Gait in dogs with neurologic disease is often assessed with subjective numeric rating scales. Olby et al1 proposed a disability index with the advantage that the dog can be scored without ambulation. That system uses a numeric rating system in which the dog is assigned a number on the basis of the level of neurologic dysfunction. The index has several disadvantages, including nonlinear levels of disability represented by a linear scale, evaluator subjectivity, and low sensitivity.2,3 Subjective evaluation of patients may allow evaluator bias because it is not always possible to adequately mask an evaluator. For example, if a comparison were being made between surgical and nonsurgical management of a neurologic disorder, masking would be difficult because of preparation and wound closure. Sham surgery avoids this complication, but in clinical trials, sham surgery is difficult to justify ethically. Additionally, intra- and interobserver variation may make it difficult to distinguish between groups, collaborate with multiple locations, or compare studies.
Numeric rating systems have inherent disadvantages, including the requirement of nonparametric statistical testing if there are < 5 categories. The Olby et al1 disability index uses a 14-point numeric scoring system; however, the improvement between adjacent scores may not be equivalent. For example, to improve from a score of 4 to 5, a dog with non–weight-bearing activity in > 1 joint < 50% of the time must improve to weight bearing ≥ 50% of the time. To improve from a score of 5 to 6, the dog must become 100% weight bearing. Moving from a score of 5 to a score of 6 is more of an improvement than from a score of 4 to a score of 5. Subjective rating scales are also insensitive at detecting subtle differences between groups.2–4
Kinematic gait analysis is a sensitive, objective measure of gait. Two- or 3-D kinematic gait analysis is used to quantify neurologic dysfunction in ambulatory dogs.5,a Cameras are used to record ambulation with joints identified by use of reflective markers or digital identification. Specialized software is then used to measure characteristics such as changes in joint angles during the various phases of gait as well as angular limb velocities. Range of motion in each joint and some spatiotemporal characteristics can be measured but not ground reaction forces. Error is introduced with placement and movement of markers relative to the actual joint, and some subjectivity is inherent in the videography digitizing process.2–4
Force platform gait analysis is considered the gold standard for quantifying limb function in dogs with orthopedic disease. It is objective and far more sensitive than visual assessment of gait.6–8 Pressure walkways have been validated as an alternative to the force platform, which uses symmetry of ground reaction forces, peak vertical force, and vertical impulse.9,10 A single pressure walkway can accommodate dogs of greater size and measure consecutive footfalls, whereas force platforms of multiple sizes placed in series are necessary for dogs of different sizes and for consecutive footfalls. This can be cost prohibitive. Pressure platforms also measure spatiotemporal gait characteristics. These characteristics have not been used to assess gait in dogs with orthopedic or neurologic disease.
The spatiotemporal characteristics velocity, stance time, swing time, stride time, and stride length have been used in human and basic science research to evaluate neurologic gait disturbance.11–14 In rats, stance time, swing time, and stride time are shorter after induced spinal cord injury.15 Velocity, stride length, swing time, and stance time have been used to help evaluate physical therapy patients and measure response to treatment for neurologic disorders.11–13
We hypothesized that spatiotemporal characteristics may be used to assess gait in dogs with neurologic disease. The objective of the study reported here was to determine the spatiotemporal gait characteristics and associated covariates of clinically normal dogs and dogs with spinal cord disease.
Materials and Methods
Dogs were recruited at 3 veterinary teaching hospitals with institutional animal care and use committee approval and informed owner consent. All dogs included in the study were either clinically normal (no clinical signs or history of neurologic or orthopedic disease) or had neurologic disease localized to the T3-L3 spinal cord segment including hind limb ataxia. Dogs were excluded if they were nonambulatory or aggressive. Age, breed, height, weight, and group were recorded for each dog. Height was measured with the dog standing normally on a flat surface, from the surface to the top of the withers.
All dogs walked at their own preferred velocity 3 to 5 times (each instance being termed a pass) over 1 of 2 commercially available pressure walkways referred to as walkway 1b and walkway 2c after being acclimated to the walkway. The pass was valid if the dog walked without noticeable distraction, visible acceleration, or assistance for a minimum of 2 m and stayed within the bounds of the pressure mat of the walkway with all 4 feet. Dog handlers were trained and experienced. Stride time, stance time, swing time, stride length, and velocity were recorded from the pressure walkway.
