Force platform gait analysis is an important tool for objective assessment of lameness in dogs. The GRFs obtained by use of force platform gait analysis correlate with limb function.1–3 In addition, PVF and VI are commonly evaluated.4–6 Peak vertical force represents the maximal load exerted by a paw during the stance phase, whereas VI represents the area under the force-versus-time curve. The GRF data can be affected by altered limb kinetics, kinematics, and subject comfort during locomotion.7 Ground reaction forces are often evaluated in clinical trials of dogs to assess lameness and to help determine response to treatment.8–12 Typically, the mean of 5 valid trials is used for data collection.10,13–17
Morphological variation is substantial among dog breeds and is an important consideration in clinical studies, which often contain several breeds of dogs. Additional between-subject variation in preferred V, trial repetition, and interday variability can also alter GRFs.3,11,18–22 The technique to best standardize GRF data in heterogeneous canine populations has been studied18,20,23 because these variations can influence interpretation of GRF data. Calculation of an SI is 1 method commonly used to assess treatment effects in clinical studies.2 Use of the SI eliminates the need to normalize data between subjects because the affected limb is compared to the contralateral limb within an animal, with the assumption that the contralateral limb is clinically normal or is considered a suitable control limb. If both of the thoracic or both of the pelvic limbs of a dog are abnormal and subject to variation, perhaps as a result of a treatment effect, then SI values may be unreliable.2,3,10,24 In studies that involve a heterogeneous population of dogs in which SI may be unreliable, GRF data need to be normalized to account for the variation among individual dogs to minimize variance when groups of dogs are compared. It is common to normalize GRF data on the basis of body weight, but this incompletely accounts for subject conformation.20 Use of multiple body measurements better accounts for morphological variation and reduces GRF variability among subjects, although it incompletely accounts for all variance.23
Trunk V is typically measured and considered during the design of force platform gait analysis studies. Narrow V ranges have historically been used to reduce variance.25 However, use of narrow ranges can lead to increased trial repetition and patient fatigue.26 Furthermore, wider V or V* ranges can be used because they do not necessarily have substantial effects on GRF variance, but they do improve acquisition of valid trials and reduce the total number of repetitions needed to collect 5 valid trials for data analysis.26,27
Limb V may also be a source of variance in subjects of different height or body conformation. A dog with shorter limbs requires a more rapid limb V to reach a specific trunk V, compared with the limb V needed for a subject with longer limbs.11,22,28 Furthermore, it has been hypothesized that limb V can fluctuate to some degree among trials and may influence braking and propulsion GRFs in particular, even in a homogenous group of dogs.18 However, braking and propulsion GRFs are less commonly evaluated in detail in clinical trials. Use of limb V in lieu of trunk V in GRF data analysis may help reduce this variation. It is difficult to obtain direct measurements of individual limb V with standard force platform gait analysis equipment and software.11 There is a strong negative correlation between ConT and trunk V.15 Contact time may be used as a surrogate for limb V.11
Normalizing GRF and V data to dimensionless values may help to control for morphological variations in a heterogeneous population of dogs.3,23,27 Consequently, scaling ConT to body size28,29 may also reduce between-subject variability during force platform gait analysis. Fluctuations in trunk V may contribute to GRF variation1,25; therefore, development of ConT and ConT* ranges may be helpful in limiting GRF variance during force platform gait analysis of dogs. The purpose of the study reported here was to evaluate ConT and ConT* ranges for use in analysis of GRF data.
Materials and Methods
Animals
Twenty-nine client-owned dogs with no history of orthopedic disease were recruited for the study. The dogs represented a heterogeneous sample population. Data for these dogs were reported in a previous study.27 In the study reported here, force platform gait analysis was performed at the UW Veterinary Care Hospital at the University of Wisconsin-Madison. The study was approved by the Animal Care and Use Committee of the School of Veterinary Medicine (protocols V1070 and V1600; dates of approval, May 8, 2013, and July 9, 2013, respectively), and owners provided informed consent for inclusion of their dogs.27
Procedures
An orthopedic examination was performed on each dog. Then, height of each dog was measured from the ground contact point of the palmar surface of the paws to the most dorsal aspect of the scapulae (withers). The ridge between the scapulae is the most dorsal point on the trunk and is the standard place to measure height of dogs and horses. The measurement was obtained by use of a framing square with each dog in a square stance. Force platform gait analysis was then conducted with each dog. Dogs were excluded if the SI for PVF was > 15%,27 ConT or ConT* was > 10%, or there was a significant difference in PVF, VI, ConT, or ConT* within a limb pair.
Force platform gait analysis
Trials were collected by use of a single force platforma that measured 3-D forces and impulses. The force platform was embedded level with the floor in the center of a runway that was 7.5 m in length. The V was measured by use of 3 photoelectric cells mounted 1 m apart and centered over the force platform. Dogs were habituated to the force platform and procedures before data were collected.
