Musculoskeletal injuries are the major cause of racehorse death and attrition, with numerous potential contributing factors.1 Epidemiologic evidence indicates that horse characteristics (age, sex, and performance quality),2–10 training and racing history,4–6,9,a hoof management,11 horseshoe characteristics,12 preexisting musculoskeletal injuries,13 racetrack characteristics (geometry, condition, and surface),5,8–10,14–16 and race features (eg, class of race and purse)3,4,17 may be associated with risk for injury. Factors that can be altered are attractive candidates for further study because management of true risk factors is likely to reduce injuries. Racetrack surface is one factor that is intensively managed and is practical to change via modification of surface material properties. However, epidemiologic relationships between track surface material and injuries are inconsistent.3,8–10,15,18–21,a Several factors could contribute to the disparate results including experimental design, analytic approach, injury and case definitions, and confounding factors. Clearly, the relationships between race surface materials and likelihood for injury prevention need further elucidation.
Forelimb hoof impact accelerations and GRFs are affected by race surface material properties.22–27 Consequently, race surface materials are likely to affect risk for injury by altering the load transfer from the ground to the hoof and the propagation of forces to bones, tendons, and ligaments of the limb. However, the ideal surface material properties for injury prevention are unknown.
The racehorse industry recognizes the potential for injury reduction through racetrack surface changes. Dirt surfaces have been replaced with synthetic materials at several racecourses over the past 3 years. Early anecdotal observations28 indicate a lower incidence of fatal injuries with the introduction of synthetic race surfaces in the United States. However, variation within racetracks with ambient temperature changes and variation among racetracks with different synthetic surfaces are also anecdotally reported.b The manufacture and selection of synthetic surface materials have been largely based on empirical evidence and marketing factors. Nevertheless, development of standards for race surface materials is needed to ensure consistency among racetracks, optimize horse performance, and minimize risk for injury. The first step is to understand the properties of existing race surface materials.
The objective of the study reported here was to compare hoof accelerations and GRFs during trot and canter among the 3 primary track surfaces in the United States: dirt, synthetic, and turf. A secondary goal was to advance the scientific evaluation of new track surfaces to aid in the prevention of musculoskeletal injuries to racehorses.
Materials and Methods
Study design—Hoof accelerations, GRFs, and horse speed were measured for 3 racehorses during trot and canter on a dirt racetrack, synthetic training track,c and turf racetrack at 1 racecourse. Thoroughbred racehorses were accustomed to the data collection equipment by application of dummy horseshoes and instrumentation cables during daily treadmill exercise for a period of 2 weeks prior to data collection. Data were collected at Keeneland Race Course, Lexington, Ky, during trot and canter on track surfaces while horses were ridden by exercise riders. On each day, data were collected from 1 horse traveling on a straight corridor between timing gates on all 3 surfaces. Because of track management practices, track availability and time constraints during normal daily racecourse routines, and transport of data collection equipment among surfaces, surface order could not be changed among horses. For the dirt and synthetic surfaces, measurements were obtained in the late morning on freshly harrowed dirt and synthetic surfaces. The synthetic surface evaluated was a proprietary mixture of wax-coated silica sand, polypropylene fibers, and recycled rubber. The turf track had long grass (approx 4 to 6 inches in height) in preparation for an upcoming race meet, and no special maintenance was performed before testing. Subjectively, track conditions were good to fast for the dirt racetrack and synthetic training track and yielding to firm for the turf racetrack. During data collection, the ambient temperature ranged from 21° to 29°C. Horse speed was limited to canter to minimize horse fatigue among data collection on all 3 track surfaces during 1 daily exercise session. Accelerations and GRFs were compared among track surfaces.
Animals—Three 3-year-old female Thoroughbred racehorses (mean ± SD mass, 486 ± 6.7 kg) were studied. All horses were clinically sound, with no subjectively observed gait abnormalities. The procedures used were approved by the Institutional Animal Care and Use Committee at the University of California-Davis. Owner consent was obtained.
Data collection—A triaxial piezoelectric-based accelerometerd with a dynamic range of ± 500g in each axis was screwed onto the dorsal part of the hoof wall of the left forelimb with known orientation relative to the hoof wall. To correct any misalignments and transform the accelerometer coordinate system to the hoof coordinate system, photographs were taken from the front and the side of the hoof.
A dynamometric (force-measuring) horseshoe weighing 860 g described previously29 was used for collection of GRF data in 3 orthogonal directions. The hoof plate of the horseshoe was screwed onto the solar surface of the left forelimb hoof. A horseshoe of equal mass and geometry was fixed to the right forelimb hoof. An experienced farrier trimmed the hooves and attached the accelerometer and horseshoes. Epoxy was used to strengthen the attachment of the accelerometer to the hoof and to fill in areas of mismatch between the dorsal surface of the hoof and the outer ridge of the dynamometric horseshoe.
The dynamometric horseshoe (force shoe) and the accelerometer were connected via wires to a custom signal conditioning box located in a backpack worn by the exercise rider. Wires were taped to the forelimb of the horse. The signal conditioning box was connected to an analog-to-digital data acquisition carde located in a mini-notebook computer,f which was also contained in the exercise rider's backpack. Acceleration and force signals were triggered, acquired at 1,000 Hz, and recorded with custom software.g
A spherical kinematic marker was affixed to the skin over the dorsal spinous process of the sixth lumbar vertebra to calculate horse speed. Kinematic data were collected with a high-speed, 500-Hz video camerah for a 2-second period during trot and canter through a straight corridor on the straight portion of the training track or racetrack. A calibration frame (2.2 m wide and 1.9 m tall) was imaged within the corridor at the beginning of each data collection session on each surface. Instrumented horses were walked, trotted, and cantered through the data collection corridor until settled at each gait; the horses were limited to 3 to 4 trials of trot and 3 to 4 trials of canter to prevent fatigue. Trials were omitted if the horse bucked or if a lead or gait change was observed during the trial. Kinematic data collection was triggered by horse interference with an infrared beam on the first timing gatei crossed. The exercise rider manually triggered data collection from the accelerometer and force shoe, which were sampled synchronously at 1,000 Hz for 15 seconds. The experimental setup ensured that the first 5 strides of acceleration and GRF data would include the 1 to 3 strides of kinematic data.
