Many researchers have reported that unilateral below-knee amputees (BKA) walk asymmetrically and differently from able-bodied people. 1–5 The reasons given for asymmetrical walking vary. It is generally believed that socket fit, prosthetic alignment, and prosthetic components (including prosthetic parts’ weight and design) can all influence the gait of amputees. Others predicted that degenerative changes in the lumbar spine and knees would occur due to the asymmetrical walking that overloads the musculoskeletal system. 6 Hurley et al. 7 investigated the role of the contralateral limb in amputee gait by determining lower-limb joint reaction forces and symmetry of motion in amputee and nonamputee populations. The subjects in this study demonstrated more asymmetry than nonamputees. It is proposed that this asymmetrical movement may be attributed to the inherent variability of the actions of the prosthetic lower limb. The results suggested that although amputees demonstrate an asymmetrical gait pattern, forces acting across the joints of the contralateral limb are not significantly higher than those of nonamputees. Furthermore, providing the adult amputee has a good prosthetic fit, there is no reason to expect the long-term wearer to develop premature degenerative arthritis.
In a general review, the studies in prosthetic gait show no significant differences in gait between right- or left-sided amputation in either above or below-knee amputation. However, they have shown that the incidence of amputation is demonstrably greater in the male population. 8–10
The purpose of this study was to identify kinetic and kinematic characteristics of the gait of BKA and compare the results with the findings of other studies on BKA and normal population. Furthermore, this paper indicates the importance and feasibility of the measurement of kinetic factors in order to evaluate the gait of amputees, which have been neglected by many researchers. To date, the majority of studies have not addressed the kinetic variations of the amputee gait.
REVIEW OF THE LITERATURE
TEMPORAL AND DISTANCE FACTORS
It is frequently reported that prosthetic step length of BKA is slightly longer compared with the sound limb. 11,12 The stance time on the prosthetic side is reported to be less than the unaffected limb, and, therefore, the swing phase on the prosthetic side lasts longer. 9,11,12 Breakey 1 reported that single limb support time was 37% of the gait cycle for the affected limb and 43% for the unaffected limb. Self-selected walking velocity of amputees is lower than mean normal values. 13–15 Tibarewala and Ganguli 16 showed a decreased cadence in the gait of below and above knee amputees compared with normal subjects. Kegel et al. 17 indicated that the self-selected walking velocity of amputees improve (13%) by muscle strengthening through isometric exercises. Kegel et al. 17 also concluded that amputees have lower than normal walking velocity even after an isometric exercise program.
Results of amputees biomechanics gait studies reveal subtle departures from the gait pattern of the able-bodied. The knee flexion of the amputee is less than the mean normal amount during early stance. 1 This occurs because the prosthetic foot does not produce the controlled plantarflexion obtained naturally by eccentric contraction of the dorsiflexors. 12 During late stance, knee flexion is also less than mean normal values. Usually knee motion coordinates with foot motion. Unlike the anatomical foot, which plantarflexes at toe-off, the prosthetic ankle cannot move when weight has been transferred to the toe section. If the amputee were to allow the knee to flex to the mean normal extent, the torso would lower excessively, producing an inefficient, abnormal gait. 1
Smidt 12 reported that all the lower limb joint moments among amputees on the affected side are different from mean normal. Lewallen et al. 18 found that the nonamputated side generates a greater dorsiflexion moment than mean normal. Robinson et al. 11 reported that the steps taken by the involved (prosthetic) side were longer and accomplished in less time across the subjects. Therefore, they suggested that the acceleration of the involved lower extremity was greater than the acceleration of the uninvolved lower extremity during swing phase of the gait cycle. Winter and Sienko’s 19 study shows a somewhat smoother knee moment curve for BKA than normal people but within the normal range.
Normal subjects rapidly reduce the dorsiflexor moment at the ankle after initial contact. In this way, they reach a foot-flat position early in stance and stabilize the foot by using both dorsi- and plantar-flexors. However, BKA have a rigid ankle that generates an internal dorsiflexor moment from heel contact until foot flat and later than normal foot flat time. 20 Winter 20 also reported a rise in the ankle internal plantarflexion moment during the foot flat of BKA when the center of pressure is moving forward (between 20% and 40% of the gait cycle). This is about 30% to 40% more than that of able-bodied people and is believed to be due to the need for a rapid plantarflexion. The results of another study by Winter 20 on one BKA showed an extended dorsiflexor pattern in the ankle moment lasting 25% of the stride and followed by a plantarflexor pattern that reached a peak of about 2/3 of mean normal at about 45% of stride period. The knee moment was zero for the first half of the stance and in the normal zone for the balance of the stride period. The hip moment was lower than mean normal (performing an extensor pattern) during 90% of stance phase and mean normal for the remainder.