A handler guided each dog across the force platform; each dog trotted at its preferred trotting V. An observer evaluated each pass over the force platform to confirm that there were ipsilateral thoracic and pelvic limb paw strikes on the left or right side and to verify the dog's gait. A successful trial was defined as an acceleration of ± 0.5 m/s2 and the thoracic limb striking the force platform followed by the ipsilateral pelvic limb striking the platform. Trials were excluded if the dog walked across the force platform or if partial paw strikes were identified.
The force platform was connected by a cable to a data acquisition system and a computer with gait analysis software.b The computer software generated PVF, VI, and ConT data. The PVF and VI were normalized as a percentage of body weight.
The ConT* was calculated by use of the following equation28,29:
where withers height is measured in meters, and g is the gravitational force (9.81 m/s2).
The PVF and VI of the thoracic and pelvic limb pairs were examined for symmetry by use of the following equation13:
where GRFR is the mean of GRFs for the 5 valid trials for the right limb normalized on the basis of body weight, and GRFL is the mean of the GRFs for the 5 valid trials for the left limb normalized on the basis of body weight.
Similarly, SI for the ConT (or ConT*) of the thoracic limbs and pelvic limbs was performed by use of the following equation:
where ConTR (or ConTR*) is the mean of the ConT (or ConT*) for the 5 valid trials for the right limb, and ConTL (or ConTL*) is the mean of the ConT (or ConT*) for the 5 valid trials for the left limb.
An SI value of 0% indicates perfect symmetry between limbs, whereas positive or negative values represent lameness of the right or left limb, respectively. Distributions of dependent variables (PVF, VI, ConT, and ConT*) were examined by use of histogram plots to detect outliers. Shape of the force time curve for individual trials suspected to be outliers was examined.
After analysis of the ConT and ConT* ranges was completed, 2 ConT ranges were deemed superior (thoracic limb, 0.22 to 0.29 seconds; pelvic limb, 0.19 to 0.25 seconds). These ConT ranges were compared with 2 previously described ranges: a withers height V* range27 of 0.6 to 1.05 and a ConT range17 centered on the mean ConT with a distribution of ± 5%. The latter method yielded ranges of 0.25 to 0.27 seconds and 0.21 to 0.23 seconds for the thoracic and pelvic limbs, respectively. All experimental trials were coded into these ranges. On the basis of these ranges, the first 5 valid trials for the left and right limbs were selected for GRF analysis. When 5 valid trials for the left and right limbs were not available for a subject, then data for that subject were not analyzed further.
ConT* range development
Four ranges were created (thoracic limb ConT, pelvic limb ConT, thoracic limb ConT*, and pelvic limb ConT*). Mean thoracic and pelvic limb ConT and ConT* were determined. Ranges of ConT and ConT* were centered on these mean values and involved a multiple of the SD. Ranges included both narrow and wide ranges. Experimental trials then were coded into 1 or more of these ranges.
Statistical analysis
Initially, 5 trials for the left and right limb pairs were obtained on the basis of trials that most closely approximated the mean ConT* for each dog. The PVF, VI, ConT, and ConT* for these selected trials were then analyzed by use of the Student t test for paired data. Dogs were excluded if the SI for PVF was > 15%,27 ConT or ConT* was > 10%, or a significant difference was identified between PVF, VI, ConT, or ConT* within a limb pair. Significance was set at P < 0.0021 by use of the Dunn-Šidák correction for multiple independent testing. A Dunn-Šidák correction of P < 0.05 was calculated by use of the following equation:
where α1 is the Dunn-Šidák corrected P value, α is 0.05, and n is the number of dogs (which was 25 because 4 dogs were excluded from further analysis).
Repeated-measures ANCOVA was used for data analysis. Dog, trial number, limb (left or right), ConT, and ConT* were analyzed as factors in the ANCOVA model. Subsequently, variance effects of the ConT and ConT* ranges were examined in the statistical model. Effect size of each factor in the model was calculated. The Pearson correlation statistic was used to examine the relationship between trunk V and ConT or ConT*. Range efficiency was determined by the proportion of subjects for which there was GRF data. The effect of method for controlling V on PVF and VI was analyzed by use of a repeated-measures ANOVA. All analyses were performed by use of computer software.c,d Data were reported as mean ± SD. Results were considered significant at P < 0.05.
Results
Dogs
Four dogs did not meet the inclusion criteria and were excluded. Two dogs had PVF SI > 15% (27.21% and −17.36%), 1 dog had ConT and ConT* SI > 10% (10.42%), and 1 dog had significant (P = 0.001) differences between limbs for ConT and ConT*. Histogram plots for PVF, VI, ConT, and ConT* for 679 trials were evaluated, and 9 trials were determined to be outliers. A partial paw strike was suspected in 5 trials for 2 dogs because PVF was considerably lower than in the remaining trials for those dogs. Four trials for 3 dogs were also considered outliers. These 9 trials for these 5 dogs were excluded from further analysis.
