were significant at the cutoff of P < 0.20 were summarized (Table 2). The type of yearling exercise, sex, number of days in which the water walker was used, cumulative high-speed events (at 16.7 m/s), swim events, or jump outs were not associated with voluntary interruptions in the univariable analysis. Increasing total hand walking and mechanical horse-walker times were associated with decreased and increased risk of involuntary interruptions, respectively. Although the P values for total hand walking and mechanical horse-walker times were greater than the cutoff, these were the main exposures of interest and were considered for inclusion in multivariable models.

The variables included in the final multivariable model for involuntary interruptions were summarized (Table 3). After adjusting for the cumulative distance at canter, total mechanical horse-walker time, and trainer, total hand walking time was no longer associated with involuntary interruptions. Increasing total mechanical horse-walker time was associated with an increased risk of involuntary interruptions (Wald test P value only). Although the LRS P value indicated only potential improvement in the model fit, there was a stronger association of cumulative distance at canter, and clustering within trainer increased with the inclusion of this variable in the final model. As the cumulative distance at canter increased, horses were less likely to have an involuntary interruption. The frailty term for trainer was significant, indicating clustering at the trainer level.

Model fit—The assumption of proportion hazards was met in both the final models. The plots of the Cox-Snell residuals showed slight deviation. Refitting the final voluntary interruption model without stud farm resulted in a slightly better fit, indicating that the lack of fit might be due to the increased degrees of freedom associated with the inclusion of stud farm. The sensitivity analysis for informative censoring showed little change in the hazard ratios for the voluntary interruption model. Total mechanical horse-walker time was no longer significant in the final involuntary model, indicating that some informative censoring may be present.

Discussion

The aim of the present study was to investigate the effect of early exercise during yearling sales preparation on the risk of interruptions during race training of Thoroughbred racehorses. An earlier study19 revealed an association between race training exercise and interruptions during training. The present study expands on that research19 by investigating the effect of exposure of Thoroughbreds to exercise at an early age, before entering race training, on later performance. The results of the present study indicated that there was an association between hand walking or mechanical horse-walker exercise time and voluntary or involuntary interruptions during training.

Increasing hand walking exercise time was associated with a lesser likelihood of a voluntary interruption in the present study. Hand walking during sales preparation is most often given to train and to modify the behavior of the yearlings,13 and some stud managers use it to improve fitness of the horses. Yearlings that have been exercised may be or may appear to be more developed than those undertaking less hand walking and may be better able to withstand training without any voluntary interruptions. Results of a recent study24 indicate that both voluntary and involuntary interruptions increase the time to and reduce the likelihood of a trial and race start, and there is an association between performing trials and racing at 2 years of age and future racing performance and success.16–18 Elements within exercise programs during sales preparations represent factors that could be modified to reduce the proportion of horses with interruptions before their first trial. Before this could be done, possible associations between early exercise programs and racing success should be investigated.

Conversely, a greater amount of time spent exercising on a mechanical horse-walker was associated with an increased likelihood of an involuntary interruption from race training in the present study. Caution should be applied when interpreting this finding because the LRS P value only suggested that the inclusion of this variable might improve the model fit. The lack of significance in the final model may be due to a lack of power as a result of few involuntary interruptions. Data regarding the effect of mechanical horse-walkers on the benefits or potential negative effects on the musculoskeletal system, in a horse of any age, are limited. A retrospective study25 in dressage horses revealed that increased mechanical horse-walker time was associated with risk of lameness, although those authors speculated that this may have been due to an effect of rehabilitation for lameness, rather than a cause of lameness. Observations of yearlings on a mechanical horse-walker have previously resulted in a decrease in exercise time, because of caution over the use of the walker.15 Different strains on the third metacarpal bone in horses and changes in kinematics have been observed during tight turns or circles.26,27 Whether the use of a mechanical horse-walker resulted in any underlying problems that led to involuntary interruptions during training cannot be determined from the findings of the present study. It is possible that the trainers perceived these yearlings as being more developed for training, in terms of musculoskeletal tissues, movement, and temperament, which resulted in imposition of a higher training load than for other horses that had undergone less mechanical horse-walker exercise and were less developed. The most common reason reported for involuntary interruptions was shin soreness, and exercise involving increasing distances at high speed within short time periods has been identified as a risk factor for dorsal metacarpal disease (a variety of pathological changes in the third metacarpal bone, including periosteal new bone formation).28 No association between high-speed events and involuntary interruptions was established by the analysis in the present study.

