Contributions of Lower-Limb Muscle Power in Gait of People Without Impairments.

Gait analyses of subjects without lower-extremity impairments are done to characterize gait to provide a framework to understand musculoskeletal disorders in order to improve rehabilitation outcomes. Similarities or dissimilarities between the lower limbs of people without impairments might affect a clinician's interpretations of data obtained from people whose gait is affected by pathology.

Differences between the right and left limbs have been reported for subjects without impairments. For example, Herzog et al[1] found that asymmetries were much larger than expected in 34 gait variables for a control group. Variables such as maximum vertical force showed unexpected variability in asymmetry. Gundersen et al[2] also documented gait asymmetry in temporal and kinematic variables, challenging the assumption that right and left limbs function symmetrically.

Differences between the lower limbs during gait ca be explained, in part, by the relative contribution each limb makes to the control and propulsion required during Gait. Hirasawa[3] provided evidence to support this assumption, claiming that the left lower limb contributed mainly to body weight transfer during walking, whereas the right lower limb was responsible for propulsion. In evaluating the lateral component of the ground reaction force in 28 subjects, Matsuska et al[4] reported that the mediolateral balance in walking was controlled mostly by the left limb. For 53 male subjects and 39 female subjects walking at slow, free, and fast speeds, Hirokawa[5] associated propulsion with the right limb, whereas the left limb was found to be responsible for support. Unfortunately, these researchers usually focused on a single variable at a time. Recently, Sadeghi et al[6] suggested that gait asymmetry in people without impairments can be explained in terms of actions taken by the lower limbs to propel the body segments and to control their forward progression. However, none of the studies discussed provided data on how these actions are managed within each limb to achieve propulsion and support.

Muscle powers that combine both kinetic information (eg, moments) and kinematic information (eg, angular velocity) appear to us to be good indicators of people's ability to propel and control their lower limbs.[7,8] Muscle powers have been used to describe the gait of people without impairments,[9,10] people with amputations,[11] and patients fitted with a total hip prosthesis.[12] The interaction among muscle powers could reflect specific propulsion and control strategies related to each limb. In gait associated with pathologies, these interactions could be perturbed, resulting in compensatory actions. Understanding lower-limb muscle power relationships contributing to control and propulsion during gait in people without pathologies could be useful in distinguishing between asymmetries due to pathologies and impairments and those resulting from compensations. A well-known example would be the lack of push-off in people with lower-limb amputations compared with the exaggerated hip flexion compensation in propelling the affected limb forward.[13] Although examples of movement in the sagittal plane are easier to assess and understand because motion is assumed to occur in a single plane and about a single axis of rotation, 3-dimensional data can highlight more complex interactions occurring in the other planes as well.

We postulated that gait asymmetry in people without impairments can be explained in terms of actions taken by the lower limbs to propel the body segments and to control their forward progression. We also hypothesized that the muscle powers developed within a limb to accomplish control and propulsion tasks are responsible, in part, for gait asymmetries. Using muscle powers, the purpose of our study was to test the hypothesis that one limb (leading limb) contributes mainly to forward progression, whereas the other limb (trailing limb) provides control and contributes to propulsion to a lesser extent.

Method

Subjects

The 19 men who participated in this study had an average age of 26.18 years (SD = 3.2, range = 21-34), an average height of 1.77 m (SD = 0.06, range = 1.70-1.87), and an average body mass of 82.4 kg (SD = 13 range = 66.7-112). Subjects had no recent history off orthopedic ailments such as a recent injury or surgery that could affect their walking pattern. All subjects were right-hand and right-leg dominant, as determined by 5 activities commonly used[14,15] to identify limb preference (ie, kicking a ball, hopping on one foot, throwing a ball, writing, and opening a jar). Although this report does not focus on limb dominance, we wanted to have a uniform sample of subjects in case dominance made a difference.

Procedure

A 3-dimensional (3-D) 7-body segment model consisting of the trunk, thighs, shanks, and feet was defined by means of 20 reflective markers with a diameter of 2.5 cm.[6] For each foot, markers were placed over the lateral malleolus, the heel, and the lateral border of the fifth metatarsophalangeal joint. Markers also were placed over the apex of the lateral epicondyle and the mid-lateral sides of the tibia to locate the shanks as well as over the mid-lateral sides of the thighs and the greater trochanters. For the pelvis, markers were put over the anterior superior iliac spines and iliac crests. Markers of the pelvis as well as those put over the lateral border of the shoulders identified the trunk. To calculate motion in the joint coordinate system, measurements were taken between the external markers and the estimated joint center of rotation of each lower limb.