Thus, 670 trials for 25 dogs were analyzed. Mean ± SD number of trials per dog was 27 ± 4 (range, 19 to 37 trials/dog). Mean age was 5.7 ± 2.9 years (range, 1.4 to 11.1 years). Mean body weight was 25.6 ± 7.5 kg (range, 14.8 to 46.2 kg). Mean withers height was 0.56 ± 0.07 m (range, 0.47 to 0.80 m). Breeds included were Labrador Retriever (n = 6), Australian Shepherd (3), and 1 each of Doberman Pinscher, Springer Spaniel, Nova Scotia Duck Tolling Retriever, Siberian Husky, Portuguese Water Dog, and pit bull-type dog; there were 10 mixed-breed dogs. Thirteen dogs were neutered males, 10 dogs were spayed females, and 2 dogs were sexually intact females.
Trial capture for ConT ranges
Mean ± SD ConT for the thoracic and pelvic limbs was 0.26 ± 0.04 seconds (range, 0.17 to 0.37 seconds) and 0.22 ± 0.03 seconds (range, 0.16 to 0.33 seconds), respectively. Four ConT ranges for the thoracic limb and 4 ConT ranges for the pelvic limb were created (Table 1). As the ConT ranges became increasingly narrower, the proportion of trials captured decreased (Table 2). Four of 8 ConT ranges captured > 60% of trials per dog (thoracic limb ConT ranges, 0.22 to 0.29 and 0.19 to 0.33 seconds; pelvic limb ConT ranges, 0.19 to 0.25 and 0.17 to 0.27 seconds). The ranges of 0.19 to 0.33 seconds and 0.17 to 0.27 seconds captured the greatest proportion of trials for the thoracic and pelvic limbs, respectively, with a total capture rate of 644/670 (96.1%) trials and 638/670 (95.2%) trials, respectively. Mean percentage capture rate per dog for the ConT ranges of 0.19 to 0.33 seconds and 0.17 to 0.27 seconds was 96.3 ± 5.5% and 95.3 ± 11.8%, respectively. The thoracic limb ConT range of 0.22 to 0.29 seconds and pelvic limb ConT range of 0.17 to 0.27 seconds resulted in a total capture rate of 455/670 (67.9%) trials and 514/670 (76.7%) trials, respectively. The ConT range of 0.22 to 0.29 seconds for the thoracic limb and 0.17 to 0.27 second for the pelvic limb resulted in a mean percentage capture rate per dog of 68.7 ± 25.4% and 76.5 ± 23.0%, respectively.
Ranges of ConT and ConT* for the thoracic and pelvic limbs of a heterogeneous population of 25 clinically normal dogs as determined by use of various methods.
| ConT | ConT* | |||
|---|---|---|---|---|
| Method | Thoracic limb | Pelvic limb | Thoracic limb | Pelvic limb |
| Mean ± 0.5SD | 0.24–0.28 | 0.21–0.23 | 1.01–1.15 | 0.87–0.97 |
| Mean - 0.5SD to mean + SD | 0.24–0.29 | 0.21–0.25 | 1.01–1.21 | 0.87–1.02 |
| Mean ± SD | 0.22–0.29 | 0.19–0.25 | 0.95–1.21 | 0.82–1.02 |
| Mean ± 2SD | 0.19–0.33 | 0.17–0.27 | 0.82–1.34 | 0.73–1.12 |
Values for ConT represent the number of seconds, whereas values for ConT* are a dimensionless unit. Values for ConT* were calculated as follows: ConT/(withers height/g)0.5, where withers height is the height measured at the most dorsal aspect of the scapulae in meters, and g is the gravitational force (9.81 m/s2).
Trial capture data and associated GRF values for ConT and ConT* of 670 trials in a heterogeneous population of 25 clinically normal dogs.
| Variable | Range | Total No. of trials | Valid trials per dog (%) | PVF (%BW) | VI (%BW) |
|---|---|---|---|---|---|
| Thoracic limb | |||||
| ConT | 0.24–0.28 | 295 | 44.9 ± 27.3 | 107.3 ± 8.6 | 15.8 ± 1.0 |
| 0.24–0.29 | 333 | 50.8 ± 28.3 | 106.8 ± 8.7 | 15.9 ± 1.0 | |
| 0.22–0.29 | 455 | 68.7 ± 25.4 | 107.9 ± 8.8 | 15.5 ± 1.3 | |
| 0.19–0.33 | 644 | 96.3 ± 5.5 | 107.4 ± 9.3 | 15.6 ± 1.7 | |
| ConT* | 1.01–1.15 | 270 | 40.4 ± 24.8 | 107.6 ± 8.9 | 15.8 ± 1.2 |
| 1.01–1.21 | 348 | 52.4 ± 27.6 | 106.7 ± 8.8 | 15.9 ± 1.2 | |
| 0.95–1.21 | 460 | 69.4 ± 26.2 | 107.9 ± 9.2 | 15.6 ± 1.3 | |
| 0.82–1.34 | 638 | 95.4 ± 8.9 | 107.4 ± 9.5 | 15.7 ± 1.7 | |
| Pelvic limb | |||||
| ConT | 0.21–0.23 | 194 | 28.8 ± 19.4 | 108.3 ± 7.9 | 15.7 ± 1.4 |
| 0.21–0.25 | 333 | 49.6 ± 30.6 | 106.4 ± 8.1 | 16.2 ± 1.4 | |
| 0.19–0.25 | 514 | 76.5 ± 23.0 | 107.9 ± 8.9 | 15.7 ± 1.6 | |
| 0.17–0.27 | 638 | 95.3 ± 11.8 | 107.5 ± 9.4 | 15.6 ± 1.8 | |
| ConT* | 0.87–0.97 | 252 | 37.4 ± 18.6 | 107.5 ± 8.3 | 15.7 ± 1.5 |
| 0.87–1.02 | 343 | 50.9 ± 22.5 | 106.7 ± 8.7 | 15.9 ± 1.5 | |
| 0.82–1.02 | 459 | 68.0 ± 22.7 | 108.1 ± 9.3 | 15.6 ± 1.6 | |
| 0.73–1.12 | 651 | 97.0 ± 5.8 | 107.2 ± 9.5 | 15.7 ± 1.8 | |
Values reported are mean ± SD. Values for the ConT range represent the number of seconds, whereas values for the ConT* range are a dimensionless unit.