It has been hypothesized that the associations between sex and interruptions during training may be a reflection of the difference between colts and fillies in the cumulative amount of exercise during a sales preparation.15,19 The present study did not reveal any significant associations between sex and either involuntary or voluntary interruptions.

In the present study, the most common specific reason cited by trainers for voluntary interruptions was so a horse could grow and strengthen. The data obtained in the present study indicated that there was a significant association between the cumulative number of days off and voluntary interruptions. It is speculated that horses perceived as not coping well or not sufficiently strong would be given more days off but still kept in training. Similarly, increasing canter distance was associated with lesser risk of voluntary interruptions, suggesting that horses with greater cumulative canter distances may be perceived by the trainers as performing well (eg, apparent cooperation during training or indication of progression or improved fitness) and continue to be exercised. This association could be a result of being in training for a longer period than horses that had an interruption, rather than a cause of greater cumulative canter distances.

The percentage of horses reaching their first trial without an interruption during training in the present study was < 30%, as previously reported.19 These and other discrepancies between study findings, such as the lack of association between high-speed events or age with interruptions, may be due to differences in the sample population of the 2 studies and the inclusion of the early exercise exposures in the present study.

Stud farm was measured as a fixed effect in the models, which may have affected the generalizability of the results because it only measures the effects of the stud farms included in the study. Furthermore, the inclusion of stud farm as a fixed effect increased the degrees of freedom lost and would have influenced the SEs and CIs obtained in the models. However, the inclusion of stud farm as a fixed effect would have been unlikely to affect the conclusions regarding the main factors of interest in the models. Stud farm was associated with voluntary interruptions in the final model, and there was significant clustering in both final models at the trainer level in the present study. Unmeasured effects at the stud farm level, such as management and feeding practices (not recorded), and the production focus of the farm (eg, sprinters vs long-distance equine athletes) may have contributed to the association with stud farm. The unmeasured factors and clustering at the trainer level are common in studies6,7,19 of training and racing performance, and although it may represent management differences, such as the reasons cited for voluntary interruptions, the level of involvement from the owners should also be considered.

The present study was part of a larger, 2-part investigation. The first part of the investigation used a convenience sample of stud farms, the implications of which have been discussed.15 The sample of trainers in the present study was determined by the horses enrolled in part 1 of the investigation, and trainers were enrolled in the present study on the basis of the horses being registered with them. Given that trainers were not being recruited only on the basis of their willingness to participate, loss to follow-up from yearling sales to trainers was expected. As illustrated by the present study, even with enthusiasm and willingness to participate, the amount of work required to record training exercise did result in withdrawal of trainers and subsequent loss or exclusion of data, a common problem in other studies29,30 of this kind. Withdrawal of trainers from the present study reduced the size of the sample of horses followed; thus, the study may have lacked statistical power to investigate multiple risk factors and identify subtle effects. Furthermore, exclusions can result in selection bias because the trainers who did not participate may have differed in various aspects from those who remained in the study. Giving preference to trainers who already have recording systems in place may introduce selection bias29 but would improve compliance and the accuracy of the data. Although trainers could not be selected in this manner for the present study, several trainers already had recording systems in place. Movement in the direction of a central training database, as developed by The Jockey Club for racetrack injuries,31 would assist future studies and allow investigators to use larger, randomly selected samples of trainers.