Bilateral kinetic gait data were collected by means of 2 AMTI force plates,(*) which were centrally located along a 13-m walkway, and an 8-camera video system.([dagger]) To cover 2 consecutive strides, 4 cameras were placed on each side of the subject at an average distance of 4.5 m and located along an arc of about 120 degrees. After camera calibration, the subjects, who wore comfortable shorts and running shoes, were asked to walk along the walkway at their own free walking pace. Because all subjects were right-side dominant, the right lower extremity was arbitrary chosen as the leading lower extremity and the left lower extremity was arbitrarily chosen as the trailing lower extremity. Videotaped data were recorded at 90 Hz, and the synchronized force data were sampled at 360 Hz. Although data were collected for 5 walking trials, data for the 3 bilateral gait trials that were closest to a subject's mean walking speed were selected for analysis.

The Direct Linear Transformation software of the Expert-Vision Motion Analysis System([dagger]) was used to reconstruct the image markers into 3-D coordinates. Noise in the videotape and force data was reduced by means of a fourth-order zero-phase-lag Butterworth filter with cutoff frequencies of 6 and 30 Hz, respectively.[16] The root mean square absolute error of the 3-D coordinates was less than 5 mm, and the relative measurement error corresponding to the distance between 2 known markers was less than 1 mm.

Body segment, kinematic, and force plate data were used to calculate the net muscle moment at each joint of the lower limbs and in each plane throughout the gait cycle.[6,9] Instantaneous muscle powers were calculated as the product of the net muscle moment and joint angular velocity. The mechanical energy generated and absorbed at each joint was calculated from the muscle powers developed in each plane. For averaging purposes, muscle powers and their respective mechanical energies were normalized with respect to each subject's body mass.[6,9]

To illustrate the 3-D muscle powers developed during the gait cycle and to identify the peak values that were selected for analysis, the mean trailing limb curves were overlaid on the same time scale of the mean leading limb values with their standard deviations, as shown in the Figure. This was done by normalizing the power curves of each limb with respect to the duration of the gait cycle. The gait cycle of the leading limb was 1.45 m (SD=0.7) in duration and normalized to 100%, with a mean stance phase at 60.7% (SD=1.7%). The gait cycle of the trailing limb was 1.47 m (SD=0.7) in duration and also was normalized to 100% but began at 10.6% of the gait cycle of the leading limb and ended at 110.6% of the gait cycle of the leading limb. The mean stance phase was 61.0% (SD=1.5%) and occurred at 71.6% of the gait cycle on the leading limb scale. This type of simultaneous and bilateral representation of the muscle power curves was previously reported and discussed by Allard and colleagues.[10,17] These curves generally correspond to those reported by Eng and Winter.[9]

A 2-step approach was used to identify the main muscle power contributions of each lower limb during gait and to determine their interactions within each limb. For each limb, a principal-component analysis (PCA) was first used to classify the 52 gait variables into smaller data sets or principal components. These variables, which are listed in Table 1, consisted of 8 spatiotemporal variables and 22 peak muscle powers for each limb and their corresponding mechanical energies.[10]

Table 1.

Names of 8 Spatiotemporal Variables and 22 Peak Muscle Powers and Their Corresponding Energies Calculated for Each Lower Limb(a)

Spatiotemporal                                       Mechanical
Variable              Muscle Power                   Energy

Speed                 H1S (hip extensors)            H1S
Stride length         H2S (hip flexors)              H2S
Stance                H3S (hip flexors)              H3S
Step length           H1F (hip abductors)            H1F
Cadence               H2F (hip abductors)            H2F
Breaking phase        H3F (hip abductors)            H3F
Double support time   H1T (thigh medial rotators)    H1T
Terminal double       H2T (thigh lateral rotators)   H2T
  support time        H3T (thigh medial rotators)    H3T
                      K1S (knee extensors)           K1S
                      K2S (knee extensors)           K2S
                      K3S (knee extensors)           K3S
                      K4S (knee extensors)           K4S
                      K1F (knee adductors-           K1F
                       abductors)
                      K2F (knee adductors-           K2F
                       abductors)
                      K1T (knee medial rotators)     K1T
                      K2T (knee lateral rotators)    K2T
                      K3T (knee medial rotators)     K3T
                      A1S (ankle dorsiflexors)       A1S
                      A2S (ankle plantar flexors)    A2S
                      A1F (ankle evertors)           A1F
                      A2F (ankle invertors)          A2F