WORK AND POWER
The more important phases of power generation (+ve) and absorption (−ve) at ankle, knee, and hip joints during normal gait have been identified by Winter 20 as follows. A1 is the first major power absorption that occurs between 5% and 40% of the stride as the result of a lengthening contraction of the ankle plantarflexors (dorsiflexion occurring) when the leg rotates over the foot under eccentric (lengthening) control of plantarflexors. A2 is the only major power generation of the ankle, which occurs between 40% and 60% of the stride as the result of rapid shortening contraction of the ankle plantarflexors (plantarflexion occurring) before toe-off. From heel strike to 15% of the stride, some power absorption also occurs at the knee joint (K1) due to a lengthening contraction of the knee extensors (knee flexion occurring). The only major positive burst of power at the knee (K2) occurs due to a shortening contraction of knee extensors (knee extension occurring) between approximately 15% and 25% of the stride. Between 40% of the stride to toe-off a power absorption burst occurs in the power profiles that represents K3, which is caused by a small extensor moment (small lengthening contraction of the quadriceps muscle), coupled with very rapid knee flexion. In the later half of swing phase, the hamstring muscles activate to absorb the energy from the swinging leg and foot (K4). At the hip joint, the first power generation burst (H1) occurs in the first half of the stance as a result of s shortening contraction of the hip extensors. This is followed by a power absorption burst (H2) in the later half of the stance, the result of hip flexor activity, which absorbs energy of the decelerating thigh before it reaches the maximum hip extension. Major power generation at the hip occurs in late stance phase and early swing phase (H3) as a “pull-off burst” by the hip flexors which assist in propelling the swinging limb forward. To put the important parts of the power generation picture together for both limbs over the gait cycle, there are two major generation phases of power generation in the gait cycle. One occurs near the time of push-off/pull-off of the unaffected limb (A2 unaffected, H3 unaffected, K2 affected) and the other occurs near the time of push-off/pull-off of the affected limb (A2 affected, H3 affected, K2 unaffected (for more detail, see ref. 20, pp. 35–50).
With approval from the Ethics Committee at Queen’s University, five male unilateral traumatic BKA aged 32 to 77 with tibial length of 11 cm (±5 cm) in the amputated side were recruited. Table 1 indicates demographic data of each individual participant of this study. The participants had all used their prosthesis for at least a year and had no lesion on the stump or any problem on their knee joint at the time of the study. They were also walking with no need of assistive devices and their good walking performances were confirmed by a prosthetist. They were also all using a modular prosthesis with steel components and stationary attachment flexible endoskeletal (SAFE) foot.
Peak motion analysis (2D) system was used with an AMTI force plate to yield spatial, temporal, kinematic, and kinetic data. Light reflective passive spherical markers with a 3.5-cm diameter as suggested by Zheng 21 were used and the camera was set to take 60 pictures per second. Force plate data were collected at 600-Hz frequency. The video camera was synchronized with the data acquisition from the force plate via an externally triggered pulse.
Reflective markers were taped on the body on the base of the neck at C7-T1, camera side fifth metatarsal joint, the ankle lateral malleolus, the lateral epicondyle of the femur, over the greater trochanter at the hip-joint level. 22 The last of these, the hip marker, was placed mid-way across the horizontal width of the greater trochanter at the vertical level of the hip joint with the subject in standing and the limbs in a neutral position. The level of the hip joint was determined by locating the joint under the middle third of the inguinal line. This placement was performed by one person for all subjects to reduce intertester variability. Soft tissue movement, although undoubtedly present, was not excessive due to the relatively slow speeds. Although the authors recognize that more sophisticated systems permit more precise location, this method is the only one provided by the technology used and has, despite its acknowledged limitations, been responsible for much of the progress in understanding basic features of normal and pathological gait. To improve the validity of data, a circle of low-reflecting black fabric was placed around each marker. Three acceptable trials were collected for each side of the body.
