Walking is the most common form of locomotion and comprises a series of complex movements that demand control of the musculoskeletal system through information processing, pattern generation, and reflex control.1 Healthy, able-bodied individuals demonstrate walking patterns that are typically symmetrical with respect to kinematic and kinetic variables.2 Compared with the bodies of these individuals, the body of a person with a unilateral lower-limb amputation is inherently asymmetric. Despite the loss of a limb, individuals can learn to walk comfortably with a prosthesis. However, the time required to walk independently with a prosthesis has not been well documented.
The published literature has demonstrated that amputees exhibit asymmetrical temporal-distance (T-D) walking patterns and walk at a slower self-selected walking speed than do able-bodied individuals.3–5 Lower-limb amputees also take longer steps on the prosthetic side compared with the intact side and spend more time in stance on the intact limb and in swing on the prosthetic limb. Shorter prosthetic stance has been attributed to amputees trying to protect their residual limb by spending less time supporting their body weight on the prosthesis, which serves as a compensatory action for increasing gait stability.5–7
For individuals who have experienced a lower-limb amputation, skeletal links, joints and muscles of the lower limbs have been lost and replaced with prosthetic components. Despite the altered physical constraints, amputees can learn to walk proficiently and do so by showing kinematic profiles similar to those of able-bodied individuals. Sanderson and Martin8 noted that amputees walked with patterns of motion that were not easily distinguishable from those of non-amputees. Specifically, they found that ankle, knee, and hip angular position and velocity patterns exhibited only minor differences between limbs at preferred (1.2 m · s−1) and fast (1.6 m · s−1) walking speeds. For example, the prosthetic limb was maintained more vertically aligned throughout the stance phase and prevented the knee from flexing. The authors concluded that although amputees and non amputees walked with similar lower-limb kinematic profiles, they achieved the movements by employing different kinetic profiles.
Amputee patients gain proficiency using their prostheses at different rates. Therefore, acquiring information on how specific groups respond to gait training and the time taken to adapt to walking with a prosthesis is important for therapists.9 The literature does not adequately discuss the nature of change or the time course of gait adaptations an individual makes after a lower-limb amputation. A systematic review of the literature has found one group of authors who designed and studied the use of a lower-limb prosthetic simulator (PS) with able-bodied participants.10 The concept of a simulator to examine an individual's ability to learn to use a prosthesis has been studied previously with able-bodied subjects using an upper-limb prosthetic simulator.11 The lower-limb PS described by Lemaire et al.10 was a custom-made prosthetic device that provided a prosthetic-like experience by allowing non-amputees to walk like a person with a unilateral transfemoral amputation. Because of the complex issues associated with tracking individuals who have experienced a lower-limb amputation from the moment they enter the rehabilitation setting until they leave as competent walkers, little is known about the time required to adapt to walking with a prosthetic device. To counter this difficulty, a unilateral lower-limb PS was designed and created in-house for this study to allow able-bodied individuals to walk in a unique situation. Thus, the nature and development of changes in gait variables could be monitored during this novel walking task. The purpose of the study was to determine the changes in T-D and selected kinematic variables of a novel walking task with a specially designed PS during 30 minutes of continuous walking. The experiment provided a challenge for otherwise healthy, able-bodied individuals to develop new locomotor adaptations when lower-limb mechanics were altered. This could be accomplished from the moment the participant was fitted with the PS and after practice as they learned to ambulate more comfortably.
The dependent variables selected in this study were considered important in characterizing altered walking patterns in individuals with modified lower-limb mechanics. T-D gait variables included walking speed, step length, and relative stance duration on both the intact and prosthetic limbs. Kinematic variables included vertical orientation of the prosthesis and the shank of the intact limb and anterior-posterior trunk lean at instances of prosthetic contact (PC) and intact foot contact (IFC). It was hypothesized that, with time, walking speed would increase as a function of increasing step length for both limbs and that stance duration would decrease. It was also hypothesized that the prosthetic and intact shank segments initially would be more vertically oriented at PC and IFC, respectively, and trunk angle initially would show greater forward lean at PC and IFC.
