Influence of body weight support on normal human gait: development of a gait retraining strategy.

The locomotion of patients with neurological deficits is often impaired by poor muscular activation, [1] poor weight-bearing capacities, [2-4] and poor balance. [5] Although convential treatment regimens usually focus on retraining of these components, [1,5] gait deviations often persist, despite improvement in muscle activation patern [1] and patients' underlying abilities. [6] A spinal animal model, however, demonstrates that recovery of locomotor function is possible. Rossignol and colleagues [7,8] have shown that the adult spinal cat can recover a locomotor pattern similar in many aspects to normal, with proper foot placement and weight support of the hindquarters, after a complete transection of the spinal cord. Their results are mainly due to an interactive locomotor training program that consists of appropriately graded weight support, provided by supporting the cat's tail and allowing the animal to bear only the amount of weight with which it can cope during treadmill locomotion.

Such a training approach, which allows for simultaneous retraining of different components of locomotion, is needed for patients with neurological disorders. Finch and Barbeau [9] have proposed that removal of body weight may facilitate the expression of gait patterns and therefore could be considered a therapeutic tool in gait retraining in patients who have neurological impairment. Although the effects of increased loading have been studied, [2,10-12] the effects of decreased loading on the electromyographic (EMG) characteristics of normal gait remain to be investigated. Hewes et al [13] studied the effects of replicated lunar gravity on the kinematics of walking and running. Lunar gravity was simulated by using a walkway tilted laterally to 9.5 degrees from the vertical. The subjects walked while being supported in slings. The three subjects studied walked and ran 60% slower than under normal conditions. The results showed that there was a decrease in the amplitude of the hip, knee, and ankle angular movements and an increase in the forward inclination of the body. The experimental paradigm, however, did not isolate the effects attributable to decreased loading from those attributable to slower walking speeds assumed by the subjects.

The rationale for our study, which led to the development of a new gait retraining strategy, was based on findings both from the spinal animal model and from clinical gait observations. Our rationale was that supporting a percentage of body weight during gait retraining may facilitate the expression of a more normal gait pattern. Reduction in load through the lower extremities during locomotion can be achieved by the use of body-weight-support (BWS) system (*) that was developed in our laboratory. [14] We believe that before these findings can be extrapolated to patient populations, normal gait studies on the effects of decreased loads in conjunction with walking speed changes during treadmill walking are needed. Thus, the purpose of this study was to investigate the effect of BWS on the EMG, kinematic, and temporal-distance characteristic, of normal human locomotion during treadmill walking Preliminary results demonstrated that progressive BWS facilitates the expression of a more normal gait pattern in patients with spinal cord injuries [15] and in patients with hemiplegia. [16]

Method

Preliminary Tirals

Seven nondisabled volunteers participated in preliminary trials that allowed determination of the BWS levels used in this study. After walking at different BWS levels, only three of the seven subjects could walk with heel contact at 80% of BWS. Therefore, the upper BWS limit was set at 70%; 30%, and 50% of BWS were arbitrarily defined as the middle and lower limits, respectively. None of the seven subjects were able to walk at their natural, comfortable full-weight-bearing (FWB) speed at any BWS level. From these trials, a range of walking speeds for each BWS level was found: for 0% of BWS, 1.20-1.50 [m.ssup.-1]; for 30% of BWS, 0.79-1.00 [m.ssup.-1]; for 50% of BWS, 0.79-0.89 [m.ssup.-1]; and for 70% of BWS, 0.65-0.75 [m.ssup.-1]. Thus, specific walking speeds were predetermined by averaging, after 15 minutes of walking, the comfortable walking speed at each BWS level for the subjects who participated in our preliminary testing.

Experimental Protocol

Ten nondisabled male subjects, with a mean age of 31 years (SD=3.68), a mean height of 1.76 m (SD=0.04), and a mean weight of 72.7 kg (SD=7.00), walked on a Collins treadmill with 0%, 30%, 50%, and 70% of their body weight supported by a modified overhead harness. All subjects wore shorts and running shoes and were novice treadmill walkers with less than 2 hours of experience. All procedures were explained to the subjects, and each subject gave informed consent prior to experimentation. Prior to data collection, each unsupported subject was habituated to the experimental protocol by walking for 20 minutes on a motor-driven treadmill. Throughout all trials, simultaneous EMG and kinematic data from the right lower limb as well as the bilateral footswitches (+) were collected. Equipment information and specifications are presented in the Appendix.