Marker locations on the film were digitized and saved in files using the Peak motion analysis system. The raw data were filtered at 5-Hz frequency, converted to real-world coordinates, and processed to obtain gait variables based on time. To calculate kinetic and kinematics parameters of the gait, marker data files and force plate data files were synchronized and processed using “Q-KIN-2D” custom software, based on an inverse dynamic approach using a four-segment link segment model. The software calculated stride time and length, velocity, relative angles and foot segment angle, moments, power, and muscle work across the hip, knee, and ankle joints. The anthropometric convention and calculation procedures followed those of Winter et al. 22 However, for the prosthetic leg, calculations of anthropometric parameters were performed for each participant separately. Relative angles or joint angles are defined as the angles between two body parts, whereas segment angle is defined as the angle between a body segment and the horizontal. In other words, segment angles are the absolute angle of each of the segments in space. Moments are the net result of all bony, muscular, ligament, and friction forces acting to alter the angular rotation of the joint. Because joints do not usually reach positions of extreme flexion or extension, and friction forces are minimal, moment values usually represent the sum of muscle activity. Positive values represent an extensor moment, whereas negative values represent a flexion moment of the joint. 20
Power across a joint (j) is calculated as the product of the joint moment of force and joint angular velocity:EQUATION
where: P j is the joint power (Watts), M j is the joint moment of force (N·m), and ω j is the joint angular velocity (r/s).
The positive work and negative work (in joules per kilogram of body mass) performed by the muscles across each joint for each stride were determined by calculating the area under the power output curve:EQUATIONEQUATIONEQUATION
Wm = muscle work for i fields of joint, P = instantaneous net power, i = field, n = number of fields in event of interest, and t = time (seconds).
Positive power indicates that a muscle (or a group of muscles) was generating mechanical work, whereas negative power indicates that a muscle (or a group of muscles) were absorbing energy. The positive or concentric work done at each lower-extremity joint was calculated as the area under the curve of the corresponding positive power phase. Similarly, the negative or eccentric work done at each lower-extremity joint was calculated as the area under the corresponding negative power phase. The total positive work during a stride was the absolute sum of positive work across all joints; the total negative work was the absolute sum of the negative work across all joints. The total positive and negative work during a stride was the absolute sum of the total positive and total negative work.
The more important phases of power generation (+ve) and absorption (−ve) at ankle, knee, and hip joints during normal gait have been identified by Winter 20 as follows. A1 is the first major power absorption that occurs between 5% and 40% of the stride as the result of a lengthening contraction of the ankle plantarflexors (dorsiflexion occurring) when the leg rotates over the foot under eccentric (lengthening) control of plantarflexors. A2 is the only major power generation of the ankle, which occurs between 40% and 60% of the stride as the result of rapid shortening contraction of the ankle plantarflexors (plantarflexion occurring) before toe-off. From heel strike to 15% of the stride, some power absorption also occurs at the knee joint (K1) due to a lengthening contraction of the knee extensors (knee flexion occurring). The only major positive burst of power at the knee (K2) occurs due to a shortening contraction of knee extensors (knee extension occurring) between approximately 15% and 25% of the stride. Between 40% of the stride to toe-off a power absorption burst occurs in the power profiles that represents K3, which is caused by a small extensor moment (small lengthening contraction of the quadriceps muscle), coupled with very rapid knee flexion. In the later half of swing phase the hamstrings muscles activate to absorb the energy from the swinging leg and foot (K4). At the hip joint, the first power generation burst (H1) occurs in the first half of the stance as a result of shortening contraction of the hip extensors. This is followed by a power absorption burst (H2) in the later half of the stance, the result of hip flexor activity, which absorbs energy of the decelerating thigh before it reaches the maximum hip extension. Major power generation at the hip occurs in late stance phase and early swing phase (H3) as a “pull-off burst” by the hip flexors, which assist in propelling the swinging limb forward. To put the important parts of the power generation picture together for both limbs over the gait cycle, there are two major generation phases of power generation in the gait cycle. One occurs near the time of push-off/pull-off of the unaffected limb (A2 unaffected, H3 unaffected, K2 affected) and the other occurs near the time of push-off/pull-off of the affected limb (A2 affected, H3 affected, K2 unaffected (for more detail, see ref. 20, pp. 35–50).
