Interest is increasing among equine researchers and practitioners in the use of objective methods for lameness detection and evaluation in horses. Objectively obtained measurements are resistant to the expectation bias inherent to subjective evaluations1 and are potentially more sensitive given that some types of objective data can be acquired at a rate faster than is possible with the unaided eye.2 Measurement of GRFs during weight bearing of the limbs by use of stationary force plates is a common method of objective lameness assessment used in clinical research.3
Most studies in which lameness is evaluated in horses by use of stationary force plates have involved measuring and reporting amplitude and impulse (force × time or area under the curve) of vertical GRF, with less interest or emphasis on horizontal GRF. The vertical GRF pattern of a hind limb striking the force plate while a horse is trotting is much like a bell-shaped curve, with Vmax occurring near the middle of stance when the limb is perpendicular to the ground. The Vauc1 is approximately equal to Vauc2. If by convention positive is to the right and the horse is moving right to left, the horizontal GRF force pattern begins with a deceleratory or negative force and impulse (Hauc−) as the limb strikes the ground, reaching a minimum value (Hmin) in the first half of the stance phase and returning to 0 near midstance. After midstance, horizontal GRF becomes positive (an acceleratory or propulsive force), peaks (Hmax) in the second half of stance, and returns to 0 at the end of stance.
Differences in Vmax and total vertical impulse (Vauc1 + Vauc2) between right and left limbs are known to be positively associated with lameness, with lower values in the lame or most lame limb.3 The association between lameness and horizontal force, however, is less clear. Nonetheless, assuming no important changes in stance duration, lame horses with decreased force of weight bearing within the first half of stance, compared with nonlame horses, should also have decreased vertical impulse in early stance (decreased Vauc1) as well as decreased negative horizontal amplitude and impulse (more positive Hmin and decreased Hauc−). Similarly, horses with decreased force of weight bearing within the second half of stance, compared with nonlame horses, should also have decreased vertical impulse in late stance (decreased Vauc2) as well as decreased positive horizontal amplitude and impulse (decreased Hmax and decreased Hauc+). The small number of studies performed to examine relationships between the aforementioned force variables and clinical lameness can be attributed to the fact that such studies are difficult to accomplish. They are time-consuming, expensive, and limited to collections of data from a few noncontiguous strides.
Most limitations and difficulties of force plate measurement in equine lameness evaluations can be overcome or circumvented by use of body-mounted inertial sensors. Downward torso movement influences GRF patterns in the first half of stance. Ground reaction forces in the second half of stance influence upward torso movement. With body-mounted inertial sensors, vertical torso movement (deceleration and acceleration) can be measured over multiple contiguous strides, even in the natural, overground environment.
One commercially available system,a designed specifically to evaluate lameness in horses, measures vertical head and pelvic acceleration and converts those data to vertical position in the local reference frame of the horse. For hind limb lameness, differences in minimum (during hind limb stance; PDmin) and maximum (before and after hind limb stance; PDmax) pelvic height between right and left halves of a stride are determined for each trotting stride. Decreased downward force of the rear torso onto a lame or more lame hind limb in the first half of stance will result in abbreviated fall of the pelvis relative to the ground, compared with results during the first half of stance in the nonlame or less lame limb, and is measured as PDmin, which can be used to identify impact-type lameness. Likewise, a decrease in upward force from a lame or more lame hind limb in the second half of stance will result in abbreviated rise of the pelvis, so that the pelvis reaches a lower position relative to the ground than is achieved after stance for the nonlame or less lame limb and is measured as PDmax, which can be used to identify push off–type lameness.4 We propose that the nature of hind limb lameness (ie, differentiation of impact-type from push off–type lameness) can be determined from the relative amplitudes of PDmin and PDmax.
The purpose of the study reported here was to investigate relationships between PDmin and PDmax and stationary force plate measurements (vertical and horizontal) obtained from horses during the first and second halves of hind limb stance. Our hypotheses were that PDmin would be more strongly associated than PDmax with force plate variables measured in the first half of stance (Vmax, Vauc1, Hmin, and Hauc−) and that PDmax would be more strongly associated with force plate variables measured in the second half of stance (Vmax, Vauc2, Hmax, and Hauc+).
Materials and Methods
Horses
Twenty-one adult horses (mean age, 11.8 years; range, 3 to 22 years) were selected for the study, including 11 geldings and 10 mares. Breeds included Quarter Horse (n = 15), Thoroughbred (5), and Paint (1). Twelve horses had been brought to the Oklahoma State Veterinary Teaching Hospital for lameness evaluation, and 9 were owned by Oklahoma State University. Horses were included in the study if they were deemed not lame or had mild primary hind limb lameness, which was determined for all but 2 horses by simultaneous subjective lameness evaluation by 2 experienced veterinarians. For the remaining 2 horses, determination of hind limb lameness was made by 1 experienced clinician but confirmed by results of local anesthetic testing to isolate lameness within the limb.