Data processing—Accelerometer and force shoe data were filtered by use of a digital low-pass Butterworth filter with cutoff frequencies of 85 and 50 Hz, respectively.j Although the accelerations and GRFs were recorded for 3-D analysis, only 2 dimensions were analyzed in this study (Figure 1). For accelerations, the accelerometer coordinate system was transposedj to the hoof coordinate system by use of angles calculated from dorsal and lateral photographs of the hoof. The transformed acceleration coordinate system axes were perpendicular (Z-acceleration [acceleration perpendicular to the solar surface of the hoof in the sagittal plane]) and parallel (X-acceleration [acceleration parallel to the solar surface of the hoof in the sagittal plane]) to the solar surface of the hoof in the sagittal plane. For GRFs, the force shoe directly measured forces perpendicular (Z-force [GRF perpendicular to the solar surface of the hoof in the sagittal plane]) and parallel (X-force [GRF parallel to the solar surface of the hoof in the sagittal plane]) to the solar surface of the hoof in the sagittal plane. Acceleration and GRF data from the first 5 full strides were analyzed as repeated observations for each trial. Trials were omitted from analysis if the first 5 strides were incomplete or if the data appeared excessively noisy. By use of motion analysis software,k the video data were digitized and filtered with a digital low-pass Butterworth filter with a cutoff frequency of 10 Hz. Horse speed was calculated as the mean speed of the kinematic marker on the dorsal spinous process of the sixth lumbar vertebra over a full stride. Stride duration was calculated as the time between successive hoof strikes or hoof landing regions for acceleration data. For all data sets (acceleration, GRF, and kinematics), stride frequency was calculated as the inverse of stride duration, giving units of strides per second.
Photograph of a horse's left forelimb with a triaxial piezoelectric accelerometer fixed to the dorsal part of the hoof wall to record accelerations and a dynamometric horseshoe fixed to the solar surface of the hoof to record GRFs. The accelerometer coordinate system was transformed to a hoof coordinate system. Arrows depict the positive Z- and X-axes of each coordinate system, which are fixed relative to the solar surface of the hoof.
Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1220
Variables—Because same-trial GRF data were not available for all acceleration data, and because the beginning of heel-strike and the end of toe-off were not easily identifiable points in acceleration versus time profiles, acceleration data were analyzed for 4 regions (Figure 2) by use of custom software.j The 4 regions, or phases, were designated HL, MSt, HTO, and MSw. The HL region encompasses accelerations before heel-strike through early stance, as shown by plotting acceleration and GRF profiles from the same strides. The accelerations before impact can be attributed to modifications of hoof position in preparation for hoof contact with the ground surface. The larger magnitude peaks are caused by impact and the hoof sliding and decelerating on the surface.30 The MSt region is the relatively constant portion of the profile following HL in which there is little movement of the hoof with respect to the ground. The HTO region begins as the hoof begins to rotate and the heel lifts. The Z-axis becomes principally horizontal as the hoof leaves the surface and the large positive Z-acceleration peak (HTO maximum) is caused by the hoof accelerating to the forward speed of the horse.30 The HTO region ends at the termination of the respective acceleration peaks: HTO maximum peak for Z-acceleration and HTO minimum peak for X-acceleration. There is some nonzero acceleration in the following MSw region because the hoof does not move at a constant velocity throughout swing. Maxima, minima, and temporal components of respective regions, in addition to stride frequency, were determined after manual selection of region boundaries; magnitudes were calculated relative to the mean baseline values during mid-stance for each stride to account for instrument drift, and temporal components were normalized to stride duration.
Representative Z-acceleration (upper graph, thick line), X-acceleration (upper graph, thin line), Z-force (lower graph, thick line), and X-force (lower graph, thin line) versus time plots for the trailing forelimb of a horse cantering in a straight line on a racetrack surface. The different regions of the plots used for analysis are delineated. StrD = Stride duration. Braking = Braking phase of X-force versus time plot. Propulsion = Propulsion phase of X-force versus time plot. ms = Milliseconds.
Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1220
Vibrational energy (vibration) was determined to quantify hoof oscillations during HL and HTO for 2 frequency ranges: 0 to 167 Hz and 167 to 333 Hz. Vibration was calculated from the frequency spectrum of hoof acceleration by first transforming unfiltered acceleration versus time data to the frequency domain by use of the Fast Fourier Transform. The absolute values of the transformed data were integrated for each frequency range to give vibration values that could be compared among surfaces. Higher vibration values correspond to greater hoof vibrations because vibrational energy increases as hoof vibrations increase in number, magnitude, or both. The 2 frequency ranges were chosen to capture and separate the lower and higher frequency vibrations because it has been suggested that vibrations in the 30- to 40-Hz range contribute to fatigue damage of equine tendon and bone.31 Because the unfiltered acceleration data greater than approximately 333 Hz were dominated by noise, values of vibration > 333 Hz were not compared among surfaces.
For GRF data during stance, custom softwarej was used to calculate peak forces (calculated relative to baseline of swing phase for each stride), impulses (integral of force versus time), stride frequency, and temporal components of each peak in the force data, where there was 1 characteristic peak/stride for Z-force and 2/stride (braking and propulsion) for X-force (Figure 2). In addition, a measure of loading rate (load rate) was calculated by dividing peak force by peak force rise time (time between the initiation of the peak and the peak force). Force and impulse were normalized to the total mass (sum of horse, rider, saddle, and instrumentation mass), giving units of Newtons per kilogram (N/kg) and Newton-seconds per kilogram (N•s/kg), respectively. Temporal components of the stride were normalized to stride duration.
Statistical analysis—Mixed-model ANOVA methodsl were used to assess the effect of track surface on hoof acceleration (canter data) and GRF (trot and canter data, separately) variables. A separate mixed-effect model was used for each dependent variable. All models included horse as the subject to take into account the fact that there were repeated measures on individual horses, and all models included a speed variable (stride frequency) to account for the effect of speed on dependent variable values. Because horse speed was not available for some acceleration and GRF data, stride frequency, which has been found to increase linearly with speed,32 was used as a covariate in acceleration and GRF statistical models. A linear regression was used to assess the relation between speed and stride frequency.m
Post hoc comparisons among track surfaces were based on least squares means. Because all models included date, horse, trial, and stride frequency, the least squares means represent mean responses that were adjusted for variation in all of these factors. The canter ANOVA statistical models also contained lead forelimb as a covariate to account for the different accelerations and forces experienced by the lead and trailing forelimb during canter.22,33 The Akaike information criterion aided in variable selection for each ANOVA model and in finding the best overall fit of each model.34 A value of P < 0.05 was considered significant.