RESULTS AND DISCUSSION
TEMPORAL AND DISTANCE FACTORS
The average of selected values of gait parameters shows that the stride length, stride time, and velocity were similar to those of normal populations. The percent time of double support and stance were also similar, though the stance phase was slightly longer on the unaffected side than the affected side. However, the difference BKA stance phase from normal was not statistically important. The findings of this study are presented and compared with literature sources in Table 2
The results of this study showed that amputee gait was asymmetrical and different from that of able-bodied individuals, confirming the general findings of Engsberg et al., 4 Culham et al., 3 and Prince et al. 5 The average stride length of subjects was 1.33 m (when using steel components), which is shorter than the mean normal values (1.56 m) reported by Murray et al. 24 or mean normal values reported by Winter 20 for natural-speed walkers (1.51 m/s). However, it was similar to a study of BKA by Robinson et al. 11 (1.32 m), shorter than that reported by Waters et al. 13 (1.44 m), but longer than the values reported by Winter and Sienko 19 (1.27 m).
The normal speed of walking has been reported to be 1.37 m/s for men 23 and 1.325 m/s. 20 Similar to findings of Kegel et al., 17 the results of this study confirmed that the self selected speed of walking among amputees is lower than mean normal (1.11 m/s), and close to the finding of Robinson et al. 11 for BKA (1.07 m/s). The walking velocity of 1.04 m/s for BKA that was reported by Colborne et al. 25 was lower than the findings of this study. Similar to the findings of Breakey, 1 amputees demonstrated a longer stance phase on the normal limb and a shorter one on the amputated limb. This is common among pathological gait patterns. A cause of this may be a lack of trust of the affected side for weight bearing. In other words, amputees try to transfer weight to the unaffected side, thus making the stance phase of the affected side shorter. Table 2 indicates the results of this study and compares it with some others.
JOINT RELATIVE ANGLES
The results of this study indicate that the hip relative angle ranges were smaller for the affected side. However, the affected knee relative angle range was considerably larger than that of the unaffected side. When comparing the results of the affected side with studies of Winter and Sienko 19 and Colborne et al., 25 no substantial differences are noticed. Figure 1 indicates the typical hip, knee, and ankle relative angles of participants of this study for both affected and unaffected side.
Greater than mean normal positions of maximum hip flexion (by about 10°) were seen on the profiles of all subjects on the affected side during the gait cycle. Although this also occurred on the unaffected side, the differences from mean normal decreases during late stance and early swing. The increased flexion may be attributable to the tendency for amputees to increase their step length on the prosthetic side. 9,11,12,26 It may also be related to concurrent differences in knee angles. For instance, if the knee was flexed more than mean normal in stance phase, as it was in many trials of this study, the hip would be flexed more to keep the posture upright. The increased hip flexion could also be the result of a slight forward-bending posture. Because BKA are less secure on their artificial limb, they tend to lean forward to bring the center of gravity over the foot.
Similar to the findings of Breakey, 1 all of our subjects, except one, performed higher than mean normal knee flexion on the unaffected side during early stance. The most persuasive reason for this is that it provides a compensatory energetic advantage by increasing the K2 power burst. This is addressed in the section on powers. Although the prosthetic ankle was fixed, some degree of ankle dorsi- and plantar-flexion movement can be seen on the joint profiles (Figure 1). This is attributed to compression of the elastic foam of the artificial foot, especially at the forefoot. Increased dorsiflexion in mid to late stance resulted from an increase in weight bearing on the forefoot, whereas plantarflexion resulted from a decrease in load in late stance. On the unaffected side, the profiles for subjects H001 and H004 were very close to mean normal profiles. Close to mean normal stride length values were also seen in these two subjects. Subject H005, who had the shortest stride length, also demonstrated ankle relative angle profiles on the unaffected side that were very different from mean normal values.
Figure 2 indicates the hip, knee, and ankle moments of one subject for both affected and unaffected sides. The moment at the hip on the affected side varied greatly among subjects. During stance, two participants showed a higher than mean normal extensor moment, whereas in others, extensor moments were close to mean normal values. On the unaffected side, a higher than mean normal hip extensor moment during most of the stance and a higher than mean normal hip flexor moment in early swing was common among nearly all subjects. As expected, both peaks occurred approximately concurrently with the peak knee extensor moment. The hip extensor moment in early stance has two functions, 1) to provide support to the linked multi-segment system, and 2) to provide a small amount of power to contribute to the work of walking. With regard to the first, the hip extensor moment, the knee extensor moment, and the ankle plantarflexor moment usually sum to a typical value called the support moment, 20 and trade-off between joints is common. The support moment in early stance is provided almost entirely by the hip and knee extensor moment. However, the knee moment was larger than normal, and potentially no additional hip extensor moment was needed. A more likely explanation lies in the need to provide additional power for the work of walking, which will be discussed later.