(a) The peak muscle powers listed in this table are identified in the Figure, and their mechanical energies correspond to the area under the specific power burst. The main function of each muscle power is given in parentheses.

Data Analysis

To reduce the number of gait variables, the first 4 principal components that contained over 60% of the information were retained.[18] Then, within each of the extracted principal components, the gait variables with a factor loading of 0.6 or higher were selected for further analysis. The reader is referred to Olney et al[19] for a detailed clinical example of the PCA method. In the second step, the Pearson correlation (r) analysis was carried out to determine the interactions among the gait variables identified by the PCA within each limb. Those that were significant at P [is less than] .05 and had a correlation of .60 or higher were retained. This analysis was performed twice, once for the gait data of each limb. All statistical analyses were performed using Statistica software.([double dagger])

Results

The factor loadings of the significant temporal, power, and energy variables derived from the PCAs of the leading and trailing limbs are presented in Tables 2 and 3, respectively. For the leading limb, 9 of the 52 gait variables were found to be significant. These variables are listed in Table 2 and were mostly related to hip activity. Six of the 9 gait variables were associated with muscle power generation. Walking speed and step length were identified by the PCAs, though no differences were noted between limbs. More gait variables were required to describe the trailing limb activity (Tab. 3). Sixteen of the 52 variables were selected. Both absorption and generation power bursts were identified in all 3 planes, and they were about equally distributed between the hip and the knee.

Table 2. Right-Limb Factor Loadings of the Significant Peak Powers, Mechanical Energies, and Temporal Variables, With Their Corresponding Standard Deviations in Parentheses

Leading   Factor     Muscle Powers   Energy          Temporal
Limb(a)   Loadings   (W/kg)          (J/kg           Variables
                                     [10.sup.-2])

H1S       0.86        1.50 (1.10)
H1S       0.77                       17.10 (17.50)
H3S       0.74        3.02 (1.42)
H3S       0.78                       29.60 (17.90)
H3F       0.83       -0.19 (0.14)
H2T       0.83                        5.32 (3.90)
K1S       0.73                       -4.80 (3.00)
K3T       0.60        0.07 (0.07)
Speed
 (m/s)    0.71                                       1.30 (0.12)

(a) See Table 1 for explanation of abbreviations.

Table 3.

Left-Limb Factor Loadings of the Significant Peak Powers, Mechanical Energies, and Temporal Variables, With Their Corresponding Standard Deviations in Parentheses

Trailing   Factor     Muscle Powers   Energy           Temporal
Limb(a)    Loadings   (W/kg)          (J/kg            Variables
                                      [10.sup.-2])

H1S        0.81        1.23 (0.99)
H1S        0.73                        24.20 (22.60)
H1F        0.80       -0.95 (0.57)
H1F        0.79                       -20.50 (16.10)
H1T        0.60       -0.17 (0.17)
H1T        0.60                        -1.70 (2.40)
K2S        0.70        0.56 (0.53)
K2S        0.74                         5.30 (5.40)
K3S        0.73                        -5.30 (0.04)
K1F        0.87       -0.11 (0.16)
K1F        0.85                        -0.90 (1.50)
K2T        0.73       -0.07 (0.10)
K2T        0.68                        -0.60 (0.70)
K3T        0.89        0.05 (0.16)
K3T        0.89                         0.40 (1.10)
Step
 length
 (m)       0.75                                        0.71 (0.04)

(a) See Table 1 for explanation of abbreviations.

Because the PCAs did not identify any variables associated with the swing phase, our discussion will essentially be limited to the stance phase. Only the hip extensor (H1S) and knee medial (internal) rotator (K3T) variables were common to both limbs. The hip extensor (H1S) generation burst has been associated with control of the trunk and collapse of the stance limb[9] as well as with forward progression.[7,10] This generation burst occurred as the thigh extended at 12% of the gait cycle. Both limbs generated about the same amount of peak powers and energies. The knee medial rotator (K3T) power burst, which occurred at about 55% of the gait cycle, was associated by Allard et al[10] with a medial rotation of the thigh during push-off in moving the body's center of mass in preparation for the subsequent step.