The software used with walkway 2 requires manual foot identification but calculates the gait characteristics automatically after the identification is complete. Acceleration and peak vertical force are not measured with this system; however, limb symmetry is measured. In contrast, walkway 1 software uses manual measurement of stride length, stride time, and stance time. Stride length was defined as the distance from heel to heel, and stride time as the time immediately following toe off (0 pressure) of a foot to the toe off of the same foot following a stride. Stance time was measured for each foot from the time it appeared on the pressure mat (> 0 pressure) until it left the mat. Swing time was calculated by subtracting stance time from stride time. Velocity and acceleration were calculated by use of distance and time measurements from the pass over the platform. Peak vertical force was measured for each foot and expressed as a percentage of body weight. Peak vertical force was also used to calculate limb symmetry. Two trained investigators measured passes to facilitate timely data collection.
Statistical analysis—The gait characteristics (eg, stride time, stance time, stride length, and swing time) were potentially biased by the uncontrolled covariates (eg, velocity, acceleration, height, and weight), which could lead to erroneous interpretations of results and group comparisons. Therefore, regression analysis was used to determine the effect of the covariates.
The second step was to evaluate 2 statistical methods, a simple and a multivariate method, to remove bias. With the simple method, the variable values were divided by height values. Use of this method was motivated by observing that height is correlated to velocity and weight, and adjusting for height may implicitly account for the other 2 covariates. The multivariate method used multivariate regression to control for height, velocity, and weight, explicitly. The bias control methods were evaluated by comparing pairwise variable-covariate r2 values before and after adjustment. The method with the lower r2 value removed more bias. Groups were compared for differences in adjusted characteristics means by use of t tests (for equal group variances), Welsh t tests (for unequal group variances), or matched-pairs t tests (for limb symmetry).
A subset analysis with a t test was also performed in the group of clinically normal dogs to determine whether the chondrodystrophic nature of some breeds (eg, Dachshund and Pekinese) influenced the spatiotemporal variables after adjustment for covariates (eg, height and weight). For all comparisons, a value of P < 0.05 was considered significant.
Results
There were 42 clinically normal dogs, with a median age of 5 years (range, 1 to 12.7 years), including 20 males and 22 females. Breeds commonly represented in the clinically normal group include mixed-breed (n = 7), Great Dane (6), Dachshund (5), Pekinese (4), and Labrador Retriever (4). Mean ± SEM height and weight were 48.6 ± 3.4 cm and 22.8 ± 3.1 kg, respectively. Twenty-three dogs were measured by use of walkway 1 (6 at the University of Minnesota and 17 at the University of Illinois) and 19 by use of walkway 2 (at Iowa State University).
Twenty-four dogs with neurologic disease (22 with intervertebral disk disease and 2 with spinal cord trauma) with a median age of 5.5 years, ranging from 1.7 to 14 years, were admitted to the study. The spatiotemporal characteristics of 6 dogs were measured by use of walkway 2 (at Iowa State University), and the spatiotemporal characteristics of the remaining 18 dogs were measured by use of walkway 1 (at the University of Illinois). The Dachshund breed was overrepresented in the neurologic disease group (n = 17). The number of males (n = 13) and females (11) were similar. Mean ± SEM height and weight of dogs with neurologic disease were 29.9 ± 2.6 cm and 9.2 ± 1.2 kg, respectively, which were significantly (P = 0.01) smaller than those of the clinically normal group. The prevalent condition causing clinical signs was intervertebral disk disease in 22 dogs and spinal cord trauma of unknown cause in 2 dogs.
In the clinically normal dogs, the mean (range) velocity was 1.0 m/s (0.45 to 1.45 m/s) Mean acceleration (walkway 1 only [n = 14]) was 0.06 m/s2, and the range (0.01 to 0.14 m/s2) remained within acceptable guidelines used in orthopedic gait analysis.6,16 Similar to the clinically normal dogs, the velocities were wide ranging in the dogs with neurologic disease. Mean velocity of dogs with neurologic disease was 0.79 m/s, with a range from 0.36 to 1.39 m/s. Acceleration (walkway 1 only [n = 12]) was acceptable with a mean of 0.1 m/s2 and range of 0.02 to 0.2 m/s2.
Because the groups differed in height, weight, and velocity, the 10th, 25th, 50th, 75th, and 90th percentiles were determined (Table 1). For each variable, this revealed that ≥ 25% of dogs had similar values between the 2 groups, supporting the validity of the regression analysis.
