Partial body weight support with treadmill locomotion to improve gait after incomplete spinal cord injury: a single-subject experimental design.

Key Words: Gait, Partial body weight support, Single-subject experimental design, Stepping mechanism, Treadmill.

Incomplete spinal cord injury often results in gait patterns that are associated with a poor capacity to bear weight through the lower extremities and in an altered swing phase.[1-3] Other neurologic impairments also can cause gait deviations due to inadequate weight acceptance by the lower extremities during the stance phase and due to decreased flexion of the hip, knee, or ankle during the swing phase.[4-8] Deficits in weight acceptance, single-limb stance, and limb advancement are secondary to impairments in strength and motor control associated with neurological damage. These deficits are often severe enough to cause a delay in the initiation of gait training in the upright position. Gait training for persons with such deficits often focuses on training them to bear weight, shift weight, and balance as isolated tasks before these tasks are incorporated into locomotion. Gait deviations, however, often persist following conventional gait training.[5] Finch and Barbeau[9] proposed an alternative method of gait training for persons with neurological impairments, a method that allows simultaneous retraining of the various components of gait during actual locomotion and that theoretically facilitates the expression of a more normal gait pattern. During this training, the person's body weight is partially supported by an overhead harness as he or she walks on a treadmill at his or her maximum comfortable speed.

The rationale for this training is, at least in part, based on research on spinal mechanisms in animals. The generation of cyclic locomotor patterns can be attributed in some animals to rhythmic neural activity produced by central pattern generator (CPG) networks in the brain stem or the spinal cord, or both.[10,11] Grillner[12] demonstrated that kittens whose spinal cord was transected at the low thoracic level 1 week after birth were able to walk 2 days later when their hind limbs were held on a treadmill. Smith[13] studied cats with T-12 lesions and found the stepping mechanism to be well preserved 5 to 6 months postsurgery when the cats were supported while walking on a treadmill. Rossignol et al[14] found progressive recovery of locomotion in spinalized adult cats after weeks of supported walking on a treadmill. Barbeau and Rossignol[15] demonstrated the recovery of near-normal locomotor patterns in cats spinalized (T-13) as adults following an "interactive locomotor training" program. This program consisted of appropriately graded weight support during treadmill locomotion. Finch and Barbeau[9] proposed that the postural stability and balance required for gait in humans may be gained by a strategy of partial to full weight bearing in combination with training on the treadmill aimed at recovery of the stepping mechanism.

Partial body weight support (PBWS) during locomotion on a treadmill has been used in the treatment of persons with neurologic deficits[16-18] and other clinical conditions such as lower-extremity fractures,[19] osteoarthritis,[20] and lower-extremity amputations.[21] Improvements in temporal-distance gait variables have been shown during locomotion with PBWS in persons with hemiplegia,[16] spinal cord injuries,[17] and cerebral palsy.[18] These variables include gait speed, cadence, and stride length in addition to percentage of time spent in the stance and swing phases of gait. Pillar et al[16] studied subjects with hemiplegia as they ambulated with PBWS on a walkway located between parallel bars. The results indicated that gait speed during ambulation with PBWS increased an average of 70% compared with gait speed during ambulation with full weight bearing. The percentage of time spent by the involved lower extremity during the stance phase of gait increased, and the percentage of time spent on the involved lower extremity during the swing phase decreased during ambulation with PBWS. Visintin and Barbeau[17] found that people with spasticity (eg, presence of clonus and stiff lower-limb movements, with decreased angular joint excursions at 0% of body weight support) resulting from spinal cord injuries demonstrated more normal gait patterns during treadmill locomotion when a percentage of their body weight was supported. The subjects had increased walking speed, stride length, and single-limb support time and decreased double-limb support time. Kinematic data indicated improved trunk and knee alignment during stance. Subjects who walked with trunk and knee flexion at 0% of body weight support showed trunk alignment at or near neutral (0 [degrees]) and greater knee extension (35 [degrees]-15 [degrees] of flexion) during stance at 40% of body weight support. Van Faassen and Molen[22] concurred that when subjects with neurological impairments ambulated with only a partial load, their stepping was facilitated and functional training of the involved lower extremities could be enhanced.