Electromyographic and

Footswitch Data

Electromyographic activity was detected by surface electrodes (Medi-Trace pellet electrodes (++)) applied on the skin, 2 cm apart (center to center) longitudinal to the direction to muscle fibers. The electrodes were centered over the muscle belly following conventional skin preparation. Recordings were obtained for the following muscles on the subjects' right side: erector spinae (electrodes placed 2 cm lateral to the L4-5 interdiskal space), gluteus medius (4 cm posterior to the anterior superior iliac spine), vastus lateralis (10-12 cm superior to the upper edge of the patella and 5-6 cm lateral to the superior midline of hte thigh), medial hamstrings (posterior one third of thigh), tibialis anterior (2 cm lateral and 4 cm below the tibial tubercle), and medial gastrocnemius (2 cm superior to the lower edge of the muscle and 5 cm medial to the midline of the calf). A surface electrode was placed medially on the right leg over the bony surface of the tibia to serve as a ground electrode. Electromyographic signals, after passing through preamplifiers ([section]) to eliminate movement artifacts, were band-pass filtered (10-1,000 Hz), differentially amplified ([section]) (common mode rejection rate=80 dB), and recorded (+++) together with footswitch signals, time code, and auditory signals on a 14-channel magnetic tape at 9.5 cm/s (3.75 in/s (frequency response=2,500 Hz).

Footswitches were placed bilaterally beneath the heel, the head of the fifth metatarsal, and the big toe. Each footswitch produced a distinct voltage (Fig. 1), allowing for determination of temporal-distance characteristics. The time between one heel contact to the next heel contact of the same limb was considered as a 100% cycle. Percentage of stance was defined as the stance time divided by the cycle time. The two double-limb support times were summed to yield a total double-limb support time, which was then normalized to cycle time for subsequent analysis.

A representative sequence of artifact-free EMG signals was chosen for analysis from recorded data after having been played back on a polygraph. After passing through an anti-aliasing filter (450 Hz), the EMG signals were then digitized at 1 kHz for off-line computer analysis (PDP 11.34A (#)) using interactive programs. [17]

All data were displayed on a high-resolution terminal in 10-second sections. A minimum of 10 cycles were chosen for averaging for each BWS and FWS walking speed trial. Figure 1 represents the raw EMG activity recorded from the vastus lateralis muscle of one subject walking at 30% of BWS and at 0.97 [m.ssup.-1]. Our interactive computer programs allowed placement of arrows manually to define onset and termination of each EMG burst. [17] In Figure 1, the first arrow indicates the "on" time and the second arrow indicates the "off" time of the bursts of vastus lateralis muscle activity. The on-off timing for the muscle was normalized as a percentage of the gait cycle. Thus, the normalized on time of a muscle is equal to the percentage of the time from the immediately preceding right heel-strike to the start of the muscle burst divided by the cycle time. The off time was determined in a similar manner for the termination of muscle activity. Because of the wide intersubject variability, the on-off EMG timing was analyzed using descriptive statistics only.

The mean amplitude of each EMG burst, from on time to off time, was determined by the computer as the integrated area under the full-wave-rectified EMG signal divided by the burst duration. This has been a widely used method in quantifying EMG data in both animals [17] and humans. [18] The mean burst amplitude of a specific muscle, for each BWS session and for each control trial, was then normalized, within subjects, using the FWB mean burst amplitude at the speed of 1.36 [m.ssup.-1] as a reference denominator:

Normalized mean burst amplitude= Mean burst amplitude at BWS / Mean burst amplitude at FWB (1.36 [m.s.sup.-1]) x 100%

This normalized mean burst amplitude calculated for each muscle allowed for between-subject and between-trial analysis. Of the investigated muscles that have two bursts during the gait cycle, only the medial hamstring muscle burst occurring during stance and the tibialis anterior muscle burst occurring during swing were analyzed.

Joint Angle Data

To collect joint angle data, subjects were videotaped using a shutter video camera (**) placed 4 m from the center of and perpendicular to the treadmill. Reflective joint markers were placed at the greater tuberosity of the right shoulder, the midline of the rib cage half way between the iliac crest and the shoulder, the greater trochanter, the lateral femoral epicondyle, the lateral malleolus of the fibula, 2 cm above ground in line with the heel of the shoe, the fifth metatarsophalangeal joint, and 2 cm above the sole in line with the toe of the shoe. Trials were recorded on videotape (**) at a speed of 60 frames per second. A remote search controller (**) was used for frame-by-frame viewing. The sagittal angular displacement, one cycle per subject, was manually measured from the monitor screen (**) at each 2% to 5% of the gait cycle, depending on cycle duration. This technique has been used in other studies, [13] and the experimental error of the angular measurements has been reported to be [+ or -]5 degrees. Because the camera was set in line with the BWS system and perpendicular to the sagittal plane of the treadmill, parallax error was minimized. The hip and knee angles were calculated with respect to the vertical, with the neutral position in standing being taken as 0 degrees of displacement of the hip, flexion being positive and extension being negative. The hip and knee angular positions attained at the critical events of heel-strike, foot-flat, mid-stance, heel-off, and toe-off and the maximum knee swing angle (MSA) were plotted on a relative time scale with the gait cycle normalized to 100%. Only the MSAs are reported.