For each limb, the Pearson correlation was applied between all gait variable pairs listed in Tables 2 and 3 to determine the interactions within each limb. For the leading limb, 6 variables had a statistically significant Pearson correlation coefficient of .60 or higher, whereas 14 variables had a Pearson correlation coefficient of .60 or higher for the trailing limb (Tab. 4).

Discussion

The objective of our study was to test the hypothesis that the leading limb contributes mainly to forward progression, whereas the trailing limb provides control and contributes to a lesser extent to propulsion through different power burst interactions.

Leading Limb Power Interactions

The peak muscle power of the hip extensors (HIS) and its related mechanical energy, which occurred shortly after heel-strike, has been associated by other researchers with control of the forward acceleration of the trunk[20] and the potential collapse of the stance limb[9] as well as with forward progression.[7,10] The measurements of hip extensor (H1S) power and mechanical energy were moderately correlated (r=.60 and .61, P [is less than] 0.001) with the measurements of hip flexor energy (H3S). The hip flexor (H3S) activity occurred at the end of the stance phase, and this gait variable was assumed by Winter et al[20] to pull the thigh upward. The interaction between the hip extensors (H1S) and the hip flexors (H3S) can be explained, in part, by the fact that both contribute to forward progression.

Winter[16] assumed a relationship between the hip extensor (HIS) and hip flexor (H3S) power generation bursts when he reported that the sagittal extensor moments contributed to propelling the body forward. Our results support this observation, while revealing another dimension that involved the pelvis during mid-stance. The highest correlation was found between the hip extensor (H1S) peak power and the mechanical energy generated at the hip when the thigh was laterally (externally) rotating (H2T), which occurred during the second half of mid-stance (r=.93 and .60, P [is less than] .001). During this time, the pelvis was rotating forward and bringing the trailing limb forward under the influence of the thigh lateral rotators (H2T), thereby contributing to the progression of the trunk. The energy associated with the thigh's external rotator (H2T) power activity, which occurred during mid-stance, was related to the hip mechanical energy (H3S) generated at the end of the stance phase during the push-off period (r=.61, P [is less than] .001).

Pelvic rotation was recognized by Saunders et al[21] as one of the 6 gait determinants. They suggested that, because the pelvis is a rigid structure, it contributes to forward progression by its alternate rotations about each hip (H2T). Our finding supports the hypothesis that both pelvis rotation and the hip extensor of the supporting limb serve to effectively lengthen the limb and reduce excessive drop of the body's center of mass,[22] smoothing its vertical transition.[23] From these observations, we believed that gait propulsion is not limited to the push-off period. It is an activity that is initiated shortly after heel-strike when the hip extensors eccentrically contract (H1S), that is maintained during mid-stance (H2T), and that is completed at push-off by the hip flexors (H3S) when the muscles concentrically contract. Greater pelvis stability is provided by the hip abductors (H1F), which leads to improved forward progression. Pelvic tilt was also seen by Saunders et al[21] as one of the gait determinants that reduced vertical oscillation of the trunk.

The hip power absorption developed in the frontal plane at heel-strike (H1F) and throughout mid-stance was negatively correlated with measurements obtained for the hip flexors (H3S) (r=.60, P [is less than] .001). Allard et al[10] and Mackinnon and Winter[24] contended that the H1F controls pelvic tilt as the trailing limb enters the swing phase. During mid-stance, the hip would then contribute to forward progression by propelling the limb forward through hip extensor activity (H1S) while the hip abductors (H1F) stabilize pelvic tilt. In our study, the leading limb displayed a tendency for propulsion. This propulsion was achieved mainly by the hip, with a strong sagittal muscle power contribution, which occurred throughout the Stance phase.

Trailing Limb Power Interactions

Measurements of hip extensor power (H1S), which is considered a source of propulsion,[6,7] were moderately correlated (r=.60, P [is less than] .001) with measurements of knee power generation in the sagittal plane (K2S), as shown in Table 4. This K2S was associated by Winter et al[20] with knee extension following a flexion controlled by the extensors during early mid-stance. The K2S can be considered a propeller, but its contribution is less effective because its primary function is to extend the knee and prepare the lower limb for push-off.