Ten healthy, able-bodied individuals (4 males, 6 females; mean ± SD; age, 22.6 ± 2.3 years; height, 173.0 ± 6.4 cm; mass, 66.3 ± 6.8 kg) volunteered for this study. Inclusion criteria were that participants had no known neuromuscular or balance disorders, recent musculoskeletal injury to either lower limb, or any previous experience walking with a PS. Once fitted with the PS, they must have been able to walk unassisted. Individuals within a height range of 165 to 185 cm were selected for the study because of the dimensions of the PS. All participants gave informed consent to take part in this study in accordance with the University's Clinical Research Ethics Board guidelines.
The PS consisted of a plastic cuff into which the individual placed his or her dominant leg (Figure 1A). Dominant leg was determined by asking the participant which leg he/she would use to kick a soccer ball, and for all participants this was their right leg. The knee was flexed at 90° and securely maintained in this position with Velcro straps running across the posterior side of the thigh and shank. A rigid telescoping aluminium tube provided the means of support at the lower extremity and was easily adjusted according to the individual's lower-limb length. A slip-resistant rubber cap was placed at the end of the aluminium tube.
Each participant took part in two testing visits 1 week apart. For the first visit, whole-body kinematic data were recorded while participants walked first without the PS and then with the PS fitted to their leg. Reflective markers were placed on specific anatomical locations according to the Helen- Hayes full body marker set. The lab was equipped with a high- speed, eight-camera system (Motion Analysis Corporation, Santa Rosa, CA) and was used to capture three-dimensional kinematic data. The ExpertVision (EVaRT) software program (Motion Analysis Corporation) was used to collect data during the visit at a sampling rate of 60 Hz.
Participants first completed several control trials, walking barefoot at self-selected speeds along a 10-m walkway while T-D and kinematic data were collected. This was to establish preferred walking speed and kinematic profiles for able- bodied individuals. The PS was then fitted to the right leg; the participants were asked to insert the flexed leg into the plastic cuff while body weight was supported on the left leg during standing. The Velcro straps were adjusted to secure the leg in the PS. Participants could hold on to a chair to maintain their balance while the PS was being fitted. The telescoping tube of the PS was adjusted 2 cm shorter than the intact right leg. A leg length discrepancy of as much as 2 cm usually is regarded as normal12 and does not require corrective procedures, whereas discrepancies greater than 2 cm may lead to altered gait patterns.13 In the study, shortening of the prosthesis length by 2 cm was done to allow participants sufficient ground clearance on the prosthetic side. It has been reported that during able-bodied gait, 13 mm of toe clearance is sufficient to avoid a trip.12 Adequate foot clearance typically is accomplished by shortening of the limb during the swing phase through knee flexion of approximately 60°.12 In the current study, shortening of the prosthesis by 2 cm was deemed appropriate to reduce the likelihood participants would stumble in the absence of knee flexion on the prosthetic side. Once the PS was properly secured onto the participant's leg, load was borne through the knee during standing and walking. Because of the absence of a prosthetic foot, a three-coordinate marker set with markers for the heel, toe, and lateral aspect of the ankle was fastened to the end of the aluminium tube. These markers were used by the OrthoTrak software system (Motion Analysis Corporation) to indicate instances of prosthetic contact and prosthetic off, which define the stance and swing phases of the gait cycle.
Once fitted with the PS, participants were asked to walk continuously and unassisted along a 30-m loop in the gait laboratory for a total of 30 minutes. Pilot data revealed that participants increased their walking speed linearly for 20 minutes, at which point walking speed leveled. Therefore, for the current study, participants were asked to walk for a total of 30 minutes to ensure they achieved their maximal walking speed on each test visit. They were not given any instructions on how to walk with the PS. T-D and kinematic data were recorded for three consecutive strides, every time the participant completed a loop and came into the cameras' field of view. After 30 minutes, the PS was removed, reflective markers were placed on the anatomical landmarks of the right lower leg, and the participants were requested to walk though the measurement area. Walking speed was measured during the first ten strides immediately after the PS was removed to determine whether a speed after-effect could be observed after walking with the prosthetic device.
For the second test visit, only walking speed was measured while participants were instructed to walk continuously along a 30-m loop for another 30 minutes. Walking speed was determined with the use of photocells located 2 m apart. After completing another control trial, the PS was fitted onto the right leg, and the telescoping tube was adjusted to the same lower-leg length as during the first visit. Walking speed was monitored every time the participant completed a loop.