Percentile distributions of height, weight, and velocity in 24 dogs with neurologic disease and 42 clinically normal dogs.
| Percentile | Height (cm) | Weight (kg) | Velocity (m/s) | |||
|---|---|---|---|---|---|---|
| Neurologic disease | Clinically normal | Neurologic disease | Clinically normal | Neurologic disease | Clinically normal | |
| 90th | 50.33 | 80.10 | 19.70 | 57.74 | 1.25 | 1.34 |
| 75th | 33.00 | 64.63 | 9.09 | 31.94 | 0.95 | 1.29 |
| 50th | 23.50 | 54.00 | 7.23 | 19.55 | 0.71 | 1.07 |
| 25th | 21.15 | 28.00 | 5.91 | 6.83 | 0.62 | 0.68 |
| 10th | 21.00 | 20.30 | 5.13 | 4.85 | 0.50 | 0.53 |
There was no significant relationship between age and stride length, stride time, or stance time. Although significant (P = 0.01), only 25% of the variation in swing time was attributable to variation in age. Swing time and stance time were not related to sex; however, stride length was longer in females than males (P = 0.04) and stride time was longer in females, although not significantly (P = 0.06). Potential covariates, such as velocity, height, and weight, were not significantly different between males and females. Age and sex were balanced across groups.
The velocity of the neurologic disease group (0.79 m/s) was significantly (P = 0.01) lower than the clinically normal group (1.0 m/s). However, when velocity was adjusted for height by use of regression analysis, dogs with neurologic disease were no longer significantly different from clinically normal dogs. Acceleration was not significantly (P = 0.19) different between the 2 groups.
The relationships between stride length, stride time, stance time, or swing time and velocity, acceleration, height, and weight were determined for clinically normal dogs (Table 2). For simplicity, r2 values (expressed as a percentage) are given for the left forefoot because the results for all limbs were similar. There was no significant relationship between acceleration (n = 14 dogs) and any of the 4 characteristics (stride time, stride length, swing time, and stance time).
Relationship (r2 expressed as a percentage) between independent potential covariates of gait and spatiotemporal gait characteristics of the left forefoot in 42 clinically normal dogs.
| Characteristic | Covariate | |||
|---|---|---|---|---|
| Velocity | Acceleration | Height | Weight | |
| Stride length | 70 | NS | 90 | 80 |
| Stride time | 20 | NS | 70 | 70 |
| Stance time | 10 | NS | 60 | 60 |
| Swing time | 40 | NS | 70 | 50 |
| Height | 40 | NP | NP | 90 |
NP = Not performed. NS = Not significant (P > 0.05; therefore, r2 is irrelevant).
Height had the strongest relationship with stride length, stride time, stance time, and swing time. Additionally, height had a direct relationship with velocity and weight (Table 2). Therefore, the gait characteristics were expressed as a percentage of height. A regression analysis was subsequently performed to determine whether a relationship still existed between velocity or weight and the gait characteristics. Stride length as a percentage of height no longer had a relationship with weight (P = 0.65) or velocity (P = 0.23). However, for weight and velocity, respectively, the relationship persisted with stride time (r2 = 0.36 and 0.70; P = 0.01), stance time (r2 = 0.28 and 0.70; P = 0.01), and swing time (r2 = 0.45 and 0.59; P = 0.01). When compared, the stride length as a percentage of height was significantly shorter in the forelimbs of dogs with neurologic disease (P < 0.05); however, there was no difference in hind limbs between groups.
Although expressing stride length as a percentage of height allowed direct comparison of the groups, the remaining gait characteristics (stride time, stance time, and swing time) required multivariate analysis to remove the influence of velocity and weight prior to comparing groups. By use of multivariate analysis, the height, weight, and velocities of dogs were plotted against each of the 4 gait characteristics. The 2 groups were compared by use of the residuals from the multivariate plot.
Stride length, stride time, and stance time were all decreased in the forelimbs of dogs with neurologic disease, compared with clinically normal dogs. In addition, the hind limbs of dogs with neurologic disease typically had increased swing time. Because the residuals of each of the gait characteristics were compared, the factor of increase was calculated for each of the significant characteristics for evaluation of clinical differences. In the forelimbs, mean stride length of clinically normal dogs was 16 times greater and mean stride time was 20 times greater than those with neurologic disease. Stance time in clinically normal dogs was twice that of neurologic dogs in the forelimbs. In the hind limbs, swing time in the dogs with neurologic disease was 4 times greater than that of the clinically normal dogs.