Horses were included in the study if they were deemed nonlame by both clinicians or mildly lame in a hind limb by at least 1 clinician. No horse with primary forelimb lameness or with severe hind limb lameness (lameness visible at a walk) was used, providing a restricted set of horses ranging from nonlame to mildly lame in the right and left hind limb. Client-owned horses were evaluated without additional trimming and shoeing, but university-owned horses were trimmed and shod in accordance with their usual program 2 weeks before evaluation. Causes of hind limb lameness may have been known from previous histories and examinations or determined from clinical workup, but specific diagnosis was not of interest and did not factor into subject selection. Use of all horses was performed in accordance with approved institutional animal care and use protocols. Permissions for instrumentation of horses and collection and use of data were obtained from owners of all privately owned horses.
Simultaneous force plate and inertial sensor evaluation
University-owned horses had experience trotting over a force plate. Client-owned horses were given a brief introduction and practice session that involved walking and trotting over the force plate.b Data collection began once the horse freely trotted over the force plate while lightly restrained on a lead shank by the handler. Before data collection began, each horse was weighed on a digital scale that was accurate within 2.27 kg.
Horses were instrumented with a commercially available, body-mounted inertial sensor systema consisting of 3 devices. An accelerometer device (uniaxial [vertical] accelerometer sensor with a range of 6 g) was placed on the midline of the head and on the pelvic midline between the tubera sacralia with plastic hook and loop tape. A third device (uniaxial [sagittal plane] gyroscope sensor with a range of 300°/s) was placed on the dorsal surface of the pastern region of the right forelimb in a specially designed pastern wrap pouch. Sensors collected data at 200 Hz, and data were wirelessly (bluetooth class 1) transmitted after 8-bit analog-to-digital conversion. Vertical acceleration was converted to position by a moving-window, error-correcting, double-integration procedure as described elsewhere.5 The angular rate signal from the right forelimb was smoothed, and peak detection was used to identify right forelimb stance. Data from the head sensor were not used in the study. Detection of right hind limb stance was inferred from knowledge that the horse was trotting. Each sensor weighed 28 g and measured 2.4 × 2.5 × 1.5 cm. Physical design of these sensors, method of wireless transmission, and stride selection method have been described and validated previously.6,7
After instrumentation, each horse was led at a trot over the force plate surfaceb by a handler. Two observers confirmed a successful strike of an entire hind foot on the force plate surface. Acceptable speed of movement for each horse, as measured with an infrared beam interruption timer (with accuracy to within 1 millisecond),c was restricted to a 15% increase or decrease from the mean of all successful force plate strikes for that horse after a visually acceptable first force plate strike. Data collected on force plate strikes outside of this window of acceptable speed were discarded. Each horse was trotted back and forth over the force plate until at least 6 successful strikes for each hind limb were recorded. Inertial sensor data collection began before and continued a few seconds after the hind foot struck the force plate. However, only the stride with the successful force plate strike and 1 or 2 strides that immediately preceded or followed that strike were selected for analysis. Exact matching and detection of the single stride that successfully struck the force plate were not possible; however, a small number of strides (2 to 4), including the successful force plate strike, could be isolated for each trial.
Vertical pelvic position trajectory was then split into single strides. The PDmax was calculated for each stride as the maximum pelvic height preceding right hind limb stance minus the maximum pelvic height preceding left hind limb stance. The PDmin was calculated for each stride as the minimum pelvic height during right hind limb stance minus the minimum pelvic height during left hind limb stance (Figure 1). The PDmin measures the amplitude and side of hind limb lameness attributable to asymmetric downward movement during the first half of stance (hereafter referred to as impact-type lameness). The PDmax measures the amplitude and side of hind limb lameness attributable to asymmetric upward movement during the second half of stance (hereafter referred to as push off–type lameness). In accordance with equipment convention, positive values of PDmin and PDmax indicate right hind limb impact-type and push off–type lameness, respectively. Conversely, negative values of PDmin and PDmax indicate left hind limb impact-type and push off–type lameness, respectively.
Diagram illustrating the use of data from an inertial sensor system to estimate PDmin and PDmax when evaluating a horse for hind limb lameness. In this example, pelvic height trajectory of a trotting horse with lameness of the right hind limb (RH) has decreased downward movement of the pelvis in the first half of RH stance, compared with that in the first half of left hind limb (LH) stance (PDmin), and decreased upward movement of the pelvis in the second half of RH stance, compared with that in the second half of LH stance (PDmax) over approximately 2 strides. The PDmax is determined by subtracting maximum height of pelvis after RH stance (max2) from maximum height of pelvis after LH stance (max1). The PDmin is determined by subtracting minimum height of pelvis during LH stance (min2) from minimum height of pelvis during RH stance (min1).
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Diagram illustrating the use of data from an inertial sensor system to estimate PDmin and PDmax when evaluating a horse for hind limb lameness. In this example, pelvic height trajectory of a trotting horse with lameness of the right hind limb (RH) has decreased downward movement of the pelvis in the first half of RH stance, compared with that in the first half of left hind limb (LH) stance (PDmin), and decreased upward movement of the pelvis in the second half of RH stance, compared with that in the second half of LH stance (PDmax) over approximately 2 strides. The PDmax is determined by subtracting maximum height of pelvis after RH stance (max2) from maximum height of pelvis after LH stance (max1). The PDmin is determined by subtracting minimum height of pelvis during LH stance (min2) from minimum height of pelvis during RH stance (min1).