%BW = Percentage of body weight.
Trial capture for ConT* ranges
Mean ± SD ConT* for the thoracic and pelvic limbs was 1.08 ± 0.13 (range, 0.76 to 1.51) and 0.92 ± 0.10 (range, 0.65 to 1.30), respectively. Four ConT* ranges for the thoracic limb and 4 ConT* ranges for the pelvic limb were created (Table 1). Similar to results for the ConT ranges, the proportion of trials captured decreased as the CT* ranges became increasingly narrower. Four of 8 ConT* ranges captured > 60% of trials per dog (thoracic limb ConT* ranges, 0.95 to 1.21 and 0.82 to 1.34; pelvic limb ConT* ranges, 0.82 to 1.02 and 0.73 to 1.12). The ConT* ranges of 0.82 to 1.34 and 0.73 to 1.12 captured the greatest proportion of trials for the thoracic and pelvic limbs, respectively, with a total capture rate of 638/670 (95.2%) and 651/670 (97.2%), trials, respectively. Mean percentage capture rate per dog for ranges 0.82 to 1.02 and 0.73 to 1.12 was 95.4 ± 8.9% and 97.0 ± 5.8%, respectively. The thoracic limb ConT* range of 0.95 to 1.21 and pelvic limb ConT* range of 0.82 to 1.02 had a total capture rate of 460/670 (68.7%) and 459/670 (68.5%) trials, respectively. The ConT* range of 0.95 to 1.21 for the thoracic limb and 0.82 to 1.02 for the pelvic limb resulted in a mean percentage capture rate per dog of 69.4 ± 26.2% and 68.0 ± 22.7%, respectively.
Correlation of trunk V with ConT and ConT*
Mean ± SD trunk V was 1.93 ± 0.25 m/s (range, 1.31 to 2.96 m/s). Higher trunk V was correlated with lower ConT and ConT* for both the thoracic and pelvic limbs. There were negative correlations between trunk V and thoracic limb ConT (r2 = −0.214) and pelvic limb ConT (r2 = −0.158; Figure 1). Negative correlations between trunk V and thoracic limb ConT* (r2 = −0.445) and pelvic limb ConT* (r2 = −0.398) were higher than those between trunk V and ConT for the thoracic and pelvic limbs. Correlations between trunk V and ConT or ConT* were higher for the thoracic limb than for the pelvic limb.
Scatterplots of trunk V versus ConT (A and C) and ConT* (B and D) of the thoracic (A and B) and pelvic (C and D) limbs in 670 trials for a heterogeneous population of 25 clinically normal dogs. Each circle represents results of 1 trial. Notice the line of best fit for the data in each panel.
Citation: American Journal of Veterinary Research 79, 5; 10.2460/ajvr.79.5.546
Variance effects on GRF
Dog, ConT, and ConT* had significant effects on PVF and VI for the thoracic and pelvic limbs (Table 3). The effect of trial number on PVF and VI was not significant. Limb (left and right) had significant effects on thoracic limb PVF and VI and pelvic limb VI. The effect size of limb on GRF data was small. Effect sizes ConT and ConT* were also small, and the effects were of similar size. The magnitude of the size of the variance effect on GRF data (in decreasing order) was dog, ConT or ConT*, trial number, and limb.