Three-dimensional coordinate data from each completed lap were interpolated using a cubic spline algorithm and filtered using a fourth order Butterworth filter with a cutoff frequency of 6 Hz. The processed data were then exported into OrthoTrak 5.0 Gait Analysis Software (Motion Analysis Corporation), where gait events were determined. Instances of intact foot/prosthetic contact and intact foot/prosthetic off were identified from the motion analysis data. Before the three-dimensional coordinate data were exported for analysis, all gait events were individually verified in OrthoTrak with reference to motion data. Any trials in which the participant tripped were disregarded. A trip occurred when participants did not successfully clear the ground with the end of the PS during the swing phase, consequently causing them to hop on their intact leg to regain balance.
All kinematic data were calculated based on coordinate data from OrthoTrak. Projected two-dimensional segment angles of the lower limbs were determined in the sagittal plane (Figure 1B). On the intact side, the shank angle was defined as the vertical orientation of the shank segment with respect to the vertical axis through the ankle joint center. The prosthesis was defined as the segment between the hip and ankle joint centers. Prosthetic angle was defined as the vertical orientation of the prosthesis with respect to the vertical axis through the virtual ankle-joint center. The trunk segment was calculated as the segment from the shoulder center to the pelvic center. The trunk angle was defined as the vertical orientation of the trunk segment with respect to the vertical axis through the pelvic center. Values for the shank and prosthetic angle were determined at IFC and PC, respectively. The trunk angle was determined at both IFC and PC.
Visual inspection of mean group walking speed after 30 minutes of continuous walking with the PS revealed that the greatest changes occurred early during the first several laps. Therefore, to analyze the data where the greatest change occurred, one time constant for walking speed was calculated. One time constant corresponded to 63% of final walking speed, which occurred after the participants completed four loops. Walking speed was considered to be an overall indicator of the participants' adaptation to walking with the PS. The recorded values of the dependent variables for a total of 12 strides (three strides over four loops) were used for statistical analysis.
A general repeated-measures analysis of variance (ANOVA) design was conducted to compare differences in dependent variables over one time constant. A 2 × 12 (limb × stride) repeated-measures ANOVA was used for step length, stance duration, trunk, and lower-limb angles over one time constant. A 1 × 12 repeated measures ANOVA was used for walking speed. Initial and final walking speed on the two visits were compared using a 2 × 2 (time × visit) design. To determine if there were any after-effects of walking speed before and after walking with the PS, a 3 × 2 (time × visit) design was used. Huynh-Feldt adjustments were made across all statistical tests using SPSS version. 13.0 (SPSS Inc., Chicago, IL). Results were considered statistically significant at p ≤ 0.05.
All participants were immediately able to walk unassisted with the PS and did so continuously for 30 minutes without falling. On the first visit, initial group mean walking speed was 0.24 ± 0.12 m · s−1, which increased to 63% of final walking speed (0.47 ± 0.17 m · s−1) within the first four loops (after 120 m). Walking speed with the PS significantly increased over one time constant (F(8.69,78.27) = 11.73, p < 0.000). Final walking speed with the PS was 0.70 ± 0.21 m · s−1, just more than half of mean walking speed in the control condition (Figure 2). One week later, walking speed increased from 0.58 ± 0.19 m · s−1 to 0.77 ± 0.20 m · s−1. Initial walking speed during the second visit represented 83% of final walking speed from the first visit. A significant interaction for time × visit showed that the increase in walking speed was significantly greater during the first visit (F(1,9) = 50.901, p < 0.000).
Step length on the prosthetic side was significantly greater than on the intact side (F(1, 9) = 44.95, p < 0.000). There was also a significant main effect of step length during one time constant (F(6.31,56.78) = 10.77, p < 0.000) (Figure 3A). Step length on the intact side initially was negative, indicating that the intact foot did not step past the PS. Step length on the prosthetic side remained relatively constant throughout the walking task, whereas intact step length increased. Final intact step length with the PS was 73% of step length in the control condition. Furthermore, six of ten participants showed a stepping response of taking two steps with the intact limb for one step with the prosthetic limb. These participants had to compensate for a long prosthetic step length by taking two steps with their intact foot to maintain balance when walking with the PS. Because of this stepping response, step frequency initially was different between the two limbs, but this discrepancy disappeared during the first loop. Thereafter, step frequency was only 75% of the control condition.