At each BWS level, the distance from the greater trochanter to the floor was measured in a subsample of five subjects. This was done in order to determine whether the height of the greater trochanter was altered by the support system during the experiments. In addition, the distance between the toes of the left foot and the heel of the right foot was measured to record any change in "contact distance."

Body-Weight-Support System

Each subject was mechanically supported in a modified Tyrolean climbing harness over the treadmill. The harness supports the subject primarily about the pelvis and lower abdomen to avoid interfering with lower-limb movement. Force transducers (+) located between the harness and the motor indicated the amount of BWS provided by the harness. After individually calibrating the BWS system to 0% of BWS (equivalent to FWB) and 100% of BWS (total suspension), the percentage of BWS provided during walking could be adjusted accordingly. This BWS system has been described in detail previously. [14]

Experimental Trials

In order to dissociate changes attributable to slower walking speeds from those attributable to the level of BWS provided, each subject participated in two experimental trials performed within a single session. During the first trial, each subject walked at four different, randomly ordered levels of BWS and at a specific speed (1.36 [m.ssup.-1] at 0% of BWS, 0.97 [m.ssup.-1] at 30% of BWS, 0.85 [m.ssup.-1] at 50% of BWS, and 0.70 [m.ssup.-1] at 70% of BWS). During the second trial, each subject, acting as his own control, walked FWB at the four previously assigned speeds. Therefore, the interaction of BWS with walking speed was determined by comparing results across the BWS trials. The effect of walking speed was determined by comparing results across the FWB control trials. The effect of BWS was determined by comparing each BWS trial with the control FWB trial at the specific speed. To minimize fatigue, a 10-minute rest period was provided after each walking trial.

Data Analysis

Different analyses were performed using kinematic, EMG amplitude, and footswitch data from the 10 subjects walking at four BWS levels and four walking speeds. Six repeated-measures analyses of variance (ANOVAs) were performed to determine mean differences in cycle time, percentage of stance, total double-limb support time, single-limb support time, and maximum swing flexor angle of the hip and knee. Repeated-measures ANOVAs were also used to determine mean differences in normalized mean burst amplitude of EMG activity from each of six muscles of the right leg from the 10 subjects at 17 different conditions.

AN F max test (F=largest variance/smallest variance) was used to confirm homogeneity of variance for each variable. Friedman's ANOVA-by-Ranks Test was used with a Wilcoxon Matched-Pairs Signed-Rank Test as the post boc comparison to analyze the nonhomogeneous data and with the Scheffe Multiple-Comparison Test as the post boc comparison to analyze the homogeneous data. [19] All differences reported were statistically significant at .01.

Results

Temporal Data

The mean temporal components are listed in Table 1. The mean cycle time at 1.36 [m.ssup.-1] (the FWB control speed, which is equivalent to 0% of BWS) was 1,084 milliseconds. As determined by the Friedman's ANOVA-by-Ranks ([X.sup.2]=26.6) and Wilcoxon Signed-Rank tests, there was a statistically significant increase in cycle time with decreasing walking speed during the control FWB trials as well as a significant increase in cycle time across the BWS trials (Fig.2A). Nevertheless, no significant difference was found between control FWB cycle time and equivalent BWS trials at a given walking speed. Hence, there was no main BWS effect on cycle time beyond that attributable to walking speed.

There was a trend toward increased percentage of stance with decreasing walking, speed at FWB. This increase, however, was not significant across walking speeds at FWB. The percentage

[TABULAR DATA OMITTED]

of stance decreased significantly (repeated-measures ANOVA, F=32.56) across BWS trials. The percentage of stance values with BWS were all significantly decreased from the FWB control values at a given walking speed (from 60% at 0% of BWS to 52% at 70% of BWS). Thus, the decrease in percentage of stance values was affected mainly by BWS (Fig. 2B).

The total double-limb support time revealed a similar trend to that of percentage of stance. With decreasing walking speed at FWB, there was minimal change in percentage of total double-limb support time. A marked and significant decrease (repeated-measures ANOVA, F=42.86) in total double-limb support time, however, was observed across BWS trials. The total double-limb support time decreased from 22% at 0% of BWS to 9% at 70% of BWS (Fig. 2C). In addition, the percentage of single-limb support time, calculated by subtracting the percentage ot total double-limb support time from the percentage of stance, was unaffected by walking speed in the FWB session, but increased slightly (from 38.2% to 43.1%) across BWS trials. The standard deviations were calculated by obtaining the square root of the sum of the total double-limb support time and percentage of stance variances.

Hip and Knee Angle

Displacement

Figure 3 shows the hip and knee average angular displacement curves for the BWS and FWB control trials. Curves were connected through the points to denote trends. Note the open circles represent the BWS condition. These points represent the mean joint angles of the 10 subjects for each trial. The 0% of BWS plot is included in both sets of curves as a reference line. All curves were plotted normalized to the critical gait events of the 0% of BWS level, with the cycle normalized to 100%. The hip and knee joints demonstrated a similar pattern across all trials with the exception of amplitude of movement.