The theoretical basis for locomotion with PBWS is well established,[23] and the preliminary results using this approach in humans[16,17,22] are promising. Only the immediate effects of gait performance during treadmill locomotion with PBWS have been reported. Whether this method of gait training will have both short- and long-term carryover to walking off a treadmill has not been studied. Transfer of training from a treadmill to regular ambulation cannot be assumed. For example, when walking on a treadmill, push-off is assisted by movement of the treadmill belt and less planter-flexor activity is required. The passive support provided by the harness during treadmill locomotion with PBWS also may prevent the person from maintaining the postural support and balance responses that are needed during unsupported locomotion. Moreover, the difference in the amount of weight born by the lower extremities during ambulation with PBWS versus unsupported ambulation could result in learning a motor pattern that is different from that used during unsupported ambulation. Finally, the visual input from the environment is different in the two situations. During treadmill locomotion with PBWS, the ground moves relative to the subject and the visual surround is stable. During real-world locomotion, the subject moves relative to both the ground and the visual surround, thus generating a radially expanding optic flow pattern.

Our approach to answering the question of whether gait training on a treadmill with PBWS would have carryover to improve gait performance off the treadmill was to conduct an in-depth study, over several months, using a single-subject experimental design. Our hypothesis was that gait training using a treadmill and PBWS would result in improvements in the subject's gait performance off the treadmill for gait speed, cadence, stride length, and percentages of stance and swing for both lower extremities, both during treatment and during follow-up periods.

Method

Research Design

A single-subject experimental design with an A-B-A strategy was used.[24] Phase Al was the baseline phase (6 weeks), phase B was the treatment phase (6 weeks), and phase AII was the treatment-withdrawal phase (3 weeks).

Subject History

Because most of the experiments that formed the basis for the intervention used in this study were performed on spinalized animals, a subject with an incomplete spinal cord injury was recruited. To limit the potential influence of spontaneous recovery, selection criteria included that the subject be greater than 6 months postinjury and discharged from regular physical therapy due to lack of improvements for greater than 1 month.

The subject was a 28-year-old man who sustained a traumatic hyperflexion injury to the cervical spine, with subluxation of C-5 on C-6 and resultant incomplete quadriplegia, 7 months prior to participation in the study. Four days postinjury, the subject was placed in a halo brace and underwent posterior fusion of C-5 and C-6. Initially, the subject's bilateral upper-extremity strength, as determined by manual muscle testing,[25] was 4 for shoulder flexion, 4 to 4+ for elbow flexion and extension; 3+ to 4+ for wrist flexion and extension; and 0 to 1 for finger flexion and extension, except for 2- for the left extensor indices. The subject's bilateral lower-extremity strength was 2- to 2 for hip flexion; 2- for hip abduction, knee flexion, and ankle planter flexion; 4 for left knee extension and 4+ for right knee extension; and 1 for ankle dorsiflexion and great toe extension. The measurements obtained by manual muscle testing have been shown to be reliable.[26,27] The subject was dependent in all aspects of functional mobility.

After 10 days in an acute care hospital, he received 2 months of inpatient rehabilitation. He was discharged to his own apartment, and he was able to ambulate independently for short distances on level surfaces indoors with use of Loftstrand crutches and a custom-molded hinged ankle-foot orthosis for his right lower extremity. He continued outpatient physical therapy for 3 months.

The subject was given a home program and was discharged from physical therapy approximately 6 weeks prior to the start of this study. Before his injury, the subject had been active in many sports, including baseball, basketball, tennis, football, and golf.