A higher than mean normal knee extension moment peak on the unaffected side during stance occurred in all subjects. Because a lower body position occurred during prosthetic foot roll over, the knee of the unaffected side was more flexed than mean normal. This would require a larger than normal extension moment. On the unaffected side, a higher than mean normal extensor moment also can be seen during late stance. As discussed previously, downward movement of the body attributed to the heel compression and stump vertical movement on the prosthetic side would lead to increased knee flexion. Greater knee flexion during stance normally results in larger knee extension moments to prevent collapse. Another reason for the high extensor moment in late stance of the unaffected side could be the desire to create more support and security. The knee extensors are major contributors to the support of the lower limb segments. 20
The differences of the affected knee moment profiles from normal values during the first half of stance were small in most of participants. Similar to the findings of Winter and Sienko, 19 two subjects showed near zero extensor moments in early stance. The remaining three showed profiles moderately reduced from mean normal values. These variations will be discussed with power phase K2, which follows.
The ankle dorsiflexor moment in early stance on the affected side was both longer in duration and larger in amplitude than that of mean normal subjects. In normal subjects, foot flat occurs under the control of the dorsiflexor muscles, achieving an early foot flat position optimal for obtaining a balance position after heel strike. However, amputees who are using a SACH foot have no means of plantarflexing the foot; therefore, the beginning of the plantarflexor moment is delayed. The results of this study are very similar to those of Winter. 20
As the center of pressure moved forward on the prosthetic foot, an internal plantarflexor moment was generated and increased up to 60% to 80% of mean normal in nearly all subjects, which was slightly more than the 60% to 70% reported by Winter and Sienko. 19 Unlike the affected side, the moment during late stance was larger than mean normal on the unaffected side of two of the subjects. However, it was lower than mean normal in one of the subjects who also showed the slowest speed of walking. Larger than mean normal moment generation on the unaffected side during late stance is consistent with increased knee flexion and with attempts by the subject to provide compensatory power at A2.
Figure 3 indicates the hip, knee, and power profiles of one subject for both affected and unaffected side. Similar to the study of Winter and Sienko, 19 participants of this study showed a strong H1 on the affected side. When walking at self selected speed, the affected side had a higher peak power than the unaffected side at H1. A higher than mean normal H1 represents greater hip extensor activity to propel the body forward during walking and to provide additional power in partial compensation for small values of A2 on the affected side.
When compared with mean normal values, H3 on the unaffected side was higher in most of the subjects. H3 is a power burst occurring at pull-off and is a major contributor to the work of walking. If the other major contributors to the work of walking (ie, the ankle plantarflexors at A2) are deficient, H3 frequently acts as a means of compensation. This may occur as an intra-limb phenomenon (ie, on the same side as the deficiency) or inter-limb (ie, in the contralateral limb). Alternatively, because the swing phase was shorter than normal on the unaffected side, the angular velocity of the limb during swing was greater than mean normal. This would result in an increased H3, generating more power at the hip by starting a faster swing. Because a prosthetic limb weighs less than a natural limb, and given that H3 represents the power that is generated to start swing phase, lower than mean normal power generation at H3 might be expected. The results clearly indicate that the increase due to compensation outweighs any decrease due to prosthetic weight.
The K1 and K2 bursts of power on the affected side were absent or near zero in nearly all subjects. K3, an absorption phase, was higher than mean normal in all subjects. These findings do not agree with those of Gitter et al., 27 who reported a K3 within the normal range. A large K3 is usually seen associated with a need for increased stability, as in the elderly 22 and in the presence of many pathologies. 28,29 The large K3 seen also on the unaffected side can likely be attributed to the same cause. The peak power absorption at K4 on the unaffected side was larger than the affected side in all trials.