Variance effects of model factors for thoracic limb and pelvic limb ConT and ConT* on PVF and VI in a heterogeneous population of 25 clinically normal dogs.
| Thoracic limb | Pelvic limb | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| PVF | VI | PVF | VI | ||||||
| Factor analysis | Factor | ES | 95% CI | ES | 95% CI | ES | 95% CI | ES | 95% CI |
| ConT | Dog | 0.689† | 0.639–0.708 | 0.496† | 0.422–0.524 | 0.734† | 0.690–0.751 | 0.557† | 0.490–0.583 |
| ConT | 0.209† | 0.155–0.263 | 0.533† | 0.483–0.576 | 0.114† | 0.071–0.163 | 0.350† | 0.293–0.403 | |
| Trial number | 0.162 | 0.063–0.163 | 0.097 | 0.007–0.085 | 0.052 | 0–0.025 | 0.047 | 0–0.017 | |
| Limb | 0.008† | 0–0.028 | 0.013† | 0.001–0.037 | 0.003 | 0–0.163 | 0.007† | 0–0.025 | |
| ConT* | Dog | 0.703† | 0.656–0.722 | 0.649† | 0.594–0.671 | 0.765† | 0.727–0.780 | 0.570† | 0.505–0.595 |
| ConT* | 0.208† | 0.154–0.262 | 0.534† | 0.484–0.577 | 0.115† | 0.071–0.163 | 0.352† | 0.294–0.405 | |
| Trial number | 0.162 | 0.063–0.162 | 0.098 | 0.008–0.087 | 0.053 | 0–0.026 | 0.046 | 0–0.016 | |
| Limb | 0.008† | 0–0.028 | 0.014† | 0.001–0.037 | 0.003 | 0–0.017 | 0.007† | 0–0.026 | |
Limb represents variance between the left and right thoracic or pelvic limb pairs.
Effect size was significant (P < 0.05).
ES = Effect size.
Effects of ConT and ConT* ranges on GRF variance
Size of the variance effect of all ConT and ConT* ranges was small, compared with the size of the variance effect of dog (Table 4). Trial number did not have a significant effect on GRF variance in the model. Limb had significant effects on thoracic limb PVF for both ConT and ConT*. Size of the variance effect of ConT and ConT* did not decrease consistently as the ranges became increasingly narrower. Significant variance effects were detected for PVF and VI for thoracic limb ConT ranges of 0.24 to 0.29 and 0.19 to 0.33 seconds and pelvic limb ConT ranges of 0.21 to 0.25 and 0.17 to 0.27 seconds. The ConT* ranges had more significant variance effects on GRFs than did ConT ranges. Significant variance effects on PVF and VI were found for thoracic limb ConT* ranges of 1.01 to 1.21 and pelvic limb ConT* ranges of 0.87 to 1.02 and 0.82 to 1.02. Significant variance effects on only PVF were detected for the thoracic limb ConT* range of 0.82 to 1.34. Significant variance effects on only VI were detected for the thoracic limb ConT* ranges of 1.01 to 1.15 and 0.95 to 1.21 and pelvic limb ConT* range of 0.73 to 1.12. In general, size of the variance effect of the ConT* ranges was lower than that for the corresponding ConT ranges.
Variance effects for thoracic limb and pelvic limb ConT and ConT* on PVF and VI in a heterogeneous population of 25 clinically normal dogs.
| PVF (%BW) | VI (%BW) | ||||
|---|---|---|---|---|---|
| Factor analysis | Factor | ES | 95% CI | ES | 95% CI |
| Thoracic limb ConT | Dog | 0.666† | 0.613–0.687 | 0.630† | 0.572–0.653 |
| Trial number | 0.088 | 0–0.0737 | 0.155 | 0.056–0.154 | |
| Limb | 0.007† | 0–0.027 | 0.005 | 0–0.022 | |
| ConT range | |||||
| 0.24–0.28 | 0.005 | 0–0.023 | 0.001 | 0–0.011 | |
| 0.24–0.29 | 0.016† | 0.002–0.042 | 0.011† | 0.001–0.034 | |
| 0.22–0.29 | 0.004 | 0–0.020 | 0.002 | 0–0.016 | |
| 0.19–0.33 | 0.025† | 0.006–0.054 | 0.028† | 0.008–0.058 | |
| Thoracic limb ConT* | Dog | 0.661† | 0.622–0.694 | 0.662† | 0.608–0.683 |
| Trial number | 0.083 | 0–0.0670 | 0.150 | 0.051–0.149 | |
| Limb | 0.011† | 0–0.032 | 0.002 | 0–0.016 | |
| ConT* range | |||||
| 1.01–1.15 | 0.004 | 0–0.021 | 0.012† | 0.001–0.035 | |
| 1.01–1.21 | 0.010† | 0–0.031 | 0.041† | 0.016–0.077 | |
| 0.95–1.21 | 0 | 0–0.010 | 0.011† | 0.001–0.033 | |
| 0.82–1.34 | 0.012† | 0.001–0.035 | 0.003 | 0–0.017 | |
| Pelvic limb ConT | Dog | 0.734† | 0.690–0.751 | 0.534† | 0.464–0.560 |
| Trial number | 0.045 | 0–0.014 | 0.062 | 0–0.038 | |
| Limb | 0.002 | 0–0.015 | 0.005 | 0–0.022 | |
| ConT range | |||||
| 0.21–0.23 | 0.005 | 0–0.023 | 0.003 | 0–0.017 | |
| 0.21–0.25 | 0.021† | 0.004–0.049 | 0.017† | 0.003–0.043 | |
| 0.19–0.25 | 0.001 | 0–0.010 | 0 | 0–0.009 | |
| 0.17–0.27 | 0.009† | 0–0.030 | 0.010† | 0–0.032 | |
| Pelvic limb ConT* | Dog | 0.745† | 0.703–0.761 | 0.561† | 0.495–0.587 |
| Trial number | 0.042 | 0–0.009 | 0.065 | 0–0.043 | |
| Limb | 0.003 | 0–0.019 | 0.002 | 0–0.016 | |
| ConT* range | |||||
| 0.87–0.97 | 0.003 | 0–0.018 | 0.004 | 0–0.021 | |
| 0.87–1.02 | 0.010† | 0.001–0.032 | 0.014† | 0.001–0.038 | |
| 0.82–1.02 | 0.014† | 0.002–0.038 | 0.010† | 0–0.032 | |
| 0.73–1.12 | 0.001 | 0–0.011 | 0.007† | 0–0.025 | |
Values for the ConT range represent the number of seconds, whereas values for the ConT* range are a dimensionless unit. See Table 3 for remainder of key.