A significant main effect was measured for relative stance duration over one time constant (F(8.67,77.99) = 16.33, p < 0.000). Intact stance duration initially was 86% of the gait cycle, compared with 68% on the prosthetic limb, but these values decreased for both limbs. The intact limb spent significantly longer time in stance than did the prosthetic limb (F(1,9) = 370.18, p < 0.000) (Figure 3B).
Overall, there was a significant main effect for the vertical orientation of the lower limbs (F(4.92,44.24) = 8.99, p < 0.000), a significant main effect for limb (F(1,9) = 35.05, p < 0.000), and a significant interaction (F(5.94,31.33) = 5.59, p < 0.000) over one time constant. Closer inspection of Figure 4A shows that the prosthetic angle remained relatively constant while participants modified the vertical orientation of the shank, from an initial negative angle to a positive value. The initial negative shank angle indicated that the shank was oriented in front of the vertical and the knee was flexed at IFC.
Contrary to our hypothesis, there was no overall significant difference of vertical orientation of trunk angle at IFC or PC over one time constant. However, trunk angle at IFC showed significantly more forward trunk lean (represented by the negative values) than did trunk angle at PC (F(1,9) = 12.41, p < 0.006). Although not statistically different, a trend could be seen between the intact and prosthetic limbs (F(7.09,63.86) = 2.06, p < 0.060). Trunk angle at PC remained relatively constant and closer to the vertical, while there was greater forward lean at IFC (Figure 4B).
Participants' mean self-selected walking speed with two intact limbs was 1.30 ± 0.09 m · s−1. However, a significant, short-term speed after-effect was observed during the first five and ten strides immediately upon removal of the PS on the first visit (F(1.88,16.95) = 9.621, p < 0.002). Walking speed dropped to 1.13 ± 0.16 m · s−1 for the first five strides and increased to only 1.15 ± 0.09 m · s−1 during the first ten strides after removing the PS during the first visit (Figure 2).
The goal of the current study was to investigate the process of adaptation to a novel walking task by altering lower-limb mechanics with a prosthetic simulator. The results from this study show that participants were able to adjust successfully and quickly to the new set of imposed locomotor constraints by modifying T-D and kinematic variables. Walking speed during this novel walking task was considered an overall descriptor of performance and adaptation. The first time the participants walked with the prosthetic device, the final walking speed achieved was just more than half of that in the control condition. However, participants reached 63% of the final walking speed with the PS after completing only four loops of 30 m. The largest changes in gait occurred rapidly and were retained throughout a 1-week period. This suggested that learning to walk with modified lower-limb mechanics took place during the first visit.
Locomotor symmetry and walking velocity are often used to describe prosthetic gait in clinical settings.6 Walking speed is the product of step length and frequency and can be attained by different combinations of these two variables. However, as velocity increases, the number of possible combinations is reduced given the biomechanical constraints of the musculoskeletal system to further increase step length and/or frequency during walking.6 The results of this study showed that participants immediately adapted to the PS by taking very short steps with the intact foot. In six of ten participants, the intact foot was placed behind the prosthesis, resulting in a negative step length value. Step length can be negative when one foot does not pass the other foot.14 Furthermore, these same individuals showed a stepping response by taking two shorter steps with the intact foot for one long prosthetic step. The greater prosthetic step length was attributed to a long prosthetic swing phase and the effects of altered distribution of lower-limb mass on the prosthetic side. The design of the PS used in this study maintained the knee in 90° flexion and thereby decreased the moment of inertia of the prosthetic lower limb relative to the transverse axis through the hip joint. Consequently, the angular velocity of the prosthetic limb would have been greater than the intact limb during swing resulting in a longer prosthetic step length. This compensatory stepping response disappeared during the first loop as participants adapted to the altered lower-limb mechanics by increasing intact step length. Given that step frequency and prosthetic step length remained relatively constant (Figure 3A), increases in walking speed were achieved by increasing intact step length. This was contrary to our hypothesis that a faster walking speed would be attributed to step length increases for both limbs. It appeared that participants selected a prosthetic step length and modulated their intact limb to their prosthetic side. Subsequent increases in speed would have required participants to increase step length and/or frequency beyond a level they had adopted as comfortable or preferred.