To further elucidate this pattern, the dogs measured via walkway 1 (n = 14 clinically normal dogs; 12 with neurologic disease) were analyzed for the difference between groups in peak vertical force as a percentage of body weight for each limb by use of the residuals after regression analysis to eliminate the influence of velocity. There were no significant differences in weight bearing by the hind limbs between groups, but the dogs with neurologic disease had significantly higher (3 times) peak vertical force on the forelimbs (P = 0.04 for the right forelimb and P < 0.01 for the left forelimb).
Additionally, limb symmetry for these values was compared between groups by use of matched-pairs analysis. For all characteristics, dogs in both groups had ground reaction forces that were symmetric between the right and left limbs. As expected, forelimb ground reaction forces were significantly (P = 0.01) greater than hind limb ground reaction forces for dogs in both groups. Dogs in the group with neurologic disease had significantly greater forelimb-to-hind limb asymmetry than did clinically normal dogs.
In the clinically normal group, chondrodystrophic dogs (n = 11) were compared with nonchondrodystrophic dogs (31) after adjusting gait characteristics for covariates. There were no significant differences for any of the spatiotemporal characteristics.
Discussion
In this study, gait characteristics were evaluated for a relationship to potential covariates that may not be controlled between groups, including height, weight, velocity, acceleration, age, and sex. Height, weight, and velocity were the only unbalanced independent variables with a significant impact on the spatiotemporal characteristics. Because a simple method of normalizing to height did not eliminate the influence of all covariates (except for stride length), multivariate regression was performed for all 3 covariates versus the 4 gait characteristics. The residuals were then compared between clinically normal dogs and dogs with neurologic disease. This is a tool used to remove the influence of known biases.
Changes in spatiotemporal characteristics of the dogs with neurologic disease included decreased stride length, stride time, and stance time in the forelimbs and increased swing time in the hind limbs, compared with clinically normal dogs. Peak vertical force was increased in the forelimbs of dogs with neurologic disease, and the forelimb-to-hind limb symmetry was decreased in dogs with neurologic disease, compared with clinically normal dogs.
Overall, the forelimbs of dogs with neurologic disease had shorter stride lengths, decreased stride times, and decreased stance times, compared with clinically normal dogs. The changes in spatiotemporal characteristics in the forelimbs were likely the result of shifting the weight forward because of ataxia in the hind limbs. This explanation was supported by the finding that the peak vertical forces of the forelimbs of dogs with neurologic disease were 3 times greater than that of the clinically normal dogs; that is, all dogs put more weight on their forelimbs but dogs with T3-L3 spinal segment disease shifted their weight forward to a greater extent, likely as a result of hind limb ataxia.
Gait analysis in rats with experimentally induced spinal cord injury also reveals changes in spatiotemporal characteristics of the forelimbs, compared with the hind limbs.15 Rats with spinal cord injury have decreased stride length and increased stance time in the forelimbs, compared with sham-operated rats.15,17 Alternatively, stride length is not affected by spinal cord injury or sciatic nerve transection.15,17 It is difficult to interpret or compare the rat gait data with data from the present study because the rats are allowed to move at any velocity and any gait. Differing gaits between individuals and groups increase the variation in results and may interfere with comparisons. In the study reported here, velocity had a significant effect on all of the spatiotemporal characteristics.
The spatiotemporal gait variable in the hind limbs that was different in dogs with neurologic disease, compared with clinically normal dogs, was swing time, and as expected, dogs with neurologic disease had increased swing times. This was likely a direct result of ataxia and because short, quick strides in the forelimbs help to maintain balance in the hind limbs. It is possible but less likely that the increased swing time was a secondary effect of the weight shift rather than a direct result of the ataxia.
The dogs with neurologic disease were ataxic in the hind limbs, but we found no differences in stride length, stride time, or stance time, compared with clinically normal dogs. One explanation for this finding may be associated with the variability of the data in dogs with neurologic disease. The means of the residuals for stride length, stride time, and stance time were calculated for each limb of each dog prior to comparison. This was required to prevent artificial inflation of the sample size during statistical analysis. One consequence was that the variation of the individual limb was not accounted for in the analysis. For example, a hind limb in a dog with neurologic disease may have large variability in stride length from stride to stride. Some strides may be long, and others may be short; however, the mean of those long and short strides may be similar to the mean stride of a clinically normal dog with consistent stride lengths. This has also been noted in the human literature where individual variation, rather than the mean value of the variable, is used to predict falling or unsteadiness.12,14,18 It is possible that the coefficient of variation of each variable would be a better measure of difference between the 2 groups rather than the mean for the hind limbs.