Comparison of V and V* ranges and ConT and ConT* ranges
The ConT ranges derived from mean ConT ± 5%17 were 0.25 to 0.27 and 0.21 to 0.23 seconds for the thoracic and pelvic limbs, respectively. One of 25 (4%) dogs had a sufficient number of valid left and right trials for both the thoracic and pelvic limbs by use of these ranges. The ConT ranges reported here resulted in a sufficient number of left and right trials for thoracic and pelvic limbs in 19 of 25 (76%) dogs. All 25 dogs had a sufficient number of valid trials for the V* range of 0.6 to 1.05. No significant differences were found between the GRF data for any range analyzed (Table 5).
Effects of various methods for determining ConT and ConT* ranges on GRFs in a heterogeneous population of 25 clinically normal dogs.
| Thoracic Limb | Pelvic Limb | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| PVF | VI | PVF | VI | |||||||
| Method | n | Dogs (%) | Left | Right | Left | Right | Left | Right | Left | Right |
| Thoracic limb ConT range of 0.25 to 0.27 seconds and pelvic limb ConT range of 0.21 to 0.23 seconds† | 1 | 4 | 108.5 | 107.3 | 15.5 | 15.6 | 71.0 | 71.0 | 9.7 | 9.5 |
| Thoracic limb ConT range of 0.22 to 0.29 seconds and pelvic limb ConT range of 0.19 to 0.25 seconds‡ | 19 | 76 | 107.6 ± 8.3 | 107.7 ± 7.5 | 15.4 ± 1.0 | 15.5 ± 0.9 | 72.5 ± 8.7 | 71.4 ± 7.6 | 9.1 ± 0.8 | 9.0 ± 0.7 |
| V* range of 0.6 to 1.05§ | 25 | 100 | 106.3 ± 8.8 | 105.6 ± 8.6 | 15.4 ± 1.7 | 15.6 ± 1.4 | 71.4 ± 8.2 | 70.4 ± 7.9 | 9.0 ± 0.9 | 9.0 ± 0.9 |
Discussion
In another study,15 a strong inverse correlation was detected between trunk V and ConT in a homogenous population of Greyhounds. The correlation between V and ConT or ConT* was much weaker in the study reported here, which involved a heterogeneous population of client-owned dogs. Morphology and locomotion patterns differ among breeds,11,19 which could influence the relationship between V and ConT. Such variation likely would be minimized in a homogenous population of dogs of a single breed. Interestingly, when dog morphology was accounted for by transforming ConT to a dimensionless value (ie, ConT*) for the heterogeneous population of dogs of the present study, the correlation between trunk V and ConT improved for both the thoracic and pelvic limbs. This suggested that accounting for limb length during analysis of ConT data derived from a heterogeneous population of dogs will help to reduce variance in ConT among subjects. However, the inverse correlation remained weaker than previously reported,15 which suggested that this data transformation did not fully account for all variation in a heterogeneous population. We also found in the present study that thoracic limb ConT or ConT* had a higher correlation coefficient, compared with the respective correlation coefficient for the pelvic limbs. This suggested that trunk V may have more closely approximated thoracic limb ConT or ConT* than did pelvic limb ConT or ConT*. Further investigation of the normalization of the pelvic limb ConT by use of limb length measurements specific for the pelvic limbs (eg, length from the contact surface of a pelvic limb paw to the greater trochanter or wing of the ilium) may further improve the correlation between trunk V and pelvic limb ConT*. Because thoracic limb and pelvic limb ConT are inherently different across a range of trotting Vs,15 a more sophisticated method for data transformation may need to be developed.
An ideal V range would provide an efficient trial capture rate while minimizing confounding effects on GRF data.30 Efficient trial capture helps minimize trial repetition and patient fatigue, which is important during assessment of the gait for animals in which trial repetition may affect lameness severity.7,30 Use of wider V ranges will help improve efficiency for collection of valid trials but may increase variance in GRF data.25 An efficient trial capture rate for the present study was defined as capture of > 60% of the trials per dog.30 Two ranges each for the thoracic limb ConT (0.22 to 0.29 and 0.19 to 0.33 seconds), pelvic limb ConT (0.19 to 0.25 and 0.17 to 0.27 seconds), thoracic limb ConT* (0.95 to 1.21 and 0.82 to 1.34), and pelvic limb ConT* (0.82 to 1.02 and 0.73 to 1.12) yielded capture rates > 60%.