Table 4.

Pearson Correlation Coefficient Values for Right and Left Lower-Limb Peak Powers (P) and Mechanical Energies (E)

Leading
Limb               H1S (E)   H1S (P)   H3S (E)   H1F (P)

H3S (E)               .60       .61                -.60
H2T (E)               .60       .93       .61

Trailing
Limb       H1S (P)   H1F (E)   H1F (P)   K3T (E)   K3T (P)

K2S (E)    .60
K2S (P)    .60
K2T (E)               .60       .60      -.60      -.61
K2T (P)               .60       .70      -.61      -.66
K3T (P)              -.71      -.74
K3T (P)              -.61      -.74

(a) See Table 1 for explanation of abbreviations.

The hip power absorption developed in the frontal plane at heel-strike (H1F) and throughout mid-stance controls pelvic tilt as the right limb enters the swing phase.[10,23] The moderate association between measurements obtained for the hip abductors (H1F) and knee lateral rotators (K2T) (r=.60, P [is less than] .001) linked together activities related to the control and preparation of the right limb heel-strike. Prior to the double support period, pelvic tilt is controlled to ensure a safe body weight transfer. Once the initial ground contact is made and the leading limb begins to safely bear weight, the trailing knee rotates inwardly, contributing to a progressive body weight transfer under the control of a lateral rotation moment (K2T).

The medial rotation of the trailing shank actively contributes to body weight transfer by the power generation of the knee medial rotators (K3T). This observation is supported, in part, by a reasonable negative correlation between the knee lateral rotators (K2T) and knee medial rotators (K3T) (r=-.60, P [is less than] .001) and a high negative relationship between the hip abductor (H1F) and knee medial rotator (K3T) peak muscle powers (r=-.71, P [is less than] .001). Thus, the relationships between the H1F, K2T, and K3T powers can be explained in terms of control actions by the lower trailing limb. These findings were confirmed by Winter et al,[20] who suggested that transverse knee moments are passive elements that react to the hip moments.

Our description of controlled body weight transfer may be applicable to both limbs, but the interactions between these hip and knee powers were not found for the leading limb. We believe that this finding can be explained, in part, by strong leading limb propulsion. Although the trailing limb was in the swing phase, it may have deviated from the body path of progression by the leading limb push-off. A correction by the trailing limb could be achieved only during its own push-off period by the K2T and K3T power bursts where the shank was medially rotating and orienting the trunk and lower limb in line with the intended path of progression.

The trailing limb was characterized by many more peak powers and mechanical energies than the leading limb. Some of these peak powers and mechanical energies necessarily contributed to forward progression; otherwise, our subjects would have displayed evident gait perturbations. However, they were sufficient to propel the trailing limb forward at about the same speed ([bar] X=1.32 m/s, SD=0.10) as in the leading limb ([bar] X=1.30 m/s, SD=0.12). The majority of the powers and energies occurred in the transverse and frontal planes, and most of them were associated with energy absorption. These absorption powers were mainly associated with the control actions of the lower limb during the mid-stance and push-off periods. When power generation activity was present, propulsion was mostly secondary to control activities, and they may have occurred to correct actions resulting from the leading limb propulsion. We hypothesize that the principal function of the trailing limb was related to control activities.

The ankle power generation burst that developed at push-off (A2S) was not selected in the first 4 principal components of the PCA as a key variable for either limb. This finding was somewhat surprising. However, it did appear in the PCA, but only for the leading limb and only in the fifth principal component, which explained less than 10% of the total variance. The peak (A2S) power value ([bar] X=3.21 W/kg, SD=0.69) and mechanical energy did not display a good correlation with respect to the other variables. The highest correlation of the A2S was -.56 with the hip abductor (H3F) power. Considering the strong hip activity interactions and the absence of ankle power or energy variables in the PCA, these results support the passive role played by the ankle in propelling the lower limb forward.[23] These results may explain, in part, the observations of Wagner et al[25] and Prince et al,[26] who reported similar ground reaction forces for people with lower-limb amputations who were fitted with either a SACH (solid-ankle cushion heel) foot or an energy-storing foot prosthesis.