Stratification of known covariates is common in force platform gait analysis for orthopedic conditions to eliminate variation caused by morphological differences.6,19 Future studies could limit height, weight, and velocity during case recruitment to avoid complicated statistical adjustment. However, incorrectly identifying strata endpoints may lead to biased inferences. This was not a problem with the residual-based comparisons used in the present study. Also, by choosing to study dogs of all sizes, a more balanced comparison of the relationships between morphological characteristics and the spatiotemporal characteristics could be studied. This allowed us to identify the covariates that needed to be most closely controlled during population recruitment and statistical analysis.
Dogs were allowed to walk at their own preferred velocity because the typical pace of dogs with severe ataxia was unknown. By artificially slowing or increasing the walking velocity, changes in spatiotemporal characteristics may have been artificially exacerbated. This is supported by evidence in human medicine where patients are instructed to walk at a natural or comfortable pace.12–14,20 It is unknown whether velocity may be controlled without artificially changing results in dogs with a narrow range of debilitation. In the present study, dogs' clinical status ranged from severely ataxic to almost clinically normal and this likely contributed to the large range in velocities in this group. However, clinically normal dogs also had a large range of velocities, which contributed to variation. It is common in human diseases for patients to have changes in natural velocity in addition to the other spatiotemporal characteristics. Thus, it is important to note that we found no differences in velocity that could be attributed to the dogs with neurologic disease.
Conformational (eg, chondrodystrophic) differences may also affect the normal spatiotemporal patterns. However, in the clinically normal group of dogs, the spatiotemporal values were not different between chondrodystrophic and nonchondrodystrophic dogs. This provided evidence that the differences in gait were attributable to the dysfunction in the neurologic group rather than conformational differences.
In the group with neurologic disease, Dachshunds were overrepresented, and it is possible that the changes in gait in dogs with neurologic disease reflected those of Dachshunds rather than the general population. However, unlike unilateral orthopedic disease, compensation for T3-L3 myelopathy is theoretically limited to adjustments made to the forelimbs. Despite this argument, generalizing the neurologic changes in spatiotemporal characteristics of gait to the general population should be done with caution.
Although the inclusion criteria for the neurologic group included all dogs with a T3-L3 myelopathy, 92% of the affected dogs had intervertebral disk disease. Therefore, the conclusions may only be applicable to dogs with this disease. It is unknown whether other diseases that affect this region of the spinal cord may affect gait differently or whether there is a spectrum of spatiotemporal differences in gait in dogs with intervertebral disk disease that depends on the specific lesion present.
Age and sex are covariates of spatiotemporal gait analysis in humans.17,21 Swing time increased as age increased in the present study; however, the r2 value was only 0.25, so age had a limited effect. The disparity between findings in humans and the dogs in the present study may be attributable to the relative health of the clinically normal dogs, compared with aged persons. Female dogs had greater stride length and stride times than did male dogs in this study. Velocity, height, and weight were not different between male and female dogs. This was in contrast to humans, in which men typically have longer stride lengths than do women; however, there were similar ratios of males to females in both groups of dogs in the present study, so the effect of sex differences between groups was minimal. These characteristics should be evaluated to ensure equality across groups in future studies.
Other potential covariates that were not evaluated in this study included handler, location (ie, different universities), and brand of pressure walkway. All handlers were trained and experienced in gait analysis for orthopedic conditions. Locations and pressure walkways could not be evaluated because of the variations in dog size, diseases, and numbers contributed from each location. The pressure walkways have been used for recording spatiotemporal characteristics in human gait analysis.11,13,20,22 More specifically, walkway 2 has been validated for both temporal and spatial characteristics in humans and special software has been developed for dogs.11 Walkway 1 has been validated for orthopedic use in dogs.10 Although use of 1 walkway would be ideal, we believe that both walkways are reliable. Future comparisons between the 2 walkways would best be performed side by side with the same animals and handlers.
Jeffery N. Quantitative outcome measures after spinal cord injury (abstr), in Proceedings. Annu Meet Eur Coll Vet Med 2008;539–540.