Results
Forty-three trials of acceleration and GRF data were analyzed. The gaits analyzed were canter for acceleration data and canter and trot for GRF data. Stride frequencies calculated from acceleration and GRF data ranged from 1.3 to 2.1 strides/s (mean ± SD, 1.7 ± 0.2 strides/s). Same-trial acceleration, GRF, and kinematic data were not always available because of data omission. Speed and stride frequency were calculated from kinematic data of the same trials whenever a full stride of video data was available (28/43 trials); kinematic data were obtained from horse 1 on the synthetic surface and horses 2 and 3 on all 3 surfaces. These stride frequencies were plotted versus speed by surface for trot and canter data (Figure 3). The observed relationships between stride frequency and speed were not significantly different among surfaces. Therefore, the relationship between stride frequency and speed for all surfaces was used to approximate speed from stride frequencies in the context of this study.
Illustration of stride frequency versus horse speed for 3 racetrack surfaces (dirt, synthetic, and turf) for trot and canter. The linear line of best fit is given for all data, trot data, and canter data. Coefficients and r2 values for all data are b0 = 1.2052, b1 = 0.0967, and r2 = 0.84; for trot data are b0 = 0.6626, b1 = 0.2766, and r2 = 0.78; and for canter data are b0 = 1.4136, b1 = 0.0661, and r2 = 0.50.
Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1220
Acceleration—Hoof acceleration data from 16 canter trials from 4 days and 3 horses were used for analysis (Figure 4; Table 1). Data were obtained from horse 1 on all 3 surfaces and from horses 2 and 3 on the synthetic and turf surfaces. Ambient temperatures during synthetic surface data collection ranged from 21° to 24°C. Stride frequency ranged from 1.7 to 2.1 strides/s (mean ± SD, 1.9 ± 0.1 strides/s).
Least squares mean ± SE acceleration variables (adjusted for date, horse, trial, stride frequency, and lead limb) for horses cantering on dirt, synthetic, and turf track surfaces.
| Variable | Z-acceleration | X-acceleration | ||||||
|---|---|---|---|---|---|---|---|---|
| Track | P value | Track | P value | |||||
| Dirt | Synthetic | Turf | Dirt | Synthetic | Turf | |||
| HL | ||||||||
| Maximum (g) | 35.3 ± 5.8a,b | 28.5 ± 2.9a | 42.9 ± 3.8b | 0.007 | 8.7 ± 3.4a | 17.2 ± 1.7b | 15.6 ± 2.2a,b | 0.077 |
| Minimum (g) | −22.7 ± 4.8 | −15.2 ± 2.3 | −21.8 ± 3.1 | 0.111 | −36.7 ± 7.4 | −32.0 ± 4.3 | −35.2 ± 5.4 | 0.745 |
| Duration over StrD (%) | 28.7 ± 1.7 | 29 ± 0.8 | 30.5 ± 1.1 | 0.406 | 28.8 ± 1.8 | 29.1 ± 0.9 | 29.3 ± 1.2 | 0.973 |
| Vibration (0–167 Hz; g•Hz) | 2,788 ± 713a,b | 1,938 ± 307a | 4,198 ± 422b | 0.001 | 2,744 ± 1,116a,b | 2,409 ± 511a | 4,539 ± 693b | 0.039 |
| Vibration (167–333 Hz; g•Hz) | 5,071 ± 1,237a | 3,035 ± 542a | 8,296 ± 741b | < 0.001 | 4,930 ± 1,930a | 3,436 ± 1,013a | 9,723 ± 1,325b | 0.001 |
| HTO | ||||||||
| Maximum (g) | 52.8 ± 8.4a | 41.2 ± 7.5b | 44.6 ± 8.0a,b | 0.079 | 9.6 ± 1.1 | 8.3 ± 0.5 | 10.0 ± 0.7 | 0.125 |
| Minimum (g) | –13.9 ± 3.1a | −9.5 ± 2.7b | −5.2 ± 2.9a,b | 0.259 | −39.4 ± 6.3 | −31.8 ± 4.4 | −39.1 ± 5.2 | 0.189 |
| Duration over StrD (%) | 19.8 ± 1.0 | 20.9 ± 0.5 | 21.1 ± 0.6 | 0.531 | 24.2 ± 1.9a | 26.8 ± 1.0a | 30.4 ± 1.3b | 0.014 |
| Vibration (0–167 Hz; g•Hz) | 1,627 ± 503 | 1,088 ± 222 | 827 ± 303 | 0.339 | 1,385 ± 607 | 1,057 ± 288 | 765 ± 387 | 0.616 |
| Vibration (167–333 Hz; g•Hz) | 2,508 ± 499 | 1,654 ± 217 | 2,010 ± 297 | 0.218 | 2,688 ± 628 | 1,547 ± 286 | 1,944 ± 388 | 0.211 |
| MSt | ||||||||
| Duration over StrD (%) | 14.2 ± 1.4 | 15.7 ± 0.8 | 15.8 ± 1.0 | 0.534 | 11.4 ± 1.2a | 14.6 ± 0.6b | 14.3 ± 0.8b | 0.063 |
StrD = Stride duration.
Surface values within a variable row and within an acceleration (Z or X) with different superscripts are significantly (P < 0.05) different.
Illustration of Z-accelerations (thick lines) and X-accelerations (thin lines) of a horse cantering on a dirt surface, synthetic surface, or turf surface at a mean ± SD stride frequency of 1.79 ± 0.03 strides/s, which corresponds to a velocity of approximately 6 m/s. Approximately 2 strides, starting with HL, are illustrated. See Figure 1 for key.
Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1220
Z-acceleration—Maximum and minimum peak Z-acceleration magnitudes during canter were predominantly smaller for the synthetic surface than for the dirt or turf surfaces (Table 1). The HL maximum for the synthetic surface was 81% and 66% of those for the dirt and turf surfaces, respectively, but only the synthetic and turf surfaces were significantly different. The HTO maximum and minimum peak magnitudes were significantly larger for the dirt surface than for the synthetic surface. The Z-acceleration HL, HTO, and MSt durations as a percentage of stride duration were not significantly different among surfaces.