There was more energy absorption than mean normal at A1 on the affected side. This energy absorption occurred too late to be attributed to the cushion heel. However, it may be attributed to the compression of the elastic materials of the foot. Part of this energy was stored and returned in the small A2 phase, but most would be dissipated by those materials. On the unaffected side, there was also more than mean normal energy absorption at A1 in most subjects. Unlike the affected side in which the power was generated mostly by the compressed material of the prosthetic foot, on the unaffected side, it is attributable to eccentric muscle activity of the limb. This finding is consistent with the increased dorsiflexion in stance and associated with a long stride of the prosthetic side.
The A2 of the affected side in this study was larger than those reported by Winter and Sienko, 19 and Winter 20 (Figure 3). These two studies have reported a very low peak A2 on the affected side for amputees compared with the findings of this study. In another study by Gitter et al., 27 the peaks A1 and A2 are both reported to be less than the findings of this study. The peak A2 that was reported by Gitter et al. 27 was below mean normal values. This can be attributed to the foot material. In all of those studies, amputees were using SACH feet, whereas in this study, they were all using a SAFE foot.
The large A2 of the unaffected side that occurred in most of subjects can be explained as an attempt by the sound limb to compensate for the deficient power generation of the prosthetic ankle. A complementary explanation is that this action enables the amputee to reach a more safe and secure position. Initially, during heel strike, the small heel contact area of the prosthesis is an insecure situation for weight bearing, whereas the next phase (ie, foot flat) is more secure because of a larger foot contact area with ground. In an attempt to reach a more secure situation on the affected side, the amputee may generate more power on the unaffected side.
Table 3 summarizes the major findings of this study. The total positive-negative work done by the lower limb was greater on the unaffected side than affected side. Similar to the findings of Colborne et al. 25 on the unaffected side, the hip produced most of the positive work while the ankle output was below normal.
The findings of this study confirmed that BKA perform an asymmetrical gait. The major kinematic and kinetic variations from normal gait found in this study can be grouped within two phases of the gait cycle: 1) early stance of the affected side, ie, the prosthetic foot is spatially ahead of the unaffected foot; and 2) early stance of the unaffected side, ie, the unaffected foot is spatially ahead of the prosthetic foot.
In early stance of the affected side, more hip flexion was evident, due to a longer step, prosthetic fitting of the knee is flexion, and/or attempts to increase energy at K2. There was more knee flexion on the affected side, probably attributable to the prosthetic fitting and the desire to increase the power burst at K2. In early stance of the unaffected side, more than mean normal hip flexion is evident due to the slight forward bending posture of the trunk of amputees. More than mean normal flexion of the unaffected knee during stance is likely attributable to a compensatory means of increasing the energy of walking by increasing K2. The power profiles showed that amputees have little power generation during push-off of the prosthetic side. All the power differences in the gait can be attributed to an attempt of the amputee to compensate for the missing prosthetic ankle power generation by producing more power at H1 and H3 of both affected and unaffected sides and A2 of the unaffected side.
LIMITATIONS OF THE STUDY
A number of limitations arise from the single-case nature of the design. The number of subjects in this study was very small, and generalization of the findings of this study to the BKA population cannot necessarily be made. Many factors that were expected to affect the gait performance, such as gender, prosthetic type, type of suspension, foot type, and maturity of amputation, were controlled. However, the small number of subjects made it impossible to evaluate the effects on the gait of such factors as age and muscle strength that are known to affect gait pattern. 22 The highly consistent curves for the three trials of each condition for subjects (not illustrated, correlation value > 0.95) indicate high reliability of measurement.
Results may be considered limited by the use of a two-dimensional analysis system. A three-dimensional motion analysis system could collect more accurate data and enable us to discuss the movement in the frontal plane.
1. Breakey JM. Gait of unilateral below-knee amputees. Orthot Prosthet. 1976; 30: 17–24.
2. Doane NF, Holt LE. A comparison of the SACH and single axis foot in the gait of unilateral below-knee amputees. Prosthet Orthot Int. 1983; 7: 33–36.
3. Culham EG, Peat M, Newell E. Below-knee amputation, a comparison of the effect of the SACH foot and single axis foot on electromyographic patterns during locomotion. Prosthet Orthot Int. 1986; 10: 15–22.
4. Engsberg JR, Lee AG, Patterson LJ, Harder JA. External loading comparison between able-bodied and below-knee amputee
children during walking. Arch Phys Med Rehabil. 1991; 72: 657–661.
5. Prince F, Allard P, Therrien RG, McFadyen BJ. Running gait impulse asymmetries in below-knee amputees. Prosthet Orthot Int. 1992; 16: 19–24.