Tekscan Walkway, Tekscan Inc, Boston, Mass.
GAITrite System with GAITfour software, CIR Systems Inc, Havertown, Pa.
References
- 1.↑
Olby NJ, De Risio L, Muñana KR, et al.Development of a functional scoring system in dogs with acute spinal cord injuries. Am J Vet Res 2001;62:1624–1628.
- 2.
Basso DM. Behavioral testing after spinal cord injury: congruities, complexities, and controversies. J Neurotrauma 2004;21:395–404.
- 3.
Giglio CA, Defino HLA, da-Silva CA, et al.Behavioral and physiological methods for early quantitative assessment of spinal cord injury and prognosis in rats. Braz J Med Biol Res 2006;39:1613–1623.
- 4.
Muir GD, Webb AA. Assessment of behavioral recovery following spinal cord injury in rats. Eur J Neurosci 2000;12:3079–3086.
- 5.↑
Gradner G, Bockstahler B, Peham C, et al. Kinematic study of back movement in clinically sound Malinois dogs with consideration of the effect of radiographic changes in the lumbosacral junction. Vet Surg 2007;36:472–481.
- 6.
Conzemius MG, Evans RB, Besancon MF, et al. Effect of surgical technique on limb function after surgery for rupture of the cranial cruciate ligament in dogs. J Am Vet Med Assoc 2005;226:232–236.
- 7.
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.
- 8.
Waxman AS, Robinson DA, Evans RB, et al. Relationship between objective and subjective assessment of limb function in normal dogs with an experimentally induced lameness. Vet Surg 2008;37:241–246.
- 9.
Horstman CL, Conzemius MG, Evans R, et al. Assessing the efficacy of perioperative oral carprofen after cranial cruciate surgery using noninvasive, objective pressure platform gait analysis. Vet Surg 2004;33:286–292.
- 10.↑
Besancon MF, Conzemius MG, Derrick TR, et al. Comparison of vertical forces in normal greyhounds between force platform and pressure walkway measurement systems. Vet Comp Orthop Traumatol 2003;16:153–157.
- 11.↑
Youdas JW, Hollman JH, Aalbers MJ, et al. Agreement between walkway 2 and a stopwatch-footfall count method for measurement of temporal and spatial gait characteristics. Arch Phys Med Rehabil 2006;87:1648–1652.
- 12.
Merory JR, Wittwer JE, Rowe CC, et al. Quantitative gait analysis in patients with dementia with Lewy bodies and Alzheimer's disease. Gait Posture 2007;26:414–419.
- 13.
Balasubramanian CK, Bowden MG, Neptune RR, et al. Relationship between step length asymmetry and walking performance in subjects with chronic hemiparesis. Arch Phys Med Rehabil 2007;88:43–49.
- 14.
Kim CM, Eng JJ. Symmetry in vertical ground reaction force is accompanied by symmetry in temporal but not distance variables of gait in persons with stroke. Gait Posture 2003;18:23–28.
- 15.↑
McEwen ML, Springer JE. Quantification of locomotor recovery following spinal cord contusion in adult rats. J Neurotrauma 2006;23:1632–1653.
- 16.
Evans R, Gordon W, Conzemius M. Effect of velocity on ground reaction forces in dogs with lameness attributable to tearing of the cranial cruciate ligament. Am J Vet Res 2003;64:1479–1481.
- 17.
Deumans R, Jaken RJP, Marcus MAE, et al. The CatWalk gait analysis in assessment of both dynamic and static gait changes after adult rat sciatic nerve resection. J Neurosci Methods 2007;164:2120–2130.
- 18.
Verghese J, Wang C, Holtzer R, et al. Quantitative gait dysfunction and risk of cognitive decline and dementia. J Neurol Neurosurg Psychiatry 2007;78:929–935.
- 19.
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.
- 20.
Dusing SC, Thorpe DE. A normative sample of temporal and spatial gait characteristics in children using the GAITRite electronic walkway. Gait Posture 2007;25:135–139.
- 21.
Krishnamurthy M, Verghese J. Gait characteristics in nondisabled community-residing nonagenarians. Arch Phys Med Rehabil 2006;87:541–545.
- 22.
Boyd BS, Puttlitz C, Noble-Haeusslein LJ, et al.Deviations in gait pattern in experimental models of hindlimb paresis shown by a novel pressure mapping system. J Neurosci Res 2007;85:2272–2283.