Body support times during walking with the PS were consistent with data reported in the literature for lower-limb amputees.4,6–8 A significantly greater amount of time was spent in stance on the intact limb (p < 0.000) which was likely attributable to the perceived instability associated with the prosthesis. It has been suggested that prosthetic stance time is shortened to avoid pain and minimize the need for controlling the mechanical interaction between the prosthetic limb and the environment.6 This protective component has been reported previously in the literature.6–8 Time in stance on the intact limb (constantly more than 75% of the stride cycle) was considerably higher than values reported for lower-limb amputees4,5,8 because of the novelty of the walking task with the PS and the slower walking speed. Data reported in the literature have often been recorded for proficient amputee walkers. Whether the amount of time spent in stance on the sound limb when learning to walk with a prosthesis was as high for these individuals as the values observed in our study remains unclear.
The shank and prosthetic segments were more vertically oriented at instances of IFC and PC, respectively, during the walking task (Figure 4A). The shank initially was positioned in front of the vertical (negative absolute shank angle), meaning that participants made IFC with a flexed knee during the first three steps. Positioning the shank in front of the vertical at IFC assisted in shifting the whole body center of mass (CoM) anteriorly and closer to the base of support. Greater forward trunk lean (negative absolute trunk angle) also helped to maintain the CoM above the intact supporting limb at IFC. This finding was consistent with the literature reporting that lower-limb amputees flex their trunk forward early in the stance phase to keep the prosthetic knee in a more extended position and maintain the body's CoM above the supporting limb.6 Considering a significantly greater amount of time was spent in single support on the intact limb (p < 0.000), this was an important balance strategy when walking with the PS. Conversely, participants did not modulate the vertical orientation of the prosthesis at PC. Given the reduced degrees of freedom of the prosthetic segment without an artificial ankle and/or knee joint, participants modulated their intact limb, rather than their prosthetic limb, to make changes to their gait pattern with the PS.
Previous studies have reported adaptational after-effects upon removal of a perturbation, providing strong evidence that learning had occurred.15,16 The short-term speed after-effect observed during the first ten strides after removing the PS reinforced the notion that the participants had adapted to walking with the device. Slower walking speed was an expression of the modified motor behavior of walking with the PS. Retention of walking speed 1 week later provided additional support that learning had occurred after 30 minutes of walking in a novel manner during the first visit.
The results of the current study have demonstrated that individuals quickly learn to walk with a prosthetic device. Walking speed was considered an overall indicator of walking performance with the PS. A rapid increase in speed demonstrated that individuals were able to adjust their habitual gait pattern successfully according to the imposed constraints. They did so by walking considerably more slowly and altering step and body support characteristics and the orientation of body segments. The primary adaptive strategies occurred by modulating the intact limb and maintaining a consistency of motion on the prosthetic limb. The intact limb offered more flexibility of motion, and small adjustments, needed to maintain balance, could be accomplished more easily than with the prosthetic limb.
The purpose of this study was not to make direct comparisons with lower-limb amputee gait, but rather to investigate the step-by-step gait adaptations that occurred when lower-limb mechanics were altered among a group of able-bodied individuals. We have shown that young, healthy individuals can relearn a new movement pattern relatively quickly, even under biomechanical constraints, and that adaptive strategies were developed early in the task. However, individuals who have actually undergone the amputation of a lower limb and participating in early gait training must deal with many concomitant factors that affect their rehabilitation treatment. These may include poor general health and mobility, depression, edema of the residual limb, inadequate healing of skin wounds, and phantom and/or residual limb pain. These factors may delay or complicate prosthetic fitting and gait training. Therefore, the results of the current study may have more important implications for the younger, traumatic lower-limb amputee.