Although the plots of the Cox-Snell residuals showed some lack of model fit for the data obtained in the present study, some departure toward the right-hand tail was expected because of a smaller sample resulting from the occurrence of events or censoring. Additionally, these plots can indicate a lack of model fit when there is a substantial proportion of censoring,21 as seen in the involuntary interruptions model. The sensitivity analysis for informative censoring did not indicate any changes in the hazard ratios for the voluntary interruption model, suggesting that the occurrence of an involuntary interruption did not inform a voluntary interruption. There was some indication of informative censoring in the involuntary model, but this analysis features a worst-case scenario that is unlikely to represent reality; given the reasons cited for voluntary interruptions, it is doubtful that all the horses censored for voluntary interruptions would have had an involuntary interruption the next day were they not censored.

In the present study, an association between early exercise during sales preparation and the hazard of a voluntary or involuntary interruption during training of 2-year-old Thoroughbred racehorses was identified. Although further investigation into the effects of early exercise on racing performance is needed, results of this study have suggested that modifying early exercise programs to optimize their effects could provide benefits during training of 2-year-old Thoroughbreds.

ABBREVIATIONS

Cl

Confidence interval

IQR

Interquartile range LRS Likelihood ratio statistic

a.

Microsoft Excel, Microsoft Corp, Redmond, Wash.

b.

Microsoft Access, Microsoft Corp, Redmond, Wash.

c.

Intercooled Stata, version 11, Statacorp LP, College Station, Tex.

References

  • 1.

    Hind KBurrows M. Weight-bearing exercise and bone mineral accrual in children and adolescents: a review of controlled trials. Bone 2007; 40:1427.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2.

    Gunter KBaxter-Jones ADGMirwald RL, et al. Impact exercise increases BMC during growth: an 8-year longitudinal study. J Bone Miner Res 2008; 23:986993.

    • Search Google Scholar
    • Export Citation
  • 3.

    Firth EC. The response of bone, articular cartilage and tendon to exercise in the horse. J Anat 2006; 208:513526.

  • 4.

    Firth ECRogers CW. Musculoskeletal responses of 2-year-old Thoroughbred horses to early training. Conclusions. N Z Vet J 2005; 53:377383.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5.

    Parkin TDH. Epidemiology of racetrack injuries in racehorses. Vet Clin North Am Equine Pract 2008; 24:119.

  • 6.

    Ely ERPrice JSSmith RK, et al. The effect of exercise regimens on racing performance in National Hunt racehorses. Equine Vet J 2010; 42:624629.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7.

    Verheyen KLPPrice JSWood JLN. Exercise during training is associated with racing performance in Thoroughbreds. Vet J 2009; 181:4347.

  • 8.

    Barneveld Avan Weeren PR. Conclusions regarding the influence of exercise on the development of the equine musculoskeletal system with special reference to osteochondrosis. Equine Vet J Suppl 1999;(31):112119.

    • Search Google Scholar
    • Export Citation
  • 9.

    Rogers CWFirth ECMcIlwraith CW, et al. Evaluation of a new strategy to modulate skeletal development in Thoroughbred performance horses by imposing track-based exercise during growth. Equine Vet J 2008; 40:111118.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10.

    Meredith KBolwell CFRogers C, et al. The use of allied health therapies on competition horses in the North Island of New Zealand. N Z Vet J 2011; 59:123127.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11.

    Rogers CWFirth ECMcIlwraith CW, et al. Evaluation of a new strategy to modulate skeletal development in racehorses by imposing track-based exercise during growth: the effects on 2 and 3 year old racing careers. Equine Vet J 2008; 40:119127.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12.

    Rogers CWKidd LFirth EC. Linear and temporal changes in the trot of 2-year-old Thoroughbred racehorses in relation to early exercise and race training. Comp Exerc Physiol 2010; 7:6571.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13.

    Bolwell CFRogers CWFirth EC, et al. Management and exercise of Thoroughbred yearlings during preparation for yearling sales in North Island of New Zealand, in Proceedings. 70th N Z Soc Anim Prod Conf 2010;157161.

    • Search Google Scholar
    • Export Citation
  • 14.