6. Perry J. Pathologic gait. In: Atlas of Orthotics. American Academy of Orthopaedic Surgeons, St. Louis: Mosby; 1975: 144–168.
7. Hurley GRB, McKenney R, Robinson M, Zadravec M, Pierrynowski MR. The role of the contralateral limb in below-knee amputee
gait. Prosth Orthot Int. 1990; 14: 33–42.
8. Bodily K, Burgess E. Contralateral limb and patient survival after leg amputation. Am J Surg. 1983; 146: 280–291.
9. Sanders TG. Lower Limb Amputations: A Guide to Rehabilitation. New York: F.A. Davis Company; 1986.
10. Glattly HW. A statistical study of 12000 new amputees. South Med J. 1964: 57: 1373–1378.
11. Robinson JL, Smidt GL, Arora JS. Accelerographic, temporal, and distance gait factors in below-knee amputees. Phys Ther. 1977; 57: 898–904.
12. Smidt LG. Gait in Rehabilitation. New York: Churchill Livingstone; 1990.
13. Waters RL, Perry J, Antonelli D, Hislop H. Energy cost of walking of amputees: The influence of level of amputation. J Bone Joint Surg Am. 1976; 58: 42–46.
14. Skinner HB, Effeney DJ. Gait analysis in amputees. Am J Phys Med. 1985; 64: 82–89.
15. Huang CT, Jackson JR, Moore NB, Fine PR, Kuhlemeier KV, Traugh GH, Saunders PT. Amputation: Energy cost of ambulation. Arch Phys Med Rehabil. 1979; 60: 18–24.
16. Tibarewala DN, Ganguli S. Pattern recognition in tachographic gait records of normal and lower extremity handicapped human subjects. J Biomed Eng. 1982; 4: 233–240.
17. Kegel B, Burgess EM, Starr TW, Daly WK. Effects of isometric muscle training on residual limb volume, strength, and gait of below-knee amputees. Phys Ther. 1981; 61: 1419–1426.
18. Lewallen R, Quanbury AO, Ross K, et al. A biomechanical study of normal and amputee
gait. In: Winter D, Norman R, Wells R, et al.,, eds. Biomechanics IX-A. Champaign, IL: Human Kinetics Publishers; 1985: 587.
19. Winter DA, Sienko SE. Biomechanics of below-knee amputee
gait. J Biomech. 1988; 21: 361–367.
20. Winter DA. The Biomechanics and Motor Control of Human Gait: Normal, Elderly, and Pathological. 2nd ed. Waterloo, Ontario: University of Waterloo; 1991.
21. Zheng D. The validation of peak performance two-dimensional motion measurement system. [thesis]. Kingston, Ontario: Queen’s University, School of Rehabilitation Therapy; 1992.
22. Winter DA, Patla AE, Frank JS, Walt SE. Biomechanical walking pattern changes in the fit and healthy elderly. Phys Ther. 1990; 70: 340–347.
23. Finley FR, Cody KA. Locomotive characteristics of urban pedestrians. Arch Phys Med Rehabil. 1970; 51: 423–426.
24. Murray MP, Kory RC, Clarkson BH. Walking patterns in healthy old men. J Gerontol. 1969; 24: 169–178.
25. Colborne GR, Naumann S, Longmuir PE, Berbrayer D. Analysis of mechanical and metabolic factors in the gait of congenital below knee amputees: A comparison of the SACH and Seattle feet. Am J Phys Med Rehabil. 1992; 71: 272–278.
26. Mizuno N, Aoyama T, Nakajima A, Kasahara T, Takami K. Functional evaluation by gait analysis of various ankle-foot assemblies used by below-knee amputees. Prosthet Orthot Int. 1992; 16: 174–182.
27. Gitter A, Czerniecki JM, DeGroot DM. Biomechanical analysis of the influence of prosthetic feet on below-knee amputee
walking. Am J Phys Med Rehabil. 1991; 70: 142–148.
28. Olney SJ, Griffin MP, Monga TN, McBride ID. Work and power in gait of stroke patients. Arch Phys Med Rehabil. 1991; 72: 309–314.
29. Olney SJ, MacPhail HE, Hedden DM, Boyce WF. Work and power in hemiplegic cerebral palsy gait. Phys Ther. 1990; 70: 431–438.