The angular displacements of the hip and knee at each critical even across the FWB control trials appeared to be similar, with minimal walking speed effect (top half of Figs. 3A and 3B). The MSA values of the hip and knee were not significantly different across FWB control trials (Tab. 2). With BWS, the largest differences in hip angular displacement were at heel-strike, foot-flat, and MSA (Fig. 3A), whereas the largest differences for the knee with BWS were at foot-flat, toe-off, and MSA (Fig. 3B). The ANOVAs and Scheffe comparisons (hip, F=16.95; knee, F=24.66) revealed that, for both the hip and the knee, the MSA of all BWS trials differed significantly from that of the 0% of BWS condition at 1.36 [m.ssup.-1], and the MSA of the 70% of BWS condition was significantly less than that of the FWB condition at any control walking speed (Tab. 2).

The measurement from the greater trochanter to the floor revealed an average increase of 1.5 cm at 70% of BWS across subjects. The contact distance decreased sequentially by 15% at each BWS level from 73.5 cm at 0% of BWS to 22.5 cm at 70% of BWS.

Electromyographic Timing

Results of EMG on-off timing for the muscles investigated are illustrated in Figure 4. In the control FWB trials, walking speed had a minimal effect on timing, except for a delay in both the vastus lateralis and gluteus medius muscles' off timing (Figs. 4B and 4C). The major BWS effects observed were an earlier onset in the first erector spinae muscle burst and delayed offset in the second erector spinae muscle burst (Fig. 4A). These effects led to an increased burst duration, similarly observed in the tibialis anterior and gastrocnemius muscles, as well as a delayed on-off time for the medial hamstring muscles when compared with the FWB controls.

Normalized Mean Burst

Amplitude

The normalized mean burst amplitudes are presented in Table 3. Using Friedman's ANOVA-by-Ranks Test, it was found that the mean burst amplitudes of the erector spinae, gluteus medius, gastrocnemius, and tibialis anterior muscles (Fig. 5) were not significantly affected by decreasing walking speed during the FWB control trials. It is noteworthy that, with decreasing walking speed, the amplitudes of the first and second erector spinae muscle bursts were similar, except at the lowest walking speed. Moreover, there was a trend toward a decrease in the amplitude of the medial hamstring (F=5.06) (Fig. 5E) and vastus lateralis muscles with decreasing walking speed.

Neither the first nor the second erector spinae muscle burst amplitude was significantly affected by BWS, except for the 70% of BWS condition, when the amplitude of both bursts decreased significantly ([xsup.2]=21.89). Throughout all BWS trials, the second erector spinae muscle burst amplitude was greater than the first burst amplitude (Fig. 5A). The gluteus medius muscle mean burst amplitude decreased significantly with BWS (F=12.68), to such an extent that, it the 50% and 70% of BWS trials, the amplitudes were significantly smaller than those at any FWB control trials (Fig. 5C). Vastus lateralis muscle burst amplitude was not sigificantly affected by BWS (F=1.18), because of the large between-subject variability (Tab. 3). The medial hamstring muscle burst amplitude, although remaining stable at 30% and 50% of BWS, decreased to 56.1% at 70% of BWS (Fig. 5D). The medial gastrocnemius muscle burst amplitude decreased significantly ([xsup.2]=23.3) with increasing BWS (as much as 50% of its FWB value at the 70% of BWS level), whereas the tibialis anterior muscle burst amplitude increased ([xsup.2]=21.1) (Figs. 5F and 5E, respectively,).

Discussion

Temporal Data

The progressive decrease found for percentage of stance and percentage of total double-limb support time across BWS trials may have resulted from a pure BWS effect or from a possible interaction between BWS levels and the set walking speed. The results of the combined intervention of increasing BWS and decreasing walking speed suggest that these body weight effects may have been under-estimated. For instance, in the case of percentage of stance, assuming the effects of decreasing walking speed and increasing body weight are additive, the dashed line in Figure 2B would represent the true BWS effect without the effect of walking speed (ie, the increase in percentage of stance attributable to FWB control sessions subtracted from the BWS percentage of stance values). Because our subjects were assigned a specific walking speed per BWS level, the walking speeds may not have matched the amount of body weight support removed for each subject, resulting in a higher-than-normal variability (Tab. 1). The relationship between walking speed and BWS remains to be further investigated.

The BWS effect is evident on the percentage of total double-limb support time; however, the percentage of single-limb support time increased only from 39.8% to 43.1% (Tab. 1). Thus, it would appear that the BWS effects on the overall percentage of stance are greater for the percentage of total double-limb support time, only minimally increasing the percentage of single-limb support time.