Differences among dogs contribute the most to GRF variance.23,28,30 In addition to obvious morphological differences, subtle between-subject differences (eg, the pattern of locomotion in individual dogs) may also contribute to variance.21 Although effects of ConT and ConT* ranges on GRF data were variable, in general, the narrowest ranges had fewer significant effects on GRF data. Interestingly, ConT ranges had fewer significant effects on GRFs, compared with effects for the ConT* ranges. However, the ConT* ranges often had smaller effect sizes, compared with effect sizes for the associated ConT ranges. Subtle differences in the pattern of locomotion among dogs could explain these findings. Ranges with a high capture rate, no significant effects on GRFs, and small effect sizes were the thoracic limb ConT range of 0.22 to 0.29 seconds and pelvic limb ConT range of 0.19 to 0.25 seconds. This suggested that use of ConT, rather than ConT*, would be preferred as a control variable for collection of data during force platform gait analysis.
The ConT ranges proposed in another study17 were extremely inefficient for capturing trials and would not be suitable for clinical use in a heterogeneous population of dogs. The best withers height V* range in another study27 performed more efficiently than the best ConT ranges of the present study for capturing a sufficient number of valid trials. The most efficient V* range in that study27 was associated with fewer significant effects on GRFs, compared with results for the most efficient ConT ranges of the study reported here. The V* range was, in part, more efficient because a single trial allowed collection of valid GRF data from both limb pairs. For ConT, separate ranges are used for the thoracic and pelvic limbs. It was uncommon for a single trial to be valid for both the thoracic and pelvic limbs. Thus, more trials would be necessary to collect 5 valid ConT trials for each limb. However, clinical trials typically assess either thoracic or pelvic limb lameness, which would eliminate the need for separate thoracic and pelvic limb ConT ranges when assessing only 1 limb pair. Regardless, the V* range would have a more efficient trial capture rate. This finding supported the notion that limb V varies within a trial, but that use of the withers height V* range of 0.6 to 1.05 would be a more useful approach in force platform gait analysis.
The population of the study reported here differed with regard to body weight, breed, and height, which is typical of a clinically relevant population. However, the overall number of dogs was relatively small, and a range created by use of data from a larger group of dogs may better represent overall variation in morphology of dog breeds. Ideally, the ConT or V range should capture the preferred ConT or V of all subjects in a sample population. The ConT ranges reported here could be used when designing studies for force platform gait analysis. In a previous study,17 investigators determined that ConT and V could be interchangeable as a control variable in animals with mild to moderate lameness, and results for clinically normal dogs of the present study supported this notion, although correlations are weaker in a heterogeneous population.
Results of the present study suggested that ConT ranges can be used to control GRF variance in force platform gait analysis. Although trunk V correlated more closely with ConT*, ConT* ranges had more significant variance effects on GRFs than did the ConT ranges. Additional studies are needed to determine whether results for healthy dogs without lameness would be relevant for a sample population of lame dogs, including the relative efficiencies of V* or ConT* ranges for capturing GRF data from subjects in a clinical trial without causing confounding effects on GRF data. Unless there are important differences that suggest ConT is advantageous over V* ranges in lame dogs, we believe that use of V* ranges to control for GRF variance in force platform gait analysis is the best approach.
ABBREVIATIONS
| ConT | Contact time |
| ConT* | Relative contact time |
| GRF | Ground reaction force |
| PVF | Peak vertical force |
| SI | Symmetry index |
| V | Velocity |
| V* | Relative velocity |
| VI | Vertical impulse |
Footnotes
OR6-6-1000 biomechanics platform with SGA6-4 signal conditioner/amplifier, Advanced Mechanical Technologies Inc, Waterton, Mass.
Acquire, version 7.30, Sharon Software Inc, Dewitt, Mich.
STATA, version 14.0, Stata Corp LP, College Station, Tex.
Prism, version 7.0, GraphPad Software Inc, La Jolla, Calif.
References
1. Evans R, Horstman C, Conzemius M. Accuracy and optimization of force platform gait analysis in Labradors with cranial cruciate disease evaluated at a walking gait. Vet Surg 2005;34:445–449.
2. Fanchon L, Grandjean D. Accuracy of asymmetry indices of ground reaction forces for diagnosis of hind limb lameness in dogs. Am J Vet Res 2007;68:1089–1094.
3. Voss K, Imhof J, Kaestner S, et al. Force plate gait analysis at the walk and trot in dogs with low-grade hindlimb lameness. Vet Comp Orthop Traumatol 2007;20:299–304.
4. DeCamp CE. Kinetic and kinematic gait analysis and the assessment of lameness in the dog. Vet Clin North Am Small Anim Pract 1997;27:825–840.