Most Z-acceleration vibrations during canter were smaller for the synthetic surface than for the dirt or turf surfaces, but only vibrations during HL were significantly different among surfaces (Table 1). The HL vibration for the turf surface was significantly larger than (more than double) that for the synthetic surface for both frequency ranges and was also significantly larger than that for the dirt surface in the 167- to 333-Hz range. The HTO vibration for the dirt surface was nearly double that of the turf surface in the 0- to 167-Hz range, but HTO vibrations were not significantly different among surfaces.
X-acceleration—Only the HL maximum was significantly larger for the synthetic surface than for the dirt surface (Table 1). The HL vibrations were significantly larger for the turf surface than for the synthetic surface for both frequency ranges. The HTO duration was a significantly smaller percentage of stride duration for the dirt and synthetic surfaces than for the turf surface. The MSt duration was significantly smaller for the dirt surface than for the synthetic and turf surfaces.
Z-force—For Z-force, data from 27 trials from 2 days and 2 horses were used for analysis (Figure 5; Table 2). Of these 27 trials, there were 13 trials of canter and 14 trials of trot. For both canter and trot, data were obtained from horse 2 on all 3 surfaces and from horse 3 on the dirt and synthetic surfaces. Ambient temperatures during synthetic surface data collection ranged from 27° to 29°C for force data. The stride frequency ranged from 1.3 to 2.1 strides/s (mean ± SD, 1.6 ± 0.2 strides/s) overall, from 1.4 to 2.1 strides/s (1.8 ± 0.1 strides/s) for canter, and from 1.3 to 1.6 strides/s (1.5 ± 0.1 strides/s) for trot. For canter data, all Z-force variables examined had significant differences among surfaces (Table 2). Peak Z-force for the synthetic surface was 83% and 71% of those for the dirt and turf surfaces, respectively, and the synthetic surface value was significantly different from the dirt and turf values, which were significantly different from each other (Figure 6). The impulse was significantly smaller for the synthetic surface than for the dirt and turf surfaces. The temporal components (as a percentage of stride duration) of peak force rise time and Z-force duration (or stance duration) were significantly smaller for the turf surface than for the dirt and synthetic surfaces, and Z-force duration was significantly smaller for the synthetic surface than for the dirt surface. The Z-force load rate for the turf surface was significantly larger than (> 170%) those for the dirt and synthetic surfaces.
Least squares mean ± SE Z-force variables (adjusted for date, horse, trial, stride frequency, and lead limb [canter]) for horses cantering and trotting on dirt, synthetic, and turf track surfaces.
| Variable | Canter | Trot | ||||||
|---|---|---|---|---|---|---|---|---|
| Track | P value | Track | P value | |||||
| Dirt | Synthetic | Turf | Dirt | Synthetic | Turf | |||
| Z-force | ||||||||
| Maximum (N/kg) | 13.8 ± 0.3a | 11.5 ± 0.4b | 16.1 ± 0.7c | < 0.001 | 13.5 ± 0.2a | 10.7 ± 0.2b | 13.0 ± 0.3a | < 0.001 |
| Impulse (N•s/kg) | 1.82 ± 0.06a | 1.33 ± 0.07b | 1.67 ± 0.12a | 0.004 | 2.73 ± 0.04a | 2.10 ± 0.05b | 2.89 ± 0.07a | < 0.001 |
| Peak force rise time over StrD (%) | 22.2 ± 1.2a | 19.7 ± 1.4a | 14.7 ± 2.4b | 0.073 | 28.2 ± 0.5a | 24.6 ± 0.6b | 23.1 ± 1.0b | < 0.001 |
| Duration over StrD (%) | 37.4 ± 0.5a | 35.7 ± 0.5b | 29.8 ± 1.1c | < 0.001 | 48.3 ± 0.5 | 47.5 ± 0.6 | 49.3 ± 1.0 | 0.255 |
| Load rate (N/kg/s) | 111.4 ± 7.8a | 106.4 ± 9.0a | 193.3 ± 15b | 0.002 | 70.1 ± 1.9a | 64.3 ± 2.1a | 81.8 ± 3.4b | 0.003 |
StrD = Stride duration.
Surface values within a variable row and within a gait (canter or trot) with different superscripts are significantly (P < 0.05) different.
Illustration of Z-forces (thick lines) and X-forces (thin lines) of a horse cantering on a dirt surface, synthetic surface, or turf surface at a mean ± SD stride frequency of 1.77 ± 0.03 strides/s, which corresponds to a velocity of approximately 6 m/s. Approximately 2 strides are illustrated. See Figure 2 for key.
Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1220
Illustration of peak Z-force versus stride frequency (trot and canter data combined) for 3 racetrack surfaces for trials of 2 horses (black symbols vs white symbols). Notice that the regression lines illustrate the difference in peak Z-force magnitude among surfaces over a range of similar speeds (stride frequencies).
Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1220
For trot data, nearly all Z-force variables examined had significant differences among surfaces (Table 2). Peak Z-force for the synthetic surface was 79% and 82% of those for the dirt and turf surfaces, respectively. Peak Z-force and impulse were significantly smaller for the synthetic surface than for the dirt and turf surfaces, which were not different from each other. Peak force rise time was a significantly smaller percentage of stride duration for the synthetic and turf surfaces than for the dirt surface. The Z-force load rate was significantly larger for the turf surface than for the dirt and synthetic surfaces.
X-force—For X-force, data from 22 trials from 2 days and 2 horses were used for analysis (Table 3). There were fewer X-force trials than Z-force trials because of malfunction of the × channel cable of the force shoe. Of these 22 trials, there were 11 trials of canter and 11 trials of trot. For both canter and trot, data were obtained from horse 2 on all 3 surfaces and from horse 3 on the dirt surface. The stride frequency ranged from 1.3 to 2.1 strides/s (mean ± SD, 1.6 ± 0.2 strides/s) overall, from 1.4 to 2.1 strides/s (1.8 ± 0.2 strides/s) for canter, and from 1.3 to 1.6 strides/s (1.5 ± 0.1 strides/s) for trot. For canter data, the X-force braking maximum and impulse were not significantly different among surfaces. The braking peak force rise time and duration were a significantly smaller percentage of stride duration for the synthetic and turf surfaces than for the dirt surface. The braking load rate for the turf surface was significantly larger than (> 140%) those for the dirt and synthetic surfaces. The propulsion maximum magnitude was significantly larger for the synthetic and turf surfaces than for the dirt surface.