The percentage to total double-limb support time for the FWB control sessions (ie, 27%) is not only longer than that normally reported for overground walking (ie, 20%), but is also longer than any values obtained with BWS. The longer total double-limb support time observed in the FWB control sessions during treadmill locomotion is in agreement with previously reported results and is thought [TABULAR DATA OMITTED]

to reflect a greater need for balance on a moving surface. [20] Eke-Okoro and Larsson [2] also found an increased total double-limb support time, perhaps resulting from a need to compensate for the uneven loading, and subsequently a greater balance demand. With BWS, the subjects required less balance control to walk supported on a treadmill, as reflected in the reduced total double-limb support time. Further investigation of this relationship is needed.

The decreased total double-limb support time and increased single-limb support time could have implications for balance training, as a major problem in neurological gait is the lack of balance control. [5] During the BWS trials, the subjects were forced to support their body weight (albeit less body weight) on a single limb for longer periods of time. This finding stresses the balance component of gait and thus can be incorporated during training.

By providing BWS during treadmill walking, both balance and locomotion are simultaneously being retrained, rather than separately addressed as in conventional physical therapy practice. [5] Furthermore, Eke-Okoro and Larsson [2] showed that the removal of weight from nondisabled subjects, who were previously weighted on one side (simulating a neurological condition), corrected their abnormal temporal characteristics. The BWS system may help normalize the gait of patients with neurological impairment in a similar fashion.

The fact that the subjects could not walk on the treadmill at the same speed at the different BWS levels as during FWB could partly be related to the progressively raised greater trochanter. During normal walking at a comfortable speed, the center of gravity describes a smooth vertical displacement of 5 cm, the lowest point occurring when both feet are on the ground. The harness could limit the vertical displacement of the body, which is related to the decreased contact distance. As a result, the walking speed is perceived as being faster and the maximal comfortable treadmill speed must be reduced. Hewes et al [13] also reported that their subjects, in a simulated lunar gravity situation without any change in their center-of-gravity movement, walked 60% slower than normal. Margaria and Cavagna [21] hypothesized that subjects walking under reduced gravity conditions would walk slower because of a decrease in energy expenditure at push-off. This hypothesis is supported by the decreased gastrocnemius muscle mean burst amplitude during stance observed in this study.

Most patients with neurological impairment lack an adequate push-off secondary to abnormal muscle activation. [6] Thus, both slower walking speeds and a reduced push-off demand need to be considered in developing a training strategy.

Hip and Knee Angle

Displacement

Although the angular displacement pattern for the hip and knee remained similar for both FWB and BWS trials, the amplitude of movement decreased with BWS. The consistent pattern and amplitude of movement demonstrated during the FWB trials compare favorably with the results of Smidt, [22] Murray et al, [23] and Winter. [24] Contrary to our results, however, Smidt [22] observed a decrease in the total hip movement, with a reduction in walking speed from 1.34 to 0.91 [m.s.sup.-1] in subjects walking over ground. Murray et al [23] also reported a decrease in hip displacement at heel-off for subjects walking at a comfortable speed on a treadmill, as compared with over-ground walking.

With BWS, the decrease in the hip and knee angles at foot-flat and at each respective MSA may be the result of harness constraints on the center of gravity. As previously discussed, the excursion of the center of gravity was limited, and each subject's body under BWS conditions (particularly at 50% and 70% of BWS) was prevented from moving downward at foot-flat by the harness. The resultant decrease in knee flexion at foot-flat, however, reinforces the findings of Hewes et al. [13] Their subjects showed a decrease in knee flexion at foot-flat, which the authors attributed to the decreased amount of weight supported and not to the harness constraints. The subjects in this experiment not only supported less weight, but their bodies were also prevented from descending at fool-flat. Thus, knee flexion at foot-flat was probably reduced by both factors. Another contributing factor should be considered. There may be a decrease in the transfer of kinetic energy from the reduced push-off and thus a decreased energy transfer and subsequent decrease in swing momentum and displacement. Patients with neurological impairment are often unable to control the energy transfter from the unaffected limb to affected limb. [25] These patients, however, would be able to control this transfer if it was graded to their ability using the BWS system. In addition, because most patients with acute neurological conditions lack the necessary amplitude of active joint movement, a system that progressively stimulates an increase in movement could be considered beneficial in gait retraining.

Electromyographic Timing

Despite variation, the EMG timing of flexor and extensor muscles appeared similar across walking speeds in FWB control trials. This timing demonstrated a link to the events of the gait cycle that is similar to that reported for nondisabled subjects at a comfortable walking speed [26,27] or a slower cadence. [28] The phase relationship between heel-strike and the onset of EMG activity for the subjects walking on a treadmill seems as consistent as that previously reported by Medeiros. [11]

A relationship between cycle time and lower-limb extensor muscle duration has been reported. [11, 29]31] Because the cycle time across the BWS trials in our study was not significantly different

[TABULAR DATA OMITTED]

from that for the FWB control trials, it is conceivable that the EMG timing would remain similar between the two experimental conditions. The minor increases in vastus lateralis and gluteus medius muscle off timing under both conditions may be related to the longer positive total double-limb support moment required with the increased cycle time observed with slower walking speeds.