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Diagram illustrating the use of data from an inertial sensor system to estimate PDmin and PDmax when evaluating a horse for hind limb lameness. In this example, pelvic height trajectory of a trotting horse with lameness of the right hind limb (RH) has decreased downward movement of the pelvis in the first half of RH stance, compared with that in the first half of left hind limb (LH) stance (PDmin), and decreased upward movement of the pelvis in the second half of RH stance, compared with that in the second half of LH stance (PDmax) over approximately 2 strides. The PDmax is determined by subtracting maximum height of pelvis after RH stance (max2) from maximum height of pelvis after LH stance (max1). The PDmin is determined by subtracting minimum height of pelvis during LH stance (min2) from minimum height of pelvis during RH stance (min1).
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Horses were classified as lame by use of the force plate when a difference in Vmax > 4% of the horse's body weight was identified between hind limbs. Horses were classified as lame by use of the inertial sensor system when the absolute value of PDmin or PDmax was > 3 mm.
Data regarding vertical and horizontal GRFs from successful hind limb strikes were simultaneously collected and saved for analysis. Each successful hind limb strike on the force plate was used to extract Vmax, Vauc1, Vauc2, Hmin, Hzc, Hauc−, Hmax, and Hauc+ as well as time indices for Vmax, Hmin, Hmax, and HzC (Figure 2). Means were calculated for each value over the 6 successful strikes from each hind limb. Mean right hind limb values were subtracted from mean left hind limb values so that positive differences indicated right hind limb lameness and, similarly, negative differences indicated left hind limb lameness. This approach allowed polarity to concur with that of the inertial sensor results (ie, positive for right-sided lameness and negative for left-sided lameness). The Vmax, Hmin, Hmax, and all impulse values (Vauc1, Vauc2, Hauc−, and Hauc+) were converted from Newtons to kilograms by dividing by 9.8 m/s2 and standardized by dividing by the horse's weight, yielding units that reflected percentage of body weight. All time index measurements were generated after standardization of trial data to 100 equaltime increments from beginning (first positive vertical GRF defined as baseline) to end (return to baseline) of stance, yielding decimal values that represented percentage of stance. Total stance duration prior to data reacquisition was also measured and recorded.
Diagram illustrating vertical (dashed line) and horizontal (black line) GRF patterns as measured via force plate for a typical hind limb stance of a horse. Vertical GRF starts at approximately 0% of body weight, first increases to a maximum value (Vmax), and then decreases through stance. Horizontal GRF begins as a negative (or deceleratory) force that peaks in the middle of the first half of the stance (Hmin), then becomes positive (or acceleratory) at approximately midstance, peaking in the middle of the second half of the stance (Hmax). HmaxI = Time index of Hmax. HminI = Time index of Hmin. HzcI = Time index of Hzc. VmaxI = Time index of Vmax.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Diagram illustrating vertical (dashed line) and horizontal (black line) GRF patterns as measured via force plate for a typical hind limb stance of a horse. Vertical GRF starts at approximately 0% of body weight, first increases to a maximum value (Vmax), and then decreases through stance. Horizontal GRF begins as a negative (or deceleratory) force that peaks in the middle of the first half of the stance (Hmin), then becomes positive (or acceleratory) at approximately midstance, peaking in the middle of the second half of the stance (Hmax). HmaxI = Time index of Hmax. HminI = Time index of Hmin. HzcI = Time index of Hzc. VmaxI = Time index of Vmax.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Diagram illustrating vertical (dashed line) and horizontal (black line) GRF patterns as measured via force plate for a typical hind limb stance of a horse. Vertical GRF starts at approximately 0% of body weight, first increases to a maximum value (Vmax), and then decreases through stance. Horizontal GRF begins as a negative (or deceleratory) force that peaks in the middle of the first half of the stance (Hmin), then becomes positive (or acceleratory) at approximately midstance, peaking in the middle of the second half of the stance (Hmax). HmaxI = Time index of Hmax. HminI = Time index of Hmin. HzcI = Time index of Hzc. VmaxI = Time index of Vmax.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Statistical analysis
Stepwise linear regression was performed to investigate strengths of association between inertial sensor measurements of hind limb lameness (PDmax and PDmin) and sets of force plate variables (Vmax, Hmin, and Hmax), those related to vertical and horizontal impulse (Vauc1, Vauc2, Hauc−, and Hauc+), and those related to time indices of important events in the vertical and horizontal force curves (ie, indices for Vmax, Hmin, Hmax, and Hzc). Zero-order Pearson product moment correlation coefficients were first calculated between each force plate variable and PDmin and PDmax. Variables were initially added to the model if their F values were greater than an F-to-enter threshold at P < 0.05. Because there were few variables with F values above this threshold, all possible combinations of variables with F values above the F-to-enter threshold were tested as possible models and an adjusted multiple correlation coefficient (R2a) was calculated for each. Models with highest R2a and each variable contributing independently (accounting for collinearity, with a partial F value above the F-to-remove threshold at P < 0.1) were interpreted. Departures of variables and combinations of variables from normal distribution were investigated with the Shapiro-Wilk test and determined to be significant at P < 0.05. For the restricted purpose of summarizing results, variable or model correlations with Ra2 < 0.4 were considered weak, 0.4 < Ra2 ≤ 0.7 were considered moderate, and Ra2 < 0.7 were considered strong.