Least squares mean ± SE X-force variables (adjusted for date, horse, trial, stride frequency, and lead limb [canter]) for horses cantering and trotting on dirt, synthetic, and turf track surfaces.
| Variable | Canter | Trot | ||||||
|---|---|---|---|---|---|---|---|---|
| Track | P value | Track | P value | |||||
| Dirt | Synthetic | Turf | Dirt | Synthetic | Turf | |||
| X-force: braking | ||||||||
| Maximum (N/kg) | 4.7 ± 0.3 | 4.8 ± 0.7 | 4.8 ± 0.7 | 0.986 | 3.0 ± 0.1a | 2.4 ± 0.1b | 1.1 ± 0.2c | < 0.001 |
| Impulse (N•s/kg) | 0.48 ± 0.04 | 0.38 ± 0.08 | 0.33 ± 0.09 | 0.285 | 0.52 ± 0.02a | 0.35 ± 0.03b | 0.10 ± 0.04c | < 0.001 |
| Peak force rise time over StrD (%) | 14.1 ± 0.5a | 8.5 ± 1.0b | 6.9 ± 1.2b | 0.001 | 18.1 ± 0.6a | 14.2 ± 0.9b | 13.8 ± 1.1b | 0.029 |
| Duration over StrD (%) | 31.1 ± 0.6a | 25.2 ± 1.3b | 22.5 ± 1.4b | < 0.001 | 39.6 ± 0.6a | 34.6 ± 0.9b | 30.3 ± 1.2c | < 0.001 |
| Load rate (N/kg/s) | 62.8 ± 6.0a | 88.3 ± 12.4a | 125.2 ± 14.2b | 0.003 | 24.2 ± 0.7a | 24.3 ± 1.0a | 13.1 ± 1.4b | < 0.001 |
| X-force: propulsion | ||||||||
| Maximum (N/kg) | −1.3 ± 0.1a | −2.3 ± 0.3b | −2.5 ± 0.3b | 0.010 | −1.4 ± 0.1a | −2.1 ± 0.1b | −3.0 ± 0.2c | < 0.001 |
| Impulse (N•s/kg) | −0.022 ± 0.007 | −0.058 ± 0.014 | −0.055 ± 0.016 | 0.102 | −0.038 ± 0.004a | −0.082 ± 0.006b | −0.150 ± 0.007c | < 0.001 |
| Peak force rise time over StrD (%) | 2.4 ± 0.5 | 4.4 ± 1.0 | 4.0 ± 1.1 | 0.275 | 2.9 ± 0.3a | 5.5 ± 0.4b | 9.6 ± 0.6c | < 0.001 |
| Duration over StrD (%) | 6.8 ± 0.9 | 8.9 ± 1.9 | 6.0 ± 2.2 | 0.418 | 10.4 ± 0.6a | 14.3 ± 0.8b | 22 ± 1.1c | < 0.001 |
| Load rate (N/kg/s) | −108.2 ± 8.0 | −93.7 ± 16.5 | −121.0 ± 18.9 | 0.379 | −65.3 ± 12.0a,b | −83.6 ± 16.0a | −43.9 ± 20.4b | 0.213 |
Surface values within a variable row and within a gait (canter or trot) with different superscripts are significantly (P < 0.005) different.
See Table 2 for remainder of key.
For trot data, the X-force braking maximum and impulse were significantly larger for the dirt surface than for the synthetic and turf surfaces and the synthetic surface values were significantly larger than those for the turf surface. The braking peak force rise time and duration were a significantly smaller percentage of stride duration for the synthetic and turf surfaces than for the dirt surface, and the duration was a significantly smaller percentage for the turf surface than for the synthetic surface. The braking load rate for the turf surface was significantly smaller than (< 55%) those for the dirt and synthetic surfaces. The propulsion maximum magnitude, absolute value of impulse, peak force rise time, and duration were all significantly larger for the synthetic and turf surfaces than for the dirt surface, and the turf surface values were significantly larger than those for the synthetic surface. The propulsion load rate was significantly smaller in magnitude for the turf surface than for the synthetic surface.
Discussion
Significant differences among dirt, synthetic, and turf track surfaces at 1 racecourse for hoof accelerations and GRFs were found for Thoroughbred racehorses instrumented with a hoof accelerometer and a dynamometric horseshoe. Key findings were that most peak hoof accelerations were smaller for the synthetic surface than for the dirt and turf surfaces. Hoof vibrations were largest for the turf surface during HL and largest for the dirt surface during HTO. Peak Z-forces and load rates were smallest for the synthetic surface.
The synthetic surface generally had the lowest peak magnitudes for Z-acceleration. Consequently, the hoof received less shock, or a less violent impact, on the synthetic surface than on the dirt and turf surfaces. The effective mass, or the actual mass that is subjected to this shock, is dependent on the kinematics of the horse at impact. Much of the hoof impact–related shock is likely dampened by the musculoskeletal structures of the forelimb and might only be received distal to the metacarpophalangeal (fetlock) joint.31,35 However, during HL, peak Z-acceleration results paralleled peak Z-force data; this relation indicates that peak Z-acceleration may serve as an index for the peak GRF, which is more closely related to loading of forelimb musculoskeletal structures.
The synthetic surface often had the lowest Z-acceleration and X-acceleration hoof vibrations during HL and HTO. In general, larger values of vibration mean that the amount of oscillatory movement of the hoof and its tissues is higher. Excessive vibrations may lead to injury because repetitive impulse loading causes damage to subchondral bone and articular cartilage in rabbits.36,37 Vibration in the 30- to 40-Hz range has been implicated as contributing to fatigue damage of equine tendon and bone.31 Therefore, the 0- to 167-Hz frequency range vibrations may be more biologically relevant than those in the 167- to 333-Hz range. The high vibrations for the turf surface during HL could be attributed to an irregular surface contour associated with residual indentations from previous hoof strikes and varying compliance caused by moisture variation across distance. These factors may have contributed to reflexive hoof adjustments and vibrations during hoof impact before attainment of a stable hoof-turf interaction. Compaction of the more consistent surface layer of the synthetic and dirt surfaces could dampen initial impact before high forces are transmitted to the limb. Overall, the synthetic surface appeared to provide the most stable interface for the hoof.