Body weight support did have some effect on the timing of activation of the back muscles studied. Thorstensson et al [32] demonstrated that small changes in a trunk flexion (2[degrees]-4[degrees]) could cause a change in erector spinae muscle activity. Thus, the changes in timing of the erector spinae muscles may be related to modifications required to control trunk movement while supported in a harness over the treadmill. The erector spinae muscles appear sensitive to equilibrium changes caused by the removal of body weight or the stabilizing support of the harness. Furthermore, a relationship between swing duration and muscle timing may account for increased medial hamstring muscle burst duration.

Timing of events, especially that of muscle activation, is an important factor to consider in gait retraining. Examination of EMG activity in the muscles of patients with hemiplegia revealed a mixed, unpredictable pattern of timing. In most instances, abnormalities appeared related to loading on the limb, especially early gastrocnemius muscle activation. [1] Electromyographic timing was not altered with BWS, however, possibly because of a decreased load (BWS), rate of application (slow treadmill speed), and duration of load (decreased percentage of total double-limb support time). The proper training of muscles (eg, gastrocnemius muscles) may be facilitated with BWS training, but further investigation of this hypothesis is needed.

Mean Electromyographic Burst

Amplitudes

The changes in EMG amplitude attributable to the effect of walking speed in the FWB control sessions in our study are comparable with those reported by Yang and Winter. [28] The effects of walking speed on EMG amplitude were more evident at the hip than at the ankle muscles. The proportional changes in joint acceleration and deceleration with decreasing walking speed reported by Yang and Winter would thus be expected to modify the hip and knee muscle EMG amplitudes more than those at the ankle.

The normalized mean EMG burst amplitudes of the erector spinae, gluteus medius, vastus lateralis, and gastrocnemius muscles decreased with increasing BWS. As the body weight supported by the lower limbs was reduced, the muscles active during stance, especially during the hell-strike and push-off phases, demonstrated a decrease in burst amplitude.

It has previously been proposed that the erector spinae muscle decreases the forward and lateral motion of the trunk during the heel-strike and push-off phases of the gait cycle, respectively. [27, 32] Thus, a smaller forward braking reaction would be required with decreasing body weight, as illustrated by the gradual decrease inthe first EMG burst amplitude. The second burst amplitude, which was consistently larger than the first burst amplitude, however, was less affected by BWS in our study. Thorstensson and colleagues [27, 32] have suggested that, when the second EMG burst amplitude is larger than the first burst amplitude, this is usually the result of a change from a frontal to a more sagittal plane of movement. Consequently, it seems that the removal of body weight did not decrease the need for control of lateral trunk movement. It was observed that the harness, in combination with body weight removal, may have been related to an increase in lateral trunk movement consistent with the continued second burst of erector spinae muscle activity.

The gluteus medius muscle, a postural muscle needed for stance stability at heel-strike, demonstrated a similar modification to that of the erector spinae muscle. The need for pelvic control relative to the lower-limb movement would be expected to be less because of the support system. This hypothesis may be supported by the significant decrease in gluteus medius muscle amplitude across BWS trials as compared with the more consistent amplitude across the FWB control trials.

The vastus lateralis muscle is normally active during weight acceptance in early stance. The decrease in the knee extensor activity with BWS could be explained by the decreased amount of body weight and knee flexion angle at foot-flat. The removal of body weight was also influential at push-off, when ankle plantar flexion was activated to propel the body forward. The activity in the medial gastrocnemius muscle can be influenced by the duration and magnitude of the applied force, such as body weight. [30,33] As a consequence of increasing BWS, gastrocnemius muscle EMG amplitude decreased, as seen in animal experiments. [33,34] In addition, increasing BWS may decrease the stretch on the gastrocnemius muscle during stance, which may facilitate the proper phasing of this muscle for patients involved in a gait retraining program. This effect would be important for patients with neurological impairment, because early stretch activation in the triceps surae muscle is a common problem interfering with locomotion. The removal of weight also seems to affect muscles during the swing phase, as can be seen by the increased amplitude in the tibialis anterior muscle during the swing phase. This effect would also be important, because lack of dorsiflexion is another common problem in the locomotor pattern of most patients with neurological impairment.