The Fisher exact test was used to compare proportions of horses identified through force plate and initial sensor evaluations as nonlame, lame in right hind limb, and lame in left hind limb. As additional confirmation for consistency of speed of movement between lame and nonlame limbs, mean speed of movement, stride rate, and stance duration were compared between lame and nonlame limb trials with the paired Student t test. Lame limb trials were defined as those with a lower mean Vmax, and nonlame limb trials were defined as the side with higher mean Vmax. Statistical softwared was used for all analyses. Values of P < 0.05 were considered significant for all comparisons.
Results
Of the 21 horses used in the study, 19 underwent subjective lameness evaluation by 2 clinicians. These evaluators agreed on the existence and side of hind limb lameness for 14 horses, 3 of which were believed to be nonlame, 6 were believed to be lame in the right hind limb, and 5 were believed to be lame in the left hind limb. There was disagreement for 5 horses, with one clinician indicating the horse was mildly lame in a hind limb and the other indicating that the horse was not lame. In no circumstances did one evaluator indicate that the horse was lame in one hind limb but the other indicate that the horse was lame in the other hind limb. Force place analysis revealed that 11 of the 21 adult horses included in the study were not lame in the hind limbs, 6 were lame in the right hind limb, and 4 were lame in the left hind limb. The mean difference in Vmax between hind limbs was 4.1% (SD, 2.8%; range, 0.5% to 9.9%). When data from inertial sensors were used to identify lameness, only 2 horses were identified with both a PDmin and PDmax, that suggested they were not lame in the hind limbs, with the other 19 horses identified as lame in at least 1 hind limb.
By PDmin alone, 6 horses were identified as not lame, 9 were identified as lame in the right hind limb, and 5 were identified as lame in the left hind limb. By PDmax alone, 6 horses were identified as not lame, 5 were identified as lame in the right hind limb, and 10 were identified as lame in the left hind limb. No significant differences were identified between lame and nonlame hind limb trials for speed of movement (mean, 2.9 m/s; P = 0.55), stride rate (mean, 1.7 strides/s; P = 0.55), or stance duration (0.37 seconds; P = 0.79).
For PDmin, associations with force plate variables were strongest for single-component models only (Table 1). The only association that was moderately strong was a positive association with Vmax (Figure 3; Ra2 = 60; P < 0.001). The PDmin increased approximately 1 mm for every 1% increase in difference in Vmax between the hind limbs, with Vmax smaller in the lame versus nonlame hind limb. All other significant correlations (with Hmin, Vauc1, Vauc2, and time index of Hmin) were positive but weak. Curves were plotted and represented mean vertical and horizontal GRFs for all horses with inertial sensor asymmetry indicative of impact-type lameness only (ie, PDmin greater than threshold and PDmax lower than threshold; Figure 4).
Fitted linear regression lines (solid line) and 95% confidence bounds (dotted lines) for the best associations achieved between stationary force plate measurements and inertial sensor measurements of hind limb lameness in 21 horses. A—Predicted versus actual PDmin as estimated from Vmax (PDmin = 0.7 + 0.9 Vmax; Ra2 = 0.60; P < 0.001). B—Predicted versus actual PDmax as estimated from Vauc2 and Hauc+ (PDmax = 0.2 + 7.0 VauC2 – 42.5 HauC+; Ra2. = 0.77; P < 0.001).
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Fitted linear regression lines (solid line) and 95% confidence bounds (dotted lines) for the best associations achieved between stationary force plate measurements and inertial sensor measurements of hind limb lameness in 21 horses. A—Predicted versus actual PDmin as estimated from Vmax (PDmin = 0.7 + 0.9 Vmax; Ra2 = 0.60; P < 0.001). B—Predicted versus actual PDmax as estimated from Vauc2 and Hauc+ (PDmax = 0.2 + 7.0 VauC2 – 42.5 HauC+; Ra2. = 0.77; P < 0.001).
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Fitted linear regression lines (solid line) and 95% confidence bounds (dotted lines) for the best associations achieved between stationary force plate measurements and inertial sensor measurements of hind limb lameness in 21 horses. A—Predicted versus actual PDmin as estimated from Vmax (PDmin = 0.7 + 0.9 Vmax; Ra2 = 0.60; P < 0.001). B—Predicted versus actual PDmax as estimated from Vauc2 and Hauc+ (PDmax = 0.2 + 7.0 VauC2 – 42.5 HauC+; Ra2. = 0.77; P < 0.001).