In this article, we present simultaneous 3-D bilateral data in an attempt to explain gait asymmetry in people without known pathologies or impairments in relation to tasks associated with control and propulsion. We did not focus on limb dominance but rather on normal gait variations that we consider to be adaptations during 2 consecutive strides. Only right-footed subjects were selected because data from both right- and left-footed subjects could have influenced the results due to limb dominance. Furthermore, right-side-dominant people represent a larger population than left-footed individuals. Footedness, based on the available information, is not known to have a relationship with gait asymmetry. Gundersen et al[2] concluded that asymmetry cannot be predicted by dominance. One limb behaves differently from the other limb. Because we do not have sufficient information to associate these differences with limb dominance, we wanted to express our results in terms of one limb with respect to the other limb in right-footed subjects. The interactions among these gait variables within and between limbs need to be addressed in order to improve our understanding of forward progression and control tasks in the lower extremities during walking.

Although the right limb was selected as the leading limb, it may be interesting to repeat the study with the left limb as the leading limb. By having left-limb-dominant subjects, we could have obtained additional information, but these subjects are more difficult to find.

Several factors may have contributed to the marked difference in the tasks performed by the leading and trailing limbs. Although the walkway was 13 m long and the data were collected in the central portion of the walkway, the subjects may have been influenced by the a priori knowledge that they must come to a full stop within a few meters after stepping on the second force plate. Although Grabiner et al[27] have shown that stepping on the force plate does not influence the gait pattern, this may not be the case when 2 force plates are used. Our study focused on the actions taken by the lower limbs in 2 consecutive gait cycles. To determine whether the leading limb is the principal contributor to propulsion, gait studies after 3 or more consecutive gait cycles are needed.

Conclusion

In our study, we demonstrated that the leading limb mainly contributes to forward progression, whereas the trailing limb provides control and contributes to propulsion to a lesser extent. For the leading limb, propulsion was an activity initiated by the hip shortly after heel-strike and maintained throughout the stance phase. Control was the main task of the trailing limb, as evidenced by the power absorption bursts. The trailing limb power generations were generally secondary to control activities or possibly to correct for the leading limb propulsion. Our results highlight the role of transverse- and frontal-plane actions during gait, and we believe that this role should be taken into account when evaluating gait and in considering approaches to rehabilitation and physical therapy intervention.

([double dagger]) StatSoft Inc, 2300 E 14th St, Tulsa, OK 74104.

References

[1] Herzog W, Nigg BM, Read LJ, Olsson E. Asymmetries in ground reaction force patterns in normal human gait. Med Sci Sports Exerc. 1989;21:110-144.

[2] Gundersen LA, Valle DR, Barr AE, et al. Bilateral analysis of the knee and ankle during gait: an examination of the relationship between lateral dominance and symmetry. Phys Ther. 1989;69:640-650.

[3] Hirasawa Y. Left leg supporting human straight (bipedal) standing. Saiensu. 1981;6:32-44.

[4] Matsuska N, Fujita M, Hamamura A, et al. Relationship between right and left legs in human gait, from a viewpoint of balance control. In: Winter DA, Norman R, Wells R, et al, eds. Biomechanics IX-A. Champaign, Ill: Human Kinetics Publishers; 1985:427-430.

[5] Hirokawa S. Normal gait characteristics under temporal and distance constraints. J Biomed Eng. 1989;11:449-456.

[6] Sadeghi H, Allard P, Duhaime M. Functional gait asymmetry in able-bodied subjects. Human Movement Science. 1997;16:243-258.

[7] Vardaxis VG, Allard P, Lachance R, Duhaime M. Classification of able-bodied gait using 3-D muscle powers. Human Movement Science. 1998;17:121-136.

[8] Olney SJ, Griffin MP, McBride ID. Temporal, kinematic, and kinetic variables related to gait speed in subjects with hemipeligia: a regression approach. Phys Ther. 1994;74:872-885.

[9] Eng JJ, Winter DA. Kinetic analysis of the lower limbs during walking: what information can be gained from a three-dimensional model? J Biomech. 1995;28:753-758.

[10] Allard P, Lachance R, Aissaoui R, Duhaime M. Simultaneous bilateral 3D able-bodied gait. Human Movement Science. 1996;15: 327-346.