Because the synthetic surface had the lowest peak Z-forces and load rates, it likely lessens the potential for injury if all other factors that contribute to injury remain constant. High peak GRFs and load rates are most likely associated with high forces on musculoskeletal structures and thus higher probability of injury because these forces are the largest contributor to limb loading. Surfaces with high hardness and frictional properties are believed to impart high forces to the distal portions of the limb and predispose horses to injury.24,38
There was evidence that the distribution of load between forelimbs and hind limbs differed among surfaces. Different surfaces had different peak Z-force and impulse values at the trot, although they had similar stance durations as a percentage of stride. By the principle of conservation of momentum, during steady-state motion, the total vertical impulse applied to the horse during a stride must equal the product of body weight and stride duration.39 Although impulse values in this study were not truly vertical because of the rotating force shoe coordinate system, the proportionally lower forelimb Z-force and impulse measured on the synthetic surface indicated that the hind limbs are likely to sustain proportionally higher forces and impulses. Any redistribution of force could result in a redistribution of injuries.
Although the Z-force canter and trot data were similar among surfaces for most variables, the X-force variables were less similar between canter and trot data. The X-force variables for canter and trot data differed mainly because the forelimb at the trot, the lead forelimb at the canter, and the trailing forelimb at the canter all have different horizontal interactions with the surface and thus have differing braking-propulsion loading ratios.33,40–42 Note that the canter least squares means data are values adjusted between left and right lead mean values.
The X-force results provide insight into the shear and frictional properties of the surfaces. For trot data, the dirt surface had the largest braking peak, which indicated it had the highest shear strength during impact; the turf surface had the lowest braking peak, which indicated it had lower shear strength during impact that may allow the hoof to slip on the surface. Larger braking and propulsion load rates may indicate a surface with higher frictional properties. The braking load rate for the turf surface at the canter was nearly 10 times that at the trot. It is speculated that at lower speeds (trot), the hoof only penetrates superficial layers of the turf surface. At higher speeds (canter), the hoof likely breaks through superficial grass and also interacts with underlying layers composed of dirt and roots. In contrast, the dirt and synthetic surfaces are more consistent through a greater depth, and their increases in braking load rate from trot to canter are not as large as that for the turf surface. Results revealed that surface properties and the relative differences among surfaces varied with horse speed.
The GRF profiles differed between surfaces (Figure 5). The Z-force profile for the turf surface often had a transient force spike before the force peak. A force spike was sometimes present in dirt surface GRF profiles and rarely present in synthetic surface GRF profiles. This spike was similar to the heel-strike transient described in human studies, which varies in magnitude among individuals but also varies with speed, footwear, and ground surface.43 This force spike can generate substantial forces on the interior structures of the limb and is suggested to contribute to bone and joint disorders such as degenerative joint disease.43 The presence and absence of these force spikes may reflect the difference between having a soft top layer, which cushions the initial impact, and having a firm top layer, which is less deformable and imparts a larger shock to the limb. It is suggested that just as viscoelastic materials in human footwear can attenuate these potentially damaging force spikes,43 the material properties of track surfaces can protect a galloping horse against these forces.
Although the acceleration and GRF profiles from the present study were generally comparable to other studies,22,23,33,39,40,44–46 differences in profiles among studies could be attributable to the different speeds, surface characteristics, horseshoes,45,47 or methodologies. One of the primary methodologic differences among studies is the location of the measurement device and the resulting measurement coordinate system. In the present study, the measurement devices were attached to the hoof and the coordinate systems were relative (parallel and perpendicular) to the solar surface of the hoof. The orientation of these hoof coordinate systems changed continually with the orientation of the hoof. The changing orientation of the hoof relative to the surface must be known to resolve differences between hoof coordinate systems (eg, force shoe) and ground coordinate systems (eg, force plate), which are used to measure vertical and horizontal forces relative to the ground. Although a hoof coordinate system differs from a ground coordinate system, the hoof coordinate system results are more representative of accelerations and forces present at the level of the hoof. Consequently, Z-forces and X-forces in force shoe studies contribute to the horizontal braking and propulsion during stance. Discontinuities, such as the negative dip in Z-force at terminal stance, may be observed in force shoe profiles but not in force plate profiles. The discontinuity at terminal stance may be attributable to the ground plate of the force shoe being distracted from the hoof plate of the force shoe as the hoof rotates and pulls out and away from the surface.
The force shoe was useful for measuring forces between the hoof and track surfaces under natural conditions. The force shoe is a 6-load component dynamometer capable of recording forces, moments, and center of pressure for multiple consecutive strides. The force shoe records forces relative to the hoof instead of the ground, and the shape of the bottom of the force shoe mimics that of the solar surface of the hoof to replicate the actual loads applied.29 The force shoe likely affects the dynamics of the horse because its mass and thickness are greater than training or racing shoes. However, the same force shoe was used throughout the study, so the shoe was unlikely to affect the surface comparisons in the present study.
Two challenges were encountered in this study related to the acquisition of data under field conditions. The first was the requirement to collect data in the same order for all horses because of racetrack management practices. Consequently, any effect associated with time or trial order (eg, temperature) could have influenced the results. However, temporal effects were unlikely to have obscured track surface effects because surface results were not correlated with time. The second challenge was the need to collect data from a horse on all surfaces during the same day. To minimize horse fatigue, trot and canter trials were performed at slow speeds. Consequently, the data are reflective, but not representative, of race speeds.
The largest expected contributor to variation in the data was individual horse variability; therefore, data were collected on all 3 surfaces from the same horse within a day. Statistical approaches were used to overcome the other potential variability in the data. First, because horse speed could not be narrowly controlled and horse speed affects hoof accelerations and GRFs,22,25,39,48 statistical analyses accounted for varying horse speed by use of stride frequency as a speed surrogate. Although its use should be limited to approximating speeds from stride frequencies in the context of this study, a linear relationship between horse speed and stride frequency was supported by the results of a linear regression. Second, only the left forelimb was instrumented with the accelerometer and force shoe, but the horses did not always canter on the desired left lead. In contrast to the symmetrical trot, where a high degree of symmetry in loading exists between left and right forelimbs,40 in the asymmetrical gaits of gallop and canter, the trailing forelimb receives larger peak impact accelerations and vertical GRFs than the lead forelimb.22,33 The trailing forelimb has a smaller braking peak GRF and a larger magnitude propulsion peak GRF than the lead forelimb during canter.33 Because of sparse data, left and right lead canter data could not be independently analyzed. Thus, for canter trials, lead limb was included as a covariate in the statistical model to account for these differences. Consequently, acceleration and GRF least squares means reported for canter were reflective of an adjusted mean between the lead and trailing forelimb means.