A limitation to the current study was that the majority of individuals with a major lower-limb amputation are older individuals who have experienced the loss of a limb for vascular reasons and have comorbidities, such as diabetes, peripheral vascular disease, and neuropathy. These conditions likely affect their physical capacity and ability to maintain dynamic balance. Thus, future research should examine how individuals from different age groups and with varying levels of physical ability adapt to altered lower-limb mechanics imposed by the PS.
There are walking aids used during early gait training with transtibial amputees that are similar to the PS designed for this study. In the United Kingdom, physiotherapists may use the pneumatic post-amputation mobility (PPAM) early walking aid with their transtibial patients.17 The PPAM aid is a generic type of prosthesis in which a pneumatic sleeve is inflated around the stump and positioned within a rigid, metal frame. The constraint with this type of prosthetic device is that patients are unable to flex and extend their leg during walking, so the device does not simulate a natural gait pattern. Therefore, an additional limitation to the current study involved the design of the PS, which inherently contributed to the asymmetrical walking pattern of the participants by not allowing them to flex their knee on the prosthetic limb. The prosthetic-like experience with the PS may be improved by designing a prosthetic device with a prosthetic foot and mechanical ankle and knee joints that resembles a real prosthesis for lower-limb amputees and facilitates a more natural gait pattern.
The novel approach of a prosthetic simulator is to monitor individuals as they must learn to modify their gait pattern. The results of the current study could have implications for rehabilitation practice and design of early walking aids. The findings also strengthen the importance of early prosthetic fitting and prosthetic gait training because individuals quickly learned to walk with the PS.
The authors thank Alec Black, Director of Shriner's Gait at Sunny Hill Hospital, Vancouver, for his assistance with data collection.
1. McFadyen BJ, Bélanger M. Neuromechanical concepts for the assessment of the control of human gait. In: Allard, P, Cappozzo A, Lundberg A, Vaughan C, eds. Three-dimensional Analysis of Human Locomotion.
Chichester, NY: John Wiley & Sons Ltd;1997:49–66.
2. Sadeghi H, Allard P, Prince F, Labelle H. Symmetry and limb dominance in able-bodied gait: a review. Gait Posture
3. Hale SA. The effect of walking speed on the joint displacement patterns and forces and moments acting on the above-knee amputee prosthetic leg. J Prosthet Orthot
4. Isakov E, Burger H, Krajnik J, et al. Influence of speed on gait parameters and on symmetry in trans-tibial amputees. Prosthet Orthot Int
5. Jaegers SM, Arendzen JH, de Jongh HJ. Prosthetic gait of unilateral transfemoral amputees: a kinematic study. Arch Phys Med Rehabil
6. Donker SF, Beek PJ. Interlimb coordination in prosthetic walking: effects of asymmetry and walking velocity. Acta Psychol
7. Nolan L, Wit A, Dudzinski K, et al. Adjustments in gait symmetry with walking speed in trans-femoral and trans-tibial amputees. Gait Posture
8. Sanderson DJ, Martin PE. Lower extremity kinematic and kinetic adaptations in unilateral below-knee amputees during walking. Gait Posture
9. Baker PA, Hewison SR. Gait recovery pattern of unilateral lower-limb amputees during rehabilitation. Prosthet Orthot Int
10. Lemaire ED, Nielen D, Paquin MA. Gait evaluation of a transfemoral prosthetic simulator
. Arch Phys Med Rehabil
11. Weeks DL, Wallace SA, Anderson DI. Training with an upper-limb prosthetic simulator
to enhance transfer to skill across limbs. Arch Phys Med Rehabil
12. Kirtley C. Clinical Gait Analysis: Theory and Practice
London: Elsevier Churchill Livingstone; 2006.
13. Perttunen JR, Anttila E, Södergård J, et al. Gait asymmetry in patients with limb length discrepancy. Scand J Med Sci Sports
14. Lamoureux LW. Kinematic measurements in the study of human walking. Bull Prosthet Res
15. Jensen L, Prokop T, Dietz V. Adaptational effects during human split-belt walking: influence of afferent input. Exp Brain Res
16. Lam T, Wolstenholme C, Yang JF. How do infants adapt to loading of the limb during the swing phase of stepping? J Neurophysiol
17. Scott H, Condie ME, Treweek SP, Sockalingam S. An evaluation of the Amputee Mobility Aid (AMA) early walking aid. Prosthet Orthot Int