5. Budsberg SC, Rytz U, Johnston SA. Effects of acceleration on ground reaction forces collected in healthy dogs at a trot. Vet Comp Orthop Traumatol 1999;12:15–19.
6. McLaughlin RM. Kinetic and kinematic gait analysis in dogs. Vet Clin North Am Small Anim Pract 2001;31:193–201.
7. Beraud R, Moreau M, Lussier B. Effect of exercise on kinetic gait analysis of dogs afflicted by osteoarthritis. Vet Comp Orthop Traumatol 2010;23:87–92.
8. Budsberg SC, Verstraete MC, Soutas-Little RW, et al. Force plate analyses before and after stabilization of canine stifles for cruciate injury. Am J Vet Res 1988;49:1522–1524.
9. Voss K, Damur DM, Guerrero T, et al. Force plate gait analysis to assess limb function after tibial tuberosity advancement in dogs with cranial cruciate ligament disease. Vet Comp Orthop Traumatol 2008;21:243–249.
10. Weinstein JI, Payne S, Poulson JM, et al. Use of force plate analysis to evaluate the efficacy of external beam radiation to alleviate osteosarcoma pain. Vet Radiol Ultrasound 2009;50:673–678.
11. Budsberg SC, Verstraete MC, Soutas-Little RW. Force plate analysis of the walking gait in healthy dogs. Am J Vet Res 1987;48:915–918.
12. Malek S, Sample SJ, Schwartz Z, et al. Effect of analgesic therapy on clinical outcome measures in a randomized controlled trial using client-owned dogs with hip osteoarthritis. BMC Vet Res 2012;8:185.
13. Budsberg SC, Jevens DJ, Brown J, et al. Evaluation of limb symmetry indices, using ground reaction forces in healthy dogs. Am J Vet Res 1993;54:1569–1574.
14. Rumph PF, Kincaid SA, Baird DK, et al. Vertical ground reaction distribution during experimentally induced acute synovitis in dogs. Am J Vet Res 1993;54:365–369.
15. McLaughlin RM, Roush JK. Effects of subject stance time and velocity on ground reaction forces in clinically normal Greyhounds at the trot. Am J Vet Res 1994;55:1666–1671.
16. Roush JK, McLaughlin RM. Effects of subject stance time and velocity on ground reaction forces in clinically normal Greyhounds at the walk. Am J Vet Res 1994;55:1672–1676.
17. Renberg WC, Johnston SA, Ye K, et al. Comparison of stance time and velocity as control variables in force plate analysis of dogs. Am J Vet Res 1999;60:814–819.
18. Jevens DJ, Hauptman JG, Decamp CE, et al. Contributions to variance in force-plate analysis of gait in dogs. Am J Vet Res 1993;54:612–615.
19. Mölsä SH, Hielm-Björkman AK, Laitinen-Vapaavuori OM. Force platform analysis in clinically healthy Rottweilers: comparison with Labrador Retrievers. Vet Surg 2010;39:701–707.
20. Voss K, Galeandro L, Wiestner T, et al. Relationships of body weight, body size, subject velocity, and vertical ground reaction forces in trotting dogs. Vet Surg 2010;39:863–869.
21. Voss K, Wiestner T, Galeandro L, et al. Effect of dog breed and body conformation on vertical ground reaction forces, impulses, and stance times. Vet Comp Orthop Traumatol 2011;24:106–112.
22. Kim J, Kazmierczak KA, Breur GJ. Comparison of temporospatial and kinetic variables of walking in small and large dogs on a pressure-sensing walkway. Am J Vet Res 2011;72:1171–1177.
23. Krotscheck U, Todhunter RJ, Nelson SA, et al. Precision and accuracy of ground reaction force normalization in a heterogeneous population of dogs. Vet Surg 2014;43:437–445.
24. Oosterlinck M, Bosmans T, Gasthuys F, et al. Accuracy of pressure plate kinetic asymmetry indices and their correlation with visual gait assessment scores in lame and nonlame dogs. Am J Vet Res 2011;72:820–825.
25. Riggs CM, DeCamp CE, Soutas-Little RW, et al. Effects of subject velocity on force plate measured ground reaction forces in healthy Greyhounds at the trot. Am J Vet Res 1993;54:1523–1526.
26. Hans EC, Zwarthoed B, Seliski J, et al. Variance associated with subject velocity and trial repetition during force platform gait analysis in a heterogeneous population of clinically normal dogs. Vet J 2014;202:498–502.
27. Volstad N, Nemke B, Muir P. Variance associated with the use of relative velocity for force platform gait analysis in a heterogeneous population of clinically normal dogs. Vet J 2016;207:80–84.
28. Pierrynowski MR, Galea V. Enhancing the ability of gait analyses to differentiate between groups: scaling gait data to body size. Gait Posture 2001;13:193–201.
29. Hof AL. Scaling gait data to body size. Gait Posture 1996;4:222–223.
30. Mickelson MA, Vo T, Piazza A, et al. Influence of trial repetition on lameness during force platform gait analysis in a heterogeneous population of clinically lame dogs each trotting at its preferred velocity. Am J Vet Res 2017;78:1284–1292.