It has been reported that the amount and timing of loading can have a strong influence on the locomotor pattern in animals. [34] For example, flexor muscle activity will decrease if the loading on the extensor muscle is high. [34] The unloading at push-off is critical for initiation of swing. Thus, the location of the body weight, as demonstrated in previous studies, [12,34] may have a more important effect in changing the gait pattern than the amount of body weight increased [11] or removed. If this hypothesis is correct, then the amount of body weight and its placement could be used as an important factor in gait retraining strategies. The removal of body weight would be most influential at heel-strike and toe-off. The unloading of the limbs occurring at toe-off could facilitate the activation of flexor muscles [29] in patients unable to control the unloading of their limbs. Training with BWS could facilitate not only flexion but also the gradual strengthening of extensor muscles during stance as BWS is decreased. Furthermore, an emphasis could be placed on the strengthening of muscle activity, at the appropriate joint angular position during gait, as suggested by Olney et al. [25]

An interactive BWS gait retraining strategy would allow patients lacking muscle control to develop the strength and coordination required to walk. Any BWS level below 70% can be beneficial for training, if the treadmill speed is adjusted to the patient's walking abilities and BWS condition. The amount of BWS can then be decreased and walking speed increased as the patient improves.

Summary and Conclusions

Normal gait under the influence of various BWS levels was compared with FWB gait to determine whether a BWS and treadmill stimulation strategy could be used in gait retraining. Adaptations to BWS were few and related to two factors.

First, modifications attributed directly to the removal of body weight were found. There was a reduced mean burst amplitude in muscles required for weight acceptance (ie, erector spinae and gluteus medius muscles) and push-off (ie, medial gastrocnemius muscle) and an increase in mean burst amplitude in the investigated muscle that is active during swing (ie, tibialis anterior muscle).

Second, alterations related to mechanical constraints of the BWS system and the dictated walking speed were revealed. These alterations were a raised center of gravity, leading to limited downward excursion, decreased percentage of stance, decreased total double-limb support time, decreased hip and knee angular displacement, and increased single-limb support time. These results led to the development of a BWS treadmill training scheme that may offer many clinical applications.

The training method used in this study did not produce abnormal gait. Furthermore, this interactive training protocol should provide an easier progression from stance to the swing phase of gait. Body-weight-supported gait appears to present temporal, kinematic, and EMG patterns that could be advantageous in gait retraining.

(*) Industries Auteca Ltd, 3125 Bernard Pilon, St Mathieu de Balocil, Quebec, Canada J3G 4F5.

(+) Nortel Manufacturing Ltd, 2000 Ellesmere Rd, Scarborough, Ontario, Canada N1H 2W4.

(++) Graphic Controls Canada Ltd, 215 Herbert St, PO Box 5500, Gananoque, Ontario, Canada K7G 2X1.

([section]) Centre de Recherche en Sciences Neurologiques de l'Universite de Montreal, Montreal, Quebec, Canada H3C 3J7.

(+++) Honeywell, Test Instruments Div, 740 Ellesmere Rd, Scarborough, Ontario, Canada N1H 2W4.

(#) Digital Equipment Corp, 100 Herzberg Rd, PO Box 1300, Kanata, Ontario, Canada K2K 2A6.

(**) Panasonic, 8270 Mayrand St, Montreal, Quebec, Canada H4P 2C5.

(1) Intertechnology Inc, 3675 Boul de Sources, DDO, Quebec, Canada H9B 2T6.

[TABULAR DATA OMITTED]

References

[1] Knutsson E, Richards C. Different types of disturbed motor control in gait of hemiparetic patients. Brain. 1979;102:405-430.

[2] Eke-Okoro ST, Larsson L. A comparison of the gaits of paretic patients with the gait of control subjects carrying a load. Scand J Rehabil Med. 1984;16:151-158.

[3] Carlsoo S, Dahllof AG, Holm J. Kinematic analysis of the gait in patients with hemiparesis and in patients with intermittent claudication. Scand J Rehabil Med. 1974;6:166-179.

[4] Dickstein R, Nissan M, Pillar T, Scheer D. Foot-ground pressure pattern of standing hemiplegic subjects: major characteristics and patterns of improvements. Phys Ther. 1984; 64:19-23.

[5] Winstein CJ, Gardner ER, McNeal DR. Standing balance training: effect on balance and locomotion in hemiparetic adults. Arch Phys Med Rehabil. 1989;70:755-762.

[6] Dietz V, Quintern S, Berger W. Electrophysiological studies of gait in spastic and rigidity. Brain. 1981;104:431-449.

[7] Rossignol S, Barbeau H, Julien C. Locomotion of the adult chronic spinal cat and its modifications by monoaminergic agonists and antagonists. In: Golberger M, Gorio M, Murray M, eds. Development and Plasticity of the Mammalian Spinal Cord. Padua, Italy: Liviana Editrice SpAt; 1986:323-346.

[8] Barbeau H, Rossignol S. Recovery of locomotion after chronic spinalization in the adult cat. Brain Res. 1967;412:84-95.

[9] Finch L, Barbeau H. Influences of partial weight bearing on normal human gait: the development of a gait retraining strategy. Can J Neurol Sci. 1985;12:183. Abstract.

[10] Ghori GM, Luckwill RG. Responses of the lower limb to load carrying in walking man. Eur J Appl Physiol. 1985;54:145-150.

[11] Medeiros JM. Investigation of Neuronal Mechanisms Underlying Human Locomotion: an EMG Analysis. Iowa City, Iowa: The University of Iowa; 1978. Doctoral dissertation.