At the start of the study (7 months postinjury), the evaluation of the subject's physical status was as follows. His sensation was intact bilaterally for both the upper and lower extremities. The strength of both upper extremities was graded 4+ overall. Lower-extremity strength was graded 5 on the left side and 4 on the right side except for the strength of the iliopsoas and quadriceps femoris muscles, which was graded 5. Range of motion was within normal limits bilaterally for both the upper and lower extremities, except for -5 degrees (from neutral) of ankle dorsiflexion on the right side, neutral ankle dorsiflexion on the left side, and neutral right calcaneal eversion. Hip medial (internal) rotation was 0 to 18 degrees on the right side and 0 to 25 degrees on the left side. Straight leg raise was 0 to 45 degrees on the right side and 0 to 40 degrees on the left side. A Thomas test showed -10 degrees of extension on the right side. His tone (de, velocity-dependent resistance to passive stretch of a muscle) was assessed via passive range of motion of the upper and lower extremities bilaterally and was, in our view, within normal limits on the left side and increased on the right side (with mild resistance in response to passive movement). There was intermittent right ankle clonus with weight bearing. When he was asked to demonstrate isolated ankle dorsiflexion bilaterally, there was calcaneal inversion with forefoot supination while attempting to dorsiflex the right ankle.

He was able to ambulate independently on level surfaces, ramps, and curbs without an assistive device or orthosis for 45.7 m (150 ft) without resting. Many gait deviations were evident, including delayed initiation and prolonged duration of right lower-extremity swing, absent push-off, decreased stance time on the right lower extremity, and a decreased cadence as compared with normal locomotion.[28] He was able to ascend and descend eight stairs with reciprocal motion of the lower extremities without using a rail or device, but he did so slowly and with mild dyscoordination. Although he was "functionally independent" in all locomotor activities, his gait remained markedly impaired in comparison with his gait before injury. He was unable to "keep up" with friends while walking, and he was unable to run or participate in any of his prior sports activities, with the exception of an occasional round of golf. Prior to participation in the study, the subject read and signed an institutionally approved informed consent form.

Weight Relief System

Gait training with PBWS required the following equipment: (1) a weight relief system,(*) (2) a Challenger 3.0 treadmill,[dagger] (3) a Gravity Lumbar Reduction (GLR) vest,[double dagger] (4) free weights, and (5) a scale. The subject was supported in the GLR vest by attachment to an overhead pulley and free weight system. The GLR vest was chosen because it supported the subject primarily at the upper abdomen and ribs and appeared to not interfere with lower-limb and pelvic movements. Figure 1 shows a subject without impairments related to ambulation in the experimental setup.

[Figure 1 ILLUSTRATION OMITTED]

Determination of Body Weight Support

Prior to initiation of the research, two pilot sessions (10 minutes each) of ambulation on the treadmill with PBWS were conducted as follows to determine the amount of body weight support to be used during the intervention phase. The subject weighed 91 kg and began ambulation on the treadmill with 40% of his body weight supported by the harness. This support was selected as the initial level of weight relief based on the study by Visintin and Barbeau,[17] who observed a loss of heel-ground contact in some subjects when greater than 40% of their body weight was supported by a harness. Our subject was unable to achieve heel-ground contact during ambulation on the treadmill with 40% of his body weight supported by the harness. The level of support, therefore, was reduced until the subject achieved heel-ground contact bilaterally for 10 consecutive steps. The final level of body weight support was 32% (29.5 kg), which then was used throughout the study to unweight the subject.

Gait Measures

Although observational gait analysis (OGA) is used widely for clinical assessment of gait performance, it generally has shown low reliability when tested[29] and data obtained in this manner cannot be quantified easily.[30] In contrast, some temporal-distance measurements of gait performance have been reported as reliable[6,31,32] and appear to correlate with functional ambulation status.[33] These measurements can now be obtained with minimal time, space, and equipment. More complex gait analysis systems utilizing electromyographic, kinetic, and kinematic measures, although ideal in many ways, would have been impractical for our study due to the frequent measurement required by single-subject design experiments. For these reasons, temporal-distance measures were used to assess gait performance during comfortable walking, fast walking, and running.

We used a Footswitch Stride Analyzer system[sections] to obtain temporal-distance gait measurements. The subject placed a shoe insert that contained four footswitches (heel, first and fifth metatarsal heads, great toe) into each shoe prior to ambulation. A wire from each insert connected the footswitches to a recorder located in a belt-pack worn around the subject's waist. The subject wore a light-sensitive switch on his left arm. As he ambulated, lights at the beginning and end of the walkway activated the recorder to start and stop data collection from the footswitches. After each trial, data were downloaded from the recorder to an IBM personal computer[parallel] for later analysis. The total weight of the testing equipment worn by the subject was 1.06 kg. The walkway was 10 m long for both comfortable and fast walking and 20 m long for running. Data were collected only during the middle 6 m for walking trials and the middle 12 m for running trials to eliminate speed differences at the beginning and end of the walkway.