[11] Czerniecki JM, Gitter A, Munro C. Joint moment and muscle power output characteristics of below-knee amputees during running: the influence of energy storing prosthetic feet. J Biomech. 1991;24:63-75.

[12] Loizeau J, Allard P, Duhaime M, Landjerit B. Bilateral gait patterns in subjects fitted with a total hip prosthesis. Arch Phys Meld Rehabil. 1995;76:552-557.

[13] Sadeghi H, Allard P, Duhaime M. Muscle power compensatory mechanisms in below-knee amputee gait. Am J Phys Med Rehabil. In press.

[14] Harris AJ. Harris Tests of Lateral Dominance. New York, NY: The Psychological Corporation; 1953.

[15] Dordill CB, Thoreson NS. Reliability of the lateral examination. J Clin Exp Neurophysiol. 1993;15:183-190.

[16] Winter DA. The Biomechanics and Motor Control of Human Gait: Normal, Elderly, and Pathological. 2nd ed. Waterloo, Ontario, Canada: University of Waterloo Press; 1990:39.

[17] Allard P, Lachance R, Aissaoui R, et al. Men and women able gait. In: Allard P, Cappozzo A, Lundberg A, Vaughan CL, eds. Three-Dimensional Analysis of Human Locomotion. New York, NY: John Wiley & Sons Inc; 1997:67-89.

[18] Hamilton HC. Regression With Graphics: A Second Course in Applied Statistics. Belmont, Calif: Wadsworth Inc; 1992.

[19] Olney SJ, Griffin MP, McBride ID. Multivariate examination of data from gait analysis of persons with stroke. Phys Ther. 1998;78:814-828.

[20] Winter DA, Eng JJ, Ishac MG. A review of kinetic parameters in human walking. In: Craik RL, Oatis CA, eds. Gait Analysis: Theory and Application. St Louis, Mo: Mosby-Year Book Inc; 1995:252-270.

[21] Saunders JBDM, Inman VT, Eberhart HS. The major determinants in normal and pathological gait. J Bone Joint Surg Br. 1953;67:237-241.

[22] Perry J. Gait Analysis: Normal and Pathological Function. Thorofare, NJ: Slack Inc; 1992:524.

[23] Bagley AM, Foerster SA, Skinner HB, Mote CD Jr. Energy consumption-prosthesis mass relations for above knee (AK) amputees. Trans Orthop Res Soc. 1990;15:173.

[24] Mackinnon CD, Winter DA. Control of whole body balance in the frontal plane during human walking. J Biomech. 1993;26:633-644.

[25] Wagner J, Sienko S, Supan T, Barth D. Motion analysis of SACH vs Flex-Foot in moderately active below-knee amputees. Prosthet Orthot Int. 1987;11:55-62.

[26] Prince F, Allard P, Therrien RG, McFadyen BJ. Running gait impulse asymmetry in below-knee amputees. Prosthet Orthot Int. 1994; 16:19-24.

[27] Grabiner MD, Feuerbach JW, Lundin TM, Davis B. Visual guidance to force plates does not influence ground reaction force variability. J Biomech. 1995;28:1115-1117.

H Sadeghi, PhD, is Professor, Department of Kinesiology, Tarbait Moallem University, Ministry of Sciences, Research, and Technology, Tehran, Iran, and Postdoctoral Fellow, Research Center, Sainte-Justine Hospital, 3175 Cote-Ste-Catherine, Montreal, Quebec, Canada H3T 1C5 ([e-mail]). Address all correspondence to Dr Sadeghi at the second address.

P Allard, PhD, PEng, is Professor, Department of Kinesiology, University of Montreal, and Director, Human Movement Laboratory, Research Center, Sainte-Justine Hospital, Montreal, Quebec, Canada.

M Duhaime, MD, is Orthopedic Surgeon and Professor of Orthopedic Surgery, Shriners Hospital, Division of Orthopedics, McGill University, Montreal, Quebec, Canada.

Dr Sadeghi provided concept/research design and data collection and analysis. Dr Sadeghi and Dr Allard provided writing, and Dr Allard provided project management. Dr Duhaime provided subjects, fund procurement, and consultation (including review of manuscript before submission).

This article was submitted April 24, 2000, and was accepted July 13, 2000.