[12] Neumann DA, Cook TM. Effect of load and carrying position on the electromyographic activity of the gluteus medius muscle during walking. Phys Ther. 1985;65:305-311.

[13] Hewes DE, Spady AA, Harris RL. Comparative Measurement of Man's Walking and Running Gaits in Earth and Simulated Lunar Gravity. NASA report TDN-3363. 1967:1-33.

[14] Barbeau H, Wainberg M, Finch L. Description and application of a systemfor locomotion rehabilitation. Med Biol Eng Comput. 1987;25:341-344.

[15] Visintin M, Barbeau H. The effects of body weight support on the locomotor pattern of spastic paretic patients. Can J Neurol Sci. 1989;16:315-325.

[16] Visintin M, Finch L, Barbeau H. Progressive weight bearing and treadmill stimulation during gait retraining of hemiplegics: a case study. Phys Ther. 1988;68:807. Abstract.

[17] Zomlefer MR, Provencher J, Blanchette G, Rossignol S. Electromyographic study of lumbar muscles in acute high decerebrate and in low spinal cats. Brain Res. 1984;290:249-260.

[18] Arsenault AB, Winter DA, Marteniuk RG. Characteristics of muscular function and adaptation in gait: a literature review. Physiotherapy Canada. 1987;39:5-12.

[19] Snedecor GW, Cochran WG. Statistical Method. 7th ed. Ames, Iowa: Iowa State University Press; 1982.

[20] Murray MP, Spurr GB, Gardner GM, Mollinger LA. Treadmill vs floor walking: kinematic electromyogram and heart rate. J Appl Physiol. 1985;59:87-91.

[21] Margaria R, Cavagna GA. Human locomotion in subgravity. Aerospace Medicine. 1966;35:1140-1146.

[22] Smidt GL. Hip motion and related factors in walking. Phys Ther. 1971;51:9-21.

[23] Murray MP, Kory RC, Clarkson BH, Sepic SB. A comparison of free and fast speed walking pattern in normal men. Am J Phys Med. 1966;45:8-24.

[24] Winter DA. Biomedical motor patterns in normal walking. Journal of Motor Behavior. 1983;15:302-330.

[25] Olney SJ, Jackson VG, George SR. Gait reeducation guidelines for strike hemiplegic using mechanical energy and power analysis. Physiotherapy Canada. 1987;40:242-248.

[26] Battye CK, Joseph J. An investigation by telemetry of the activity of some muscles in walking. Med Biol Eng. 1966;4:125-135.

[27] thorstensson A, Carlson H, Zomlefer MR, Nilsson J. Lumbar back muscle activity in relation to trunk movement during locomotion in man. Acta Physiol Scand. 1982;116:13-20.

[28] Yang JF, Winter DA. Surface EMG profiles during different walking cadences in humans. Electroencephalogr Clin Neurophysiol. 1985;60:485-491.

[29] Herman R, Weiter Z, Brampton S, Finley FR. Human solution for locomotion, I: single limb analysis. In: Herman R, Grillner S, Stein P, Stuart DG, eds. Neural Control of Locomotion. New York, NY: Plenum Publishing Corp; 1976:413-450.

[30] Grillner S. Control of locomotion in bipeds, tetrapods and fish. In: Handbook of Physiology, II: The Nervous System. New York, NY: American Physiological Society; 1981:1179-1234.

[31] Pearson K. The control of walking. Sci Am. 1976;235:72-86.

[32] Thorstensson A, Nilsson J, Carlson H, Zomlefer MR. Trunk movements in human locomotion. Acta Physiol Scand. 1984;121:9-22.

[33] Monster A. Loading reflexes during two types of voluntary muscle contraction. In: Herman R, Grillner S, Stein R, Stuart DG, eds. Neural Control of Locomotion. New York, NY: Plenum Publishing Corp; 1976:347-361.

[34] Duysens J, Pearson KG. Inhibition of flexor burst generation by loading ankle extensor muscles in walking cats. Brain Res. 1980; 187:321-332.

L Finch, MSc(Rehab), is Neuro-coordinator, Department of Physiotherapy, Montreal Neurological Institute.

H Barbeau, PhD, PT, is Associate Professor, School of Physical and Occupational Therapy, McGill University, 3654 Drummond St, Montreal, Quebec, Canada H3G 1Y5. He is also Chercheur Boursier of the Fonds de la Recherche en Sante du Quebec. Address all correspondence to Dr Barbeau.

B Arsenault, PhD, PT, is Associate Professor, Ecole de Readaption, Faculte de Medicine, Universite de Montreal, and Research Center, Montreal Rehabilitation Institute, 6300 Darlington Ave, Montreal, Quebec, Canada H3S 2J4.

This work was supported by the Medical Research Council of Canada and the Fonds de la Recherche en Sante du Quebec.