Five temporal-distance gait measurements were obtained from the computer analysis: (1) gait speed (in meters per second), (2) cadence (in steps per minute), (3) stride length (in meters), defined as the average distance between two consecutive ipsilateral heel contacts, (4) percentage of stance, defined as the average percentage of the gait cycle (the interval between successive ipsilateral foot contacts) when the foot was in contact with the floor, and (5) percentage of swing, defined as the average percentage of the gait cycle when the foot was off the floor. Percentages of stance and swing were measured seven times for both the right and left lower extremities.

Footswitch Stride Analyzer measurement reliability. The reliability of gait measurements taken from the Footswitch Stride Analyzer was assessed by one investigator, who tested four subjects during comfortable walking, fast walking, and running (two sessions at each speed). Results were evaluated using the intraclass correlation coefficient (ICC[3,1]), which was calculated for each temporal-distance variable at each speed. The ICC values for gait speed were .81 for comfortable walking, .90 for fast walking, and .95 for running. The ICC values for each of the remaining variables at each speed ranged from .73 to .96.

Procedure

During the study, the subject continued his home exercise program of hamstring muscle and heel cord stretching and a program of upper-extremity weight training and stationary bicycle use (1-2 times a week), but he agreed to refrain from starting any new exercise programs until the study was completed.

Experimental Protocol

The subject participated in the study for a total of 15 weeks. Temporal-distance data were collected during forward locomotion on a level corridor surface at three different speeds: (1) comfortable walking, defined as a self-selected comfortable pace, (2) fast walking, defined as the fastest possible walking pace, and (3) running as fast as possible. Three trials at each speed were performed consecutively, with a 2-minute rest between trials. The order of testing (comfortable walking, fast walking, running) was the same for each session so that possible effects of fatigue or practice were constant across sessions. During phase AI (6 weeks), the subject was seen one time per week for collection of baseline temporal-distance gait data only.

During phase B (6 weeks), the subject trained three times per week using the treadmill-PBWS system as described later. Once a week, the subject was tested before and after the treadmill training to assess immediate (pretreatment versus posttreatment) and short-term carryover (weekly) effects of the treadmill-PBWS training. During phase B, a 10-minute rest was provided between the initial temporal-distance gait test and the treadmill-PBWS training and between the treadmill-PBWS training and the final temporal-distance gait test. All testing and treatment sessions occurred at the same time of day throughout the study.

Phase All was shortened from 6 weeks to 3 weeks due to a change in the subject's availability for testing. As in phase Al, the subject was seen and tested once per week only for collection of temporal-distance data to determine any effect of treatment withdrawal.

Training Protocol

Gait training sessions on the treadmill during phase B were 20 minutes in duration (followed by a 5-minute cool-down period on the treadmill) and were conducted 3 days per week for 6 weeks. There were 16 training sessions, however, because the subject missed 2 sessions. Heart rate was used to determine whether a training effect that could influence cardiovascular status occurred during gait training. A one-lead electrocardiograph electrode was applied ([V.sub.2] placement[34]) prior to each gait training session. Heart rate was monitored only during treadmill training using a Cardiac Function Evaluation machine.(#) Heart rate data were collected for 15 seconds every 4 minutes using an electrocardiographic printout from the machine. For safety during ambulation, the subject was monitored by a physical therapist and a physical therapy volunteer.

A continuous, progressive-load, fixed-grade method was used to set the time and intensity of treadmill training (Tab. 1).[35] We believed that this protocol was necessary to accommodate the subject's motor impairment and make the training feasible. The treadmill slope remained at 0% throughout the study. The initial treadmill speed was set at 1.5 mph, based on the subject's ability to walk without "scuffing" the right foot at 1.5 mph during the pilot treadmill test. The treadmill speed was increased every 4 minutes until the preset maximum for that session was reached. The maximum speed for session 1 was set at 3.0 mph, based on the subject's mean comfortable walking speed of 2.7 mph during phase Al. For subsequent sessions, the maximum speed was increased by 0.5-mph increments when the subject was able to ambulate with zero scuffs of the involved right foot for 10 consecutive steps during the maximum speed in the prior session. The subject then was given three training sessions to accommodate to ambulation at the increased maximum speed before further increases were attempted. If performance deteriorated during a session, the maximum speed was reduced by 0.5 mph on the next session. The maximum speed attained during the training phase was 4.5 mph.