Results from the present study are specific to the track surfaces and maintenance procedures at the time of testing at Keeneland Race Course. Unfortunately, physical properties of the surfaces were not measured by other means, making extrapolation of the data to other race surfaces with known properties tenuous. Although the testing was conducted within 2 weeks of heavy rain, the surfaces subjectively appeared to be of normal moisture content. Subjectively, track conditions were good to fast for the dirt racetrack and synthetic training track and yielding to firm for the turf racetrack. Additionally, because only forelimb data were considered in this study, surface effects on other parts of the horse, such as the hind limbs or overall kinematics, must be researched further before conclusions are made about the overall surface effect on horse biomechanics and its injury implications.
The synthetic surface had lower forelimb hoof accelerations, hoof vibrations, and Z-forces than the dirt and turf surfaces studied. Hoof forces contribute to forces incurred by bones, joints, tendons, and ligaments of the equine forelimb. Although causes of injuries to racehorses are multifactorial, the magnitude of the race surface effect observed in the present study indicated that design and management of race surfaces have large potential to reduce injury risk. Under similar training regimens, the synthetic surface studied is likely to transmit lower peak GRFs and accelerations to forelimb structures than the dirt and turf surfaces studied. However, because of the unique material properties and different nature of individual track surfaces, extending the results of this study to encompass all dirt, synthetic, and turf track surfaces should be done with caution.
ABBREVIATIONS
| GRF | Ground reaction force |
| HL | Hoof landing phase of stride |
| HTO | Hoof takeoff phase of stride |
| MSt | Mid-stance phase of stride |
| MSw | Mid-swing phase of stride |
Verheyen K. Epidemiology of fractures in Thoroughbred racehorses in training in the United Kingdom. PhD thesis, Department of Veterinary Basic Sciences, The Royal Veterinary College, University of London, London, England, 2005.
Peterson M, College of Engineering, The University of Maine, Orono, Me: Personal communication, 2007.
Polytrack, Martin Collins Surfaces and Footings LLC, Lexington, Ky.
Model 3023A1, Dytran Instruments Inc, Chatsworth, Calif.
DAQCard-AI-16E-4, National Instruments Corp, Austin, Tex.
Libretto U100, Toshiba America Inc, New York, NY.
LabVIEW, National Instruments Corp, Austin, Tex.
Fastcam PCI, Photron USA Inc, San Diego, Calif.
MEK Photoelectric Infrared Cells, Sircon Controls Ltd, Mississauga, ON, Canada.
MATLAB, The MathWorks Inc, Natick, Mass.
Peak Motus, version 9.0, Vicon Motion Systems, Centennial, Colo.
SAS, version 9.1, SAS Institute Inc, Cary, NC.
SigmaPlot, version 10.0, Systat Software Inc, San Jose, Calif.
References
- 1.↑
Stover SM. The epidemiology of Thoroughbred racehorse injuries. Clin Tech Equine Pract 2003;2:312–322.
- 2.
Estberg L, Stover SM, Gardner IA, et al. Fatal musculoskeletal injuries incurred during racing and training in Thoroughbreds. J Am Vet Med Assoc 1996;208:92–96.
- 3.
Estberg L, Stover SM, Gardner IA, et al. Relationship between race start characteristics and risk of catastrophic injury in Thoroughbreds: 78 cases (1992). J Am Vet Med Assoc 1998;212:544–549.
- 4.
Bailey CJ, Reid SW, Hodgson DR, et al. Risk factors associated with musculoskeletal injuries in Australian Thoroughbred racehorses. Prev Vet Med 1997;32:47–55.
- 5.
Henley WE, Rogers K, Harkins L, et al. A comparison of survival models for assessing risk of racehorse fatality. Prev Vet Med 2006;74:3–20.
- 6.
Perkins NR, Reid SW, Morris RS. Risk factors for injury to the superficial digital flexor tendon and suspensory apparatus in Thoroughbred racehorses in New Zealand. N Z Vet J 2005;53:184–192.
- 7.
Perkins NR, Reid SW, Morris RS. Risk factors for musculoskeletal injuries of the lower limbs in Thoroughbred racehorses in New Zealand. N Z Vet J 2005;53:171–183.
- 8.
Hernandez J, Hawkins DL, Scollay MC. Race-start characteristics and risk of catastrophic musculoskeletal injury in Thoroughbred racehorses. J Am Vet Med Assoc 2001;218:83–86.
- 9.
Mohammed HO, Hill T, Lowe J. Risk factors associated with injuries in Thoroughbred horses. Equine Vet J 1991;23:445–448.
- 10.
Mohammed HO, Hill T, Lowe J. The risk of severity of limb injuries in racing Thoroughbred horses. Cornell Vet 1992;82:331–341.
- 11.↑
Kane AJ, Stover SM, Gardner IA, et al. Hoof size, shape, and balance as possible risk factors for catastrophic musculoskeletal injury of Thoroughbred racehorses. Am J Vet Res 1998;59:1545–1552.
- 12.↑
Kane AJ, Stover SM, Gardner IA, et al. Horseshoe characteristics as possible risk factors for fatal musculoskeletal injury of Thoroughbred racehorses. Am J Vet Res 1996;57:1147–1152.
- 13.↑
Hill AE, Stover SM, Gardner IA, et al. Risk factors for and outcomes of noncatastrophic suspensory apparatus injury in Thoroughbred racehorses. J Am Vet Med Assoc 2001;218:1136–1144.
- 14.
Evans DL, Walsh JS. Effect of increasing the banking of a racetrack on the occurrence of injury and lameness in Standardbred horses. Aust Vet J 1997;75:751–752.
- 15.
McKee SL. An update on racing fatalities in the UK. Equine Vet Educ 1995;7:202–204.
- 16.
Oikawa M, Ueda Y, Inada S, et al. Effect of restructuring of a racetrack on the occurrence of racing injuries in Thoroughbred horses. J Equine Vet Sci 1994;14:262–268.
- 17.
Cohen ND, Mundy GD, Peloso JG, et al. Results of physical inspection before races and race-related characteristics and their association with musculoskeletal injuries in Thoroughbreds during races. J Am Vet Med Assoc 1999;215:654–661.