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical (black lines) and horizontal (gray lines) GRFs for nonlame (solid lines) and lame (dashed lines) hind limbs of 4 adult horses with a PDmin greater than and PDmax less than the threshold used to indicate lameness (ie, difference with an absolute value of at least 3 mm). Notice the large decrease in Vmax (thick arrow) in the lame hind limbs, compared with in the nonlame hind limbs and smaller decreases in Hmin and Hmax (thin arrows) in the lame hind limbs than in the nonlame hind limbs.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical (black lines) and horizontal (gray lines) GRFs for nonlame (solid lines) and lame (dashed lines) hind limbs of 4 adult horses with a PDmin greater than and PDmax less than the threshold used to indicate lameness (ie, difference with an absolute value of at least 3 mm). Notice the large decrease in Vmax (thick arrow) in the lame hind limbs, compared with in the nonlame hind limbs and smaller decreases in Hmin and Hmax (thin arrows) in the lame hind limbs than in the nonlame hind limbs.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical (black lines) and horizontal (gray lines) GRFs for nonlame (solid lines) and lame (dashed lines) hind limbs of 4 adult horses with a PDmin greater than and PDmax less than the threshold used to indicate lameness (ie, difference with an absolute value of at least 3 mm). Notice the large decrease in Vmax (thick arrow) in the lame hind limbs, compared with in the nonlame hind limbs and smaller decreases in Hmin and Hmax (thin arrows) in the lame hind limbs than in the nonlame hind limbs.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Zero-order Pearson product correlations between stationary force plate and inertial sensor-derived measurements of hind limb lameness in 21 orthopedically normal or mildly lame horses.
| PDmin | PDmax | |||
|---|---|---|---|---|
| Variable | Correlation | P value | Correlation | P value |
| Amplitude | ||||
| Vmax | 0.600 | < 0.001 | 0.064 | 0.147 |
| Hmin | 0.153 | 0.045 | −0.570 | < 0.001 |
| Hmax | −0.035 | 0.572 | −0.394 | 0.001 |
| Impulse | ||||
| Vauc1 | 0.365 | 0.002 | 0.196 | 0.026 |
| Vauc2 | 0.323 | 0.004 | 0.440 | 0.001 |
| Hauc− | 0.086 | 0.105 | −0.575 | < 0.001 |
| Hauc+ | −0.052 | 0.935 | −0.423 | 0.001 |
| Time index | ||||
| Of Vmax | −0.046 | 0.732 | 0.187 | 0.029 |
| Of Hmin | 0.180 | 0.031 | 0.191 | 0.027 |
| Of Hzc | 0.025 | 0.235 | 0.679 | < 0.001 |
| Of Hmax | −0.004 | 0.351 | 0.181 | 0.031 |
Negative values indicate lameness of the left hind limbs, and positive values indicate lameness of the right hind limbs. Values of P < 0.05 were considered significant.
The PDmax was not significantly associated with Vmax. Moderate, single-component associations were identified between PDmax and Vauc2 and between PDmax and some horizontal GRF measurements of amplitude (Hmin, Hmax), impulse (Huc−, Hauc+), and timing (time index of Hzc). The association between PDmax and Vauc2 was positive, indicating that an increase in push off-type lameness as defined by inertial sensor data was associated with a decrease in late vertical GRF impulse. Associations between PDmax and Hmin, Hmax, Hauc−, and Hauc+ were negative, indicating that an increase in push off–type lameness as defined by inertial sensor data was associated not only with decreases in Hmin and Hauc– but also with increases in Hmax and Hauc+ in lame limbs. The association between PDmax and time index of Hzc was positive, indicating that an increase in push off–type lameness identified with the inertial sensors was associated with an earlier transition from negative horizontal GRF to positive horizontal GRF in lame hind limbs.
Strong associations were identified with PDmax in the 2-component force plate models. The PDmax was strongly associated with a decrease in Vauc2 and a simultaneous increase in Hauc+ (Figure 3; Ra2 = 0.77; P < 0.001. Moreover, the decrease in Vauc2 was small, and the increase in Hauc+ was large. The PDmax was also strongly associated with a decrease in Hmin and an increase in Hmax (Ra2 = 0.61; P = 0.03 for Hmin; P = 0.01 for Hmax) and strongly associated with earlier occurrence of Hzc and Hmax (Ra2 = 0.74; P < 0.001 for time index of Hzc; P = 0.03 for time index for Hmax). Transition of horizontal GRF from negative to positive occurred 1% to 2% earlier in the stride cycle for each 1-mm increase in PDmax. Curves were plotted, representing vertical and horizontal GRFs for horses with inertial sensor asymmetry indicative of push off–type lameness only (ie, with PDmax greater than threshold and PDmin lower than threshold; Figure 5), horses with inertial sensor asymmetry indicative of both impact-type and push off–type lameness (ie, both PDmin and PDmax were greater than threshold and of the same sign; Figure 6), and for horses with no hind limb lameness (ie, both PDmin and PDmax were less than threshold; Figure 7).