    Gibbs PGCohen ND. Early management of race-bred weanlings and yearlings on farms. J Equine Vet Sci 2001; 21:279283.

  • 15.

    Bolwell CFRogers CWFrench NP, et al. Exercise in Thoroughbred yearlings during sales preparation: a cohort study. Equine Vet J 2012; 44:2024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16.

    More SJ. A longitudinal study of racing Thoroughbreds: performance during the first years of racing. Aust Vet J 1999; 77:105112.

  • 17.

    Bailey CJReid SWHodgson DR, et al. Factors associated with time until first race and career duration for Thoroughbred racehorses. Am J Vet Res 1999; 60:11961200.

    • Search Google Scholar
    • Export Citation
  • 18.

    Tanner JCRogers CWFirth EC. The relationship of training milestones with racing success in a population of Standardbred horses in New Zealand. N Z Vet J 2011; 59:323327.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Bolwell CFRogers CWFrench N, et al. Risk factors for interruptions occurring before the first trial start of 2-year-old Thoroughbred racehorses in training. N Z Vet J 2012; 60:241246.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20.

    Perkins NRReid SWJMorris RS. Profiling the New Zealand Thoroughbred racing industry. 1. Training, racing and general health patterns. N Z Vet J 2004; 53:5968.

    • Search Google Scholar
    • Export Citation
  • 21.

    Dohoo IMartin WStryhn H. Modelling survival data. In: Dohoo IMartin WStryhn H, eds. Veterinary epidemiologic research. Charlottetown, PE, Canada: AVC Inc, 2003;439.

    • Search Google Scholar
    • Export Citation
  • 22.

    Collett D. Model checking in the Cox regression model: assessment of model fit. In: Collett D, ed. Modelling survival data in medical research. 2nd ed. Boca Raton, Fla: Chapman and Hall/CRC Press, 2003;122.

    • Search Google Scholar
    • Export Citation
  • 23.

    Dohoo IMartin WStryhn H. Modelling survival data. In: Dohoo IMartin WStryhn H, eds. Veterinary epidemiologic research. Charlottetown, PE, Canada: AVC Inc, 2003;438.

    • Search Google Scholar
    • Export Citation
  • 24.

    Bolwell CFRogers CWFrench N, et al. The effect of interruptions during training on the time to the first trial and race start in Thoroughbred racehorses. Prev Vet Med 2012; 105:in press.

    • Search Google Scholar
    • Export Citation
  • 25.

    Murray RCWalters JMSnart H, et al. Identification of risk factors for lameness in dressage horses. Vet J 2010; 184:2736.

  • 26.

    Hobbs SJLicka TPolman R. The difference in kinematics of horses walking, trotting and cantering on a flat and banked 10 m circle. Equine Vet J 2011; 43:686694.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 27.

    Davies HMSMerritt JS. Surface strains around the midshaft of the third metacarpal bone during turning. Equine Vet J 2004; 36:689692.

    • Search Google Scholar
    • Export Citation
  • 28.

    Verheyen KLHenley WEPrice JS, et al. Training-related factors associated with dorsometacarpal disease in young Thoroughbred racehorses in the UK. Equine Vet J 2005; 37:442448.

    • Search Google Scholar
    • Export Citation
  • 29.

    Reed SRJackson BFMcIlwraith CW, et al. Descriptive epidemiology of joint injuries in Thoroughbred racehorses in training. Equine Vet J 2012; 44:1319.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 30.

    Ely ERAvella CSPrice JS, et al. Descriptive epidemiology of fracture, tendon and suspensory ligament injuries in National Hunt racehorses in training. Equine Vet J 2009; 41:372378.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 31.

    AVMA Equine injury reporting system debuts at racetracks. J Am Vet Med Assoc 2007; 231:190.

Contributor Notes

Supported by New Zealand Racing Board as part of the Equine Partnership for Excellence.

The authors thank Geoff Jones for statistical advice.

Address correspondence to Dr. Bolwell (c.bolwell@massey.ac.nz).