Data Analysis

The mean value of three trials for each temporal-distance measure at each of the three speeds (comfortable walking, fast walking, running) for each test session was used in the analysis of the data. Pretreatment data only for phase B were used in the A-B-A comparisons. A separate analysis was performed to compare pretreatment and posttreatment values for phase B.

Between-phase comparison (AI-B-AII). Results were analyzed separately for each gait speed using the visual analysis method[36] and the 2-standard deviation band method (2-SDBM).[36,37] For visual analysis, the mean values for each gait measure for each session were plotted and compared across phases for any visible trend in performance pattern or magnitude. For the 2-SDBM analysis, the mean and 2 standard deviations for each temporal-distance variable were calculated for phase AI (baseline) and plotted on a graph, along with individual session values for phase B (intervention). If two or more successive data points in phase B fell outside the two-standard deviation band from phase AI, the changes from phase AI to phase B were considered significant. The same method was used to compare phase B with phase AII. Because the 2-SDBM analysis should not be used with a data series that has a significant autocorrelation coefficient, we evaluated serial dependency in our data using Bartlett's test prior to applying the 2-SDBM analysis. If Bartlett's test was significant, a first-difference transformation procedure was used to reduce the serial dependency prior to applying the 2-SDBM analysis.

Within-phase comparison (phase B). Performance during this treatment phase was analyzed by visual analysis only, which compared pretreatment and posttreatment measurements. Two characteristics were evaluated: (1) within-session pretreatment and posttreatment treadmill training values (an index of immediate treatment effects) and (2) the trend in the pretreatment values over the 6 weeks (an index of short-term carryover effects).

Results

Gait Performance

Between-phase comparison (AI-B-AII). Overall, visual analysis of the data comparing phase AI (baseline) with phase B (intervention) indicated improvements in all measures of gait performance during comfortable walking and fast walking except for percentages of stance and swing for the left lower extremity. Further changes in magnitude of performance for all variables during running were seen when comparing phase AI (baseline) with phase B (intervention). Improvements from phase B to phase AII (follow-up) were observed, as follows: (1) During comfortable walking, improvements were seen in gait speed, cadence, and percentages of stance and swing for the right lower extremity, (2) during fast walking, improvements were seen in gait speed, cadence, and stride length, and (3) during running, improvements were seen in gait speed and percentages of stance and swing for the right lower extremity. Table 2 lists composite means and standard deviations of each phase for all variables and speeds of locomotion.

Table 2.
Composite Means and Standard Deviations of Gait Variables for
Three Locomotor Speeds by Phase (AI, B [Pretreatment], AII)

                          Comfortable
                            Walking           Fast Walking

Gait Variable           [bar] X      SD      [bar] X     SD

Gait speed (m/s)
AI                      1.22        0.07     1.63       0.04
B                       1.36        0.04(a)  1.76       0.06(a)
AII                     1.37        0.04     1.80       0.04

Cadence
(steps/min)
AI                      103.1       3.3      117.2      7.3
B                       108.3       1.7      125.4      3.9
AII                     109.2       2.5      127.6      3.6

Stride length (m)
AI                      1.42        0.06     1.67       0.10
B                       1.51        0.03     1.69       0.03
AII                     1.51        0.02     1.70       0.03

Percentage of
right stance
AI                      61.3        0.6      61.4       0.6
B                       61.0        1.0      61.3       0.9
AII                     60.6        0.8      61.5       0.6

Percentage of
right swing
AI                      38.7        0.6      38.6       0.6
B                       39.0        1.0      38.7       0.9
AII                     39.4        0.8      38.5       0.6

Percentage of
left stance
AI                      65.1        0.5      63.7       0.9
B                       64.6        0.6      63.7       0.5
AII                     65.4        0.5      64.6       0.9