Mean vertical and horizontal GRFs for nonlame and lame hind limbs of 4 adult horses with a PDmax greater than and PDmin less than the threshold used to indicate lameness. Notice the small decrease in Vmax in the lame hind limbs compared with the Vmax in the nonlame hind limbs (downward arrow) and large shift of horizontal GRF to the left (large upward arrow) with a decrease in Hmin and increase in Hmax (small upward arrows). See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for nonlame and lame hind limbs of 4 adult horses with a PDmax greater than and PDmin less than the threshold used to indicate lameness. Notice the small decrease in Vmax in the lame hind limbs compared with the Vmax in the nonlame hind limbs (downward arrow) and large shift of horizontal GRF to the left (large upward arrow) with a decrease in Hmin and increase in Hmax (small upward arrows). See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for nonlame and lame hind limbs of 4 adult horses with a PDmax greater than and PDmin less than the threshold used to indicate lameness. Notice the small decrease in Vmax in the lame hind limbs compared with the Vmax in the nonlame hind limbs (downward arrow) and large shift of horizontal GRF to the left (large upward arrow) with a decrease in Hmin and increase in Hmax (small upward arrows). See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for nonlame and lame hind limbs of 6 adult horses with both PDmax and PDmin greater than the threshold used to indicate lameness. Notice the decrease in Vmax in the lame hind limbs, compared with the Vmax in the nonlame limbs (downward arrow) and shift of horizontal GRF to the left (large upward arrow) with a decrease in Hmin and increase in Hmax (small upward arrows). See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for nonlame and lame hind limbs of 6 adult horses with both PDmax and PDmin greater than the threshold used to indicate lameness. Notice the decrease in Vmax in the lame hind limbs, compared with the Vmax in the nonlame limbs (downward arrow) and shift of horizontal GRF to the left (large upward arrow) with a decrease in Hmin and increase in Hmax (small upward arrows). See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for nonlame and lame hind limbs of 6 adult horses with both PDmax and PDmin greater than the threshold used to indicate lameness. Notice the decrease in Vmax in the lame hind limbs, compared with the Vmax in the nonlame limbs (downward arrow) and shift of horizontal GRF to the left (large upward arrow) with a decrease in Hmin and increase in Hmax (small upward arrows). See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for the right and left hind limbs of 2 adult horses with both PDmax and PDmin lower than the threshold used to indicate lameness. Notice that the curves for vertical or horizontal GRFs are equivalent for the right and left hind limbs. See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for the right and left hind limbs of 2 adult horses with both PDmax and PDmin lower than the threshold used to indicate lameness. Notice that the curves for vertical or horizontal GRFs are equivalent for the right and left hind limbs. See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Mean vertical and horizontal GRFs for the right and left hind limbs of 2 adult horses with both PDmax and PDmin lower than the threshold used to indicate lameness. Notice that the curves for vertical or horizontal GRFs are equivalent for the right and left hind limbs. See Figure 4 for remainder of key.
Citation: American Journal of Veterinary Research 77, 4; 10.2460/ajvr.77.4.337
Discussion
Results of the present study involving force plate and inertial sensor evaluations partially supported the hypothesis that PDmin and PDmax reflect different attributes of hind limb lameness in horses. However, it cannot be presumed that PDmin and PDmax can be substituted for measurements of vertical and horizontal GRFs made in the first and second half of stance. The PDmin was the better estimate of the difference in Vmax between hind limbs and was more strongly associated with vertical impulse at the beginning (Vauc1) rather than at the end (Vauc2) of stance. However, PDmin was not associated with differences in any horizontal GRF variables between hind limbs. These findings agreed with those of a previous study8 involving horses with experimentally induced lameness that were trotted on a force-measuring treadmill, in which vertical deceleration of the sacrum was significantly correlated with Vmax. However, horizontal forces in that study could not be measured. An increase in PDmin, which reflects reduced fall of the pelvis during the stance phase of the lame limb, so that the local minimum becomes higher than that during the stance phase of the nonlame limb, simply represents decreased weight bearing on a hind limb.
The PDmax in the present study was more strongly associated with vertical ground reaction impulse at the end of (Vauc2) versus beginning of (Vauc1) stance but was not associated with Vmax. In situations in which the absolute value of PDmax was greater than the threshold of 3 mm, there was a transfer of impulse in the second half of stance from vertical to horizontal. An increase in PDmax reflected a small decrease in vertical ground reaction impulse but a large increase in positive horizontal impulse in the second half of stance. This can be explained in that a horse with hind limb lameness greatest in the second half of stance, when constrained to move forward at a certain speed, could accomplish torso advancement with more forward but less vertical thrust. This movement could be executed by either or a combination of increased inertia of the opposite hind limb actively swinging forward or hip joint extension without proportional limb lengthening (ie, stifle and hock extension). The torso would roll over the stationary hoof, but with reduced vertical thrust.
Horizontal GRFs of trotting horses and their alterations in horses with clinical and experimentally induced lameness have been evaluated in few studies.9–14 Although a decrease in deceleratory horizontal GRF and an increase in acceleratory horizontal GRF were identified in the present study when PDmax was greater than the threshold used to indicate lameness, decreases in both the deceleratory and acceleratory components of horizontal GRF were identified in horses with experimentally induced metacarpophalangeal lameness in a previous study.11 Also, in another study9 involving horses with experimentally induced superficial flexor tendinitis of the forelimb, a decrease was identified in Hauc− but no significant changes were detected in acceleratory forces. Both of those studies9,11 involved measurement of forelimb forces, and the mechanics, energetics (work and power), and function of the fore- and hind limbs are not equivalent. In a third study13 involving horses with hind limb lameness caused by intratendon and intraligament injections of collagenase, decreases were identified in both the acceleratory and deceleratory components of horizontal GRFs for lesions of the flexor tendon but only in the deceleratory component for lesions of the suspensory ligament. However, only 4 horses were evaluated in that study,13 suspensory ligament lameness was induced following induction of flexor tendon lameness in the same horse, and horses became more than mildly lame, with differences in Vmax between limbs of > 30%. In the present study, horses were considerably less lame, with mean differences in Vmax between hind limbs of only 4%. More severe lameness would likely cause pain in affected horses throughout the stance phase, and adaptations that are possible with mild lameness limited to the second half of stance may not be possible with severe lameness in both halves of the stride.