- 18.
Wilson JH, Robinson RA, Jensen RC, et al. Equine soft tissue injuries associated with racing: descriptive statistics from American racetracks, in Proceedings. Dubai Int Equine Symp 1996;1–21.
- 19.
Parkin TD, Clegg PD, French NP, et al. Risk of fatal distal limb fractures among Thoroughbreds involved in the five types of racing in the United Kingdom. Vet Rec 2004;154:493–497.
- 20.
Williams RB, Harkins LS, Hammond CJ, et al. Racehorse injuries, clinical problems and fatalities recorded on British racecourses from flat racing and National Hunt racing during 1996, 1997 and 1998. Equine Vet J 2001;33:478–486.
- 21.
Hill T, Carmichael D, Maylin G, et al. Track condition and racing injuries in Thoroughbred horses. Cornell Vet 1986;76:361–379.
- 22.
Ratzlaff MH, Wilson PD, Hutton DV, et al. Relationships between hoof-acceleration patterns of galloping horses and dynamic properties of the track. Am J Vet Res 2005;66:589–595.
- 23.
Barrey E, Landjerit B, Wolter R. Shock and vibration during the hoof impact on different track surfaces. In: Persson S, Jeffcott LB, Lindholm A, eds. Equine exercise physiology 3. Davis, Calif: ICEEP Publications, 1991;97–106.
- 24.
Gustås P, Johnston C, Drevemo S. Ground reaction force and hoof deceleration patterns on two different surfaces at the trot. Equine Comp Exerc Physiol 2007;3:209–216.
- 25.
Ratzlaff MH, Hyde ML, Hutton DV, et al. Interrelationships between moisture content of the track, dynamic properties of the track and the locomotor forces exerted by galloping horses. J Equine Vet Sci 1997;17:35–42.
- 26.
Kai M, Takahashi T, Aoki O, et al. Influence of rough track surfaces on components of vertical forces in cantering Thoroughbred horses. Equine Vet J Suppl 1999;(30):214–217.
- 27.
Rollot Y, Lecuyer E, Chateau H, et al. Development of a 3D model of the equine distal forelimb and of a GRF shoe for noninvasive determination of in vivo tendon and ligament loads and strains. Equine Vet J 2004;36:677–682.
- 28.↑
The Horse magazine Web site. Welfare and safety summit racehorse injury numbers revised. Available at: www.thehorse.com/ViewArticle.aspx?ID=11650. Accessed Apr 29, 2008.
- 29.↑
Roland ES, Hull ML, Stover SM. Design and demonstration of a dynamometric horseshoe for measuring ground reaction loads of horses during racing conditions. J Biomech 2005;38:2102–2112.
- 30.↑
Pratt GW. Model for injury to the foreleg of the Thoroughbred racehorse. Equine Vet J Suppl 1997;(23):30–32.
- 31.↑
Wilson AM, McGuigan MP, Su A, et al. Horses damp the spring in their step. Nature 2001;414:895–899.
- 32.↑
Witte TH, Hirst CV, Wilson AM. Effect of speed on stride parameters in racehorses at gallop in field conditions. J Exp Biol 2006;209:4389–4397.
- 33.↑
Merkens HW, Schamhardt HC, van Osch GJ, et al. Ground reaction force patterns of Dutch Warmbloods at the canter. Am J Vet Res 1993;54:670–674.
- 34.↑
Ngo L, Brand R. Model selection in linear mixed effects models using SAS® PROC MIXED, in Proceedings. SAS Tech Users Group Int 1997;1335–1340.
- 35.
Gustås P, Johnston C, Roepstorff L, et al. In vivo transmission of impact shock waves in the distal forelimb of the horse. Equine Vet J Suppl 2001;(33):11–15.
- 36.
Radin EL, Parker HG, Pugh JW, et al. Response of joints to impact loading. 3. Relationship between trabecular microfractures and cartilage degeneration. J Biomech 1973;6:51–57.
- 37.
Serink MT, Nachemson A, Hansson G. The effect of impact loading on rabbit knee joints. Acta Orthop Scand 1977;48:250–262.
- 38.
Hjertén G, Drevemo S. Semi-quantitative analysis of hoof-strike in the horse. J Biomech 1994;27:997–1004.
- 39.↑
Witte TH, Knill K, Wilson AM. Determination of peak vertical ground reaction force from duty factor in the horse (Equus caballus). J Exp Biol 2004;207:3639–3648.
- 40.↑
Merkens HW, Schamhardt HC, Van Osch GJ, et al. Ground reaction force patterns of Dutch Warmblood horses at normal trot. Equine Vet J 1993;25:134–137.
- 41.
Niki Y, Ueda Y, Masumitsu H. A force plate study in equine biomechanics. 2. The vertical and fore-aft components of floor reaction forces and motion of equine limbs at walk and trot. Bull Equine Res Inst 1982;19:1–17.
- 42.
Niki Y, Ueda Y, Masumitsu H. A force plate study in equine biomechanics. 3. The vertical and fore-aft components of floor reaction forces and motion of equine limbs at canter. Bull Equine Res Inst 1984;21:8–18.
- 43.↑
Whittle MW. Generation and attenuation of transient impulsive forces beneath the foot: a review. Gait Posture 1999;10:264–275.
- 44.
Kai M, Aoki O, Hiraga A, et al. Use of an instrument sandwiched between the hoof and shoe to measure vertical ground reaction forces and three-dimensional acceleration at the walk, trot, and canter in horses. Am J Vet Res 2000;61:979–985.
- 45.
Roepstorff L, Johnston C, Drevemo S. The effect of shoeing on kinetics and kinematics during the stance phase. Equine Vet J Suppl 1999;(30):279–285.
- 46.
Hjertén G, Drevemo S. A method to analyse forces and moments in the extremities of the horse during the stance phase at the trot. In: Gillespie R, Robinson NE, eds. Equine exercise physiology 2. Davis, Calif: ICEEP Publications, 1987;587–598.
- 47.
Back W, van Schie MHM, Pol JN. Synthetic shoes attenuate hoof impact in the trotting Warmblood horse. Equine Comp Exerc Physiol 2006;3:143–151.
- 48.
McLaughlin RM, Gaughan EM, Roush JK, et al. Effects of subject velocity on ground reaction force measurements and stance times in clinically normal horses at the walk and trot. Am J Vet Res 1996;57:7–11.