Percentage of
left swing
AI                      34.9        0.5      36.3       0.9
B                       35.4        0.6      36.3       0.5
AII                     34.6        0.5      35.4       0.9

                             Running

Gait Variable          [bar] X       SD

Gait speed (m/s)
AI                      2.64        0.27
B                       3.21        0.19(a)
AII                     3.24        0.20

Cadence
(steps/min)
AI                      165.0       5.1
B                       170.0       2.7
AII                     169.1       2.9

Stride length (m)
AI                      1.92        0.17
B                       2.27        0.11(a)
AII                     2.07        0.69

Percentage of
right stance
AI                      34.7        1.8
B                       31.2        1.2(a)
AII                     29.6        1.0

Percentage of
right swing
AI                      65.3        1.8
B                       68.8        1.2(a)
AII                     70.4        1.0

Percentage of
left stance
AI                      36.8        3.9
B                       30.3        1.8
AII                     30.9        1.7

Percentage of
left swing
AI                      63.2        3.9
B                       69.7        1.8
AII                     69.1        1.7

(a) Changes from phase AI to phase B were significant. At least two successive data points in the intervention phase fell outside the 2-standard deviation band from the baseline phase.

Using the 2-SDBM analysis, differences between phases Al and B were found in gait speed during comfortable walking, fast walking, and running and in stride length and percentages of stance and swing for the right lower extremity during running. Figure 2 illustrates the results of the 2-SDBM analysis, which compared data from phase AI with data from phase B pretreatment sessions for variables that showed differences for comfortable walking and fast walking. Figure 3 illustrates similar data for the running condition. Autocorrelation coefficients for all temporal-distance gait variables, for all speeds of locomotion, between phases AI and B were not significant. Changes between phases B and AII were not significant for any variable or any speed, as indicated by the 2-SDBM analysis.

[Figure 2 and 3 ILLUSTRATION OMITTED]

Within-phase comparison (phase B). The immediate and short-term carryover of treatment improvements are illustrated in Figures 4 through 6. Visual analysis indicated a pattern of improvement in pretreatment and posttreatment within-session measurements during comfortable walking for gait speed and percentages of stance and swing for the right lower extremity (Fig. 4). These same variables were the only variables that showed evidence of short-term carryover effects (ie, week-to-week improvements as reflected in pretreatment measurements) during comfortable walking. During fast walking, gait speed showed a pattern of decline for pretreatment and posttreatment immediate treatment effect (Fig. 5). Gait speed and cadence, however, showed evidence of week-to-week improvements for pretreatment measurements during phase B.

[Figure 4 and 5 ILLUSTRATION OMITTED]

In the running condition, pretreatment and posttreatment within-session values showed a decline for gait speed, cadence, and stride length (Fig. 6). Little or no change between pretreatment and posttreatment values was noted for percentages of stance and swing for both lower extremities. Gait speed, cadence, and stride length values showed little short-term change during the first 3 weeks of treatment, but the values for these variables demonstrated a pattern of improvement during the last 3 weeks of treatment. Values for percentages of stance and swing for both lower extremities demonstrated week-to-week improvements after the first 2 weeks of treatment.

[FIGURE 6 ILLUSTRATION OMITTED]

Heart Rate

Table 3 displays the subject's heart rate response at rest and during treadmill-PBWS training for each session (refer to Tab. 1 for corresponding treadmill speeds). Resting heart rate decreased over the 15-week training period (Fig. 7). Training heart rates increased corresponding to the training demand of each weekly treadmill session.

Table 3.
Heart Rate (HR) Response (in Beats per Minute) of
Subject During Treadmill (TM) Training

                                Time (min)

TM
Session   0 (at                                         25
No.       Rest)   3-4   7-8   11-12   15-16   19-20   (at Rest)

1          94     104   108     115     118    120      105
2          87      94   101     101     110    120      100
3          72      76    80      84     100    104       99
4          76      96   109     109     112    120      100
5          95      96   100     104     115    120      100
6          68      90    96      98     109    118      100
7          79      91    94     101     115    127      102
8          76      97   100     104     117    128       96
9          81      89    94      96     109    120      100
10         77      90    92      98     106    130      104
11         74      94   100     108     112    140      108
12         76     102   110     120     130    140      105
13         69     100    96     109     112    120      100
14         70      94    96      92     105    105       97
15         80      96   105     120     120    120      100
16         80      96   105     118     118    120      102