One method that veterinarians use to detect hind limb lameness is evaluation of the vertical movement of the entire pelvis. Differences in pelvic fall (PDmin) had a larger increase in movement asymmetry per degree of visual lameness score than did differences in pelvic rise (PDmax) in a previous study.15 However, differences in pelvic rise and a transfer of vertical horizontal GRFs may also be important in at least some lameness conditions. Exacerbation of hind limb lameness with hind limb flexion tests is more accurately reflected in an increase in PDmax than in PDmin.16,17 In a small study18 involving horses with proximal suspensory desmitis in the hind limbs, lameness was more strongly associated with an increase in PDmax than in PDmin. In a different study19 involving horses with hind limb lameness caused by pressure applied to the frog, kinematic tracking of vertical pelvic movement with horses trotting on a treadmill revealed that PDmax was better than PDmin for detection of mild hind limb lameness. These findings highlight the importance of measuring both pelvic fall and pelvic rise when evaluating horses for hind limb lameness.
Use of a stationary force plate and low threshold for differentiation of nonlame from lame horses resulted in horses being more likely to be classified as nonlame than as lame in the present study. This contrasted with the situation in another study11 in which, even after improvement in the degree of forelimb lameness following injection of endotoxin into a joint, results of subjective lameness evaluation indicate that horses were nonlame, even though use of the stationary force plate identified significant differences in vertical GRF between control and lame limbs. Hind limb lameness is generally more difficult to detect than forelimb lameness, primarily because the degree of asymmetry in head movement is greater and easier to detect than that of pelvic movement. Observer agreement for detection of lameness and determination of affected side is lower for hind limb lameness than for forelimb lameness.20–22 These findings suggest that detection of lameness would be detected earlier if a more objective method was used instead of subjective evaluation. However, comparison of the intensity and depth of lameness evaluations in studies in which the purpose was different (eg, use of horses with possible hind limb lameness vs horses with known, experimentally induced lameness) highlights the difficulty inherent when subjective lameness evaluation is used as a gold standard. Estimation of the degree of agreement between subjective evaluation, force plate, and inertial sensor results was not the primary intent of the study reported here.
In the present study, the association of PDmin with Vmax was positive and significant (Ra2 = 0.60; P < 0.001). This association was lower than that identified in a previous study21 in which comparisons were made between lame and nonlame forelimbs in vertical movement of the head as measured by inertial sensors and mean Vmax difference. However, strength of association is strongly influenced by range of measurements used in the comparison of 2 methods. The horses used in the present study had hind limb lameness that was fairly mild, with mean Vmax between hind limbs equal to the threshold that was estimated between nonlame and lame limbs. Indeed, the threshold we used for identification of hind limb lameness by use of vertical GRF was probably too high, given that the force plate method was more likely than the inertial sensor method to indicate that a horse was not lame. Also, the threshold used for the inertial sensors to indicate lameness (a difference with an absolute value of at least 3 mm for PDmin and PDmax) was probably too low, given that the inertial sensor method was more likely than the force plate method to indicate that a horse was lame.
Speed of movement is significantly correlated with hind limb vertical and braking forces in trotting horses.23 In the present study, speed of movement, stride rate, and stance duration did not differ between trials when lame or nonlame limbs of trotting horses struck the force plate.
The study reported here yielded preliminary evidence that inertial sensor measurements of asymmetric vertical pelvic movement may provide information relevant to the nature of hind limb lameness (ie, push off-type vs impact-type). The PDmin reflected changes in vertical GRF, primarily in the first half of stance, and PDmax reflected changes in both vertical and horizontal GRFs in the second half of stance. However, the clinical applicability of these findings can only be determined by comparison of vertical pelvic movement patterns in horses in which hind limb lameness has been definitively diagnosed, as supported by successful reduction of lameness with regional, nerve, or joint anesthesia and definitive imaging findings.
Acknowledgments
This manuscript represents a portion of a thesis submitted by Dr. Bell to the University of Missouri Department of Biomedical Sciences as partial fulfillment of the requirements for a Master of Science in Biomedical Sciences, with an emphasis on Veterinary Medicine and Surgery.
Supported by the E. Paige Laurie Endowed Program in Equine Lameness. Drs. Keegan, Yonezawa, and Pai developed the inertial sensor system used in this study, which is owned by the University of Missouri and licensed to Equinosis LLC for manufacturing and marketing. Drs. Keegan, Yonezawa, and Pai are unpaid minority shareholders in Equinosis LLC. No financial support for this study was obtained from Equinosis LLC.
ABBREVIATIONS
| GRF | Ground reaction force |
| Hauc− | Horizontal deceleratory or negative impulse |
| Hauc+ | Horizontal acceleratory or positive impulse |
| Hmax | Maximum horizontal acceleratory ground reaction force |
| Hmin | Minimum horizontal deceleratory ground reaction force |
| Hzc | Horizontal ground reaction force zero crossing when negative (decelerating) switched to positive (accelerating) horizontal ground reaction force |
| PDmax | Difference in maximum pelvic height before and after hind limb stance between right and left halves of a stride |
| PDmin | Difference in minimum pelvic height during hind limb stance between right and left halves of a stride |
| Vaucl | Vertical ground reaction force impulse before maximum vertical ground reaction force |
| Vauc2 | Vertical ground reaction force impulse after maximum vertical ground reaction force |
| Vmax | Maximum vertical ground reaction force |
Footnotes
Lameness Locator, Equinosis LLC, Columbia, Mo.
Multicomponent force plate (model 9287BA), Kistler Instrument Corp, Amherst, NY.
Phototiming switch system (model 49-551A), RadioShack, Fort Worth, Tex.
Stats Direct Medical Statistics Software, StatsDirect Ltd, Altrincham, Cheshire, England.
References
1. Arkell M, Archer RM, Guitian FJ, et al. Evidence of bias affecting the interpretation of the results of local anaesthetic nerve blocks when assessing lameness in horses. Vet Rec 2006; 159: 346–349.
2. Sweet AL. Temporal discrimination by the human eye. Am J Psychol 1953; 66: 185–198.
3. Clayton HM. The force plate: established technology, new applications. Vet J 2005; 169: 15–16.
4. Kramer J, Keegan KG. Kinematics of lameness In: Hinchcliff KW, Kaneps AJ, Geor RJ, eds. Equine sports medicine and surgery. 2nd ed. Philadelphia: WB Saunders Co, 2014;223–238.
5. Keegan KG, Yonezawa Y, Pai PF, et al. Accelerometer-based system for the detection of lameness in horses. Biomed Sci Instrum 2002; 38: 107–112.
6. Keegan KG, Kramer J, Yonezawa Y, et al. Assessment of repeatability of a wireless, inertial sensor-based lameness evaluation system for horses. Am J Vet Res 2011; 72: 1156–1163.
7. Keegan KG, Yonezawa Y, Pai PF, et al. Evaluation of a sensor-based system of motion analysis for detection and quantification of forelimb and hind limb lameness in horses. Am J Vet Res 2004; 65: 665–670.
8. Weishaupt MA, Wiestner T, Hogg HP, et al. Compensatory load redistribution of horses with induced weightbearing hindlimb lameness trotting on a treadmill. Equine Vet J 2004; 36: 727–733.
9. Clayton HM, Schamhardt HC, Willemen MA, et al. Kinematics and ground reaction forces in horses with superficial digital flexor tendinitis. Am J Vet Res 2000; 61: 191–196.
10. Williams GE, Silverman BW, Wilson AM, et al. Disease-specific changes in equine ground reaction force data documented by use of principal component analysis. Am J Vet Res 1999; 60: 549–555.
11. Ishihara A, Bertone AL, Rajala-Schultz PJ. Association between subjective lameness grade and kinetic gait parameters in horses with experimentally induced forelimb lameness. Am J Vet Res 2005; 66: 1805–1815.
12. Ishihara A, Reed SM, Rajala-Schultz PJ, et al. Use of kinetic gait analysis for detection, quantification, and differentiation of hind limb lameness and spinal ataxia in horses. J Am Vet Med Assoc 2009; 234: 644–651.
13. Keg PR, Barneveld A, Schamhardt HC, et al. Clinical and force plate evaluation of the effect of a high plantar nerve block in lameness caused by induced mid-metatarsal tendinitis. Vet Q 1994; 16: 70–75.
14. Khumsap S, Lanovaz JL, Rosenstein DS, et al. Effect of induced unilateral synovitis of distal intertarsal and tarsometatarsal joints on sagittal plane kinematics and kinetics of trotting horses. Am J Vet Res 2003; 64: 1491–1495.
15. Pfau T, Spicer-Jenkins C, Smith RK, et al. Identifying optimal parameters for quantification of changes in pelvic movement symmetry as a response to diagnostic analgesia in the hindlimbs of horses. Equine Vet J 2014; 46: 759–763.
16. Starke SD, Willems E, Head M, et al. Proximal hindlimb flexion in the horse: effect on movement symmetry and implications for defining soundness. Equine Vet J 2012; 44: 657–663.
17. Marshall JF, Lund DG, Voute LC. Use of a wireless, inertial sensor-based system to objectively evaluate flexion tests in the horse. Equine Vet J Suppl 2012; (43):8–11.
18. Schramme M. Proximal metatarsal lameness in sports horses: a clinical approach to diagnosis, in Proceedings. 59th Annu Meet Am Assoc Equine Pract 2013;250–255.
19. Kramer J, Keegan KG, Kelmer G, et al. Objective determination of pelvic movement during hind limb lameness by use of a signal decomposition method and pelvic height differences. Am J Vet Res 2004; 65: 741–747.
20. Peham C, Licka T, Girtler D, et al. Hindlimb lameness: clinical judgement versus computerised symmetry measurement. Vet Rec 2001; 148: 750–752.
21. Keegan KG, Wilson DA, Kramer J, et al. Comparison of a body-mounted inertial sensor system-based method with subjective evaluation for detection of lameness in horses. Am J Vet Res 2013; 74: 17–24.
22. McCracken MJ, Kramer J, Keegan KG, et al. Comparison of an inertial sensor system of lameness quantification with subjective lameness evaluation. Equine Vet J 2012; 44: 652–656.
23. McLaughlin RM Jr, 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.