[FIGURE 7 ILLUSTRATION OMITTED]

Discussion

Gait Performance

Based on the visual and statistical analyses of the data, we believe that the results indicate that the treadmill-PBWS treatment produced improvements in the subject's locomotor abilities, mainly in his running performance. We considered the improvements noted in phase B to be due to treatment, rather than to spontaneous recovery or a Hawthorne effect,[38] because (1) the results favored this judgment, (2) the magnitude of improvement during phase B (intervention) was large, especially for the running condition, and (3) the magnitude of improvement during phase AII (treatment withdrawal) diminished, although the subject retained most gains made during phase B.

Between-phase comparison (AI-B-AII). Were the effects of treadmill-PBWS training on gait performance meaningful? The changes in gait speed during comfortable and fast walking were small and, in our view, probably of minor functional significance. In making this judgment, we considered that the subject's comfortable walking speed of 1.22 m/s during phase AI was just below the value that is often considered to be normal (ie, 1.35 m/s),[33,39] thus leaving little room for improvement. Corcoran[39] found that the self-selected comfortable walking speed for subjects without impairments related to ambulation was 3.0 mph (1.35 m/s) and that this speed correlated with the lowest caloric cost of walking. The improvement in our subject's comfortable walking speed during phase B (to 1.36 m/s) thus brought him into the normal range. For fast walking, the magnitude of change from phase AI to phase B also was small (0.13 m/s).

Our subject's 0.14-m/s increase in walking speed during comfortable walking appears to be small. Changes as minor as this, however, are reported to have functional benefits for persons with less ambulatory independence than our subject had. Holden et al,[33] for example, reported gait speed values for 61 subjects with neurological impairments by functional category and noted differences between functional levels of 0.09 m/s for subjects requiring manual support of body weight versus touch support for balance, 0.14 m/s for subjects requiring touch assistance from one person versus those requiring no human assistance, and 0.26 m/s for subjects who were independent on level surfaces only versus those who were independent on all surfaces. Imms and Edholm[40] reported that an improvement in mean gait speed of 0.20 m/s was associated with a change in functional status from homebound to limited outdoor activity and that an increase of 0.33 m/s was associated with a change in functional status from limited to unlimited outdoor activity for subjects who were disabled. Brandstater et al[6] reported that an increase of 0.25 m/s correlated with a change from stage 5 to stage 6 on a motor recovery scale for persons with hemiplegia.

In contrast to modest changes in temporal-distance variables during comfortable and fast walking, changes in these variables during running were large enough, in our view, to be of benefit to the subject. For example, mean gait speed during running increased from 5.9 mph in phase AI to 7.1 mph in phase B and remained nearly the same in phase AII (ie, 7.2 mph). During phase AI, the subject's attempts to run resulted in speeds that were in the upper range of values for fast walking reported for subjects without impairments (ie, 5.0-6.0 mph),[39] but his gait speeds clearly moved out of the fast-walking range during phase B (ie, 7.1 mph) and were maintained during follow-up. Gait speed, however, remained lower than the speed of 10.3 mph recorded for an age-matched male subject without impairments who was tested under the same conditions (unpublished data). This difference, in comparison with the age-matched subject, suggests that the subject would need further gains to resume his prior sport activities, but at least he now could be expected to be able to "keep up" with his friends while ambulating at their normal speeds.

Increased running speed has been reported to result primarily from an increase in stride length rather than an increase in cadence.[41] This finding is consistent with our findings that changes in cadence during running appeared to be negligible (Tab. 2) but changes in stride length were greater (ie, a mean increase of 0.4 m from phase AI to phase B). Stride length for normal sprinting has been reported to range from 2.2 to 2.6 m.[42] Thus, our subject moved from a below-normal stride length value (1.9 m) during the baseline phase to a low normal value (2.3 m) during the intervention phase.

In contrast to the results for comfortable walking, the changes seen in the percentages of stance and swing during running appear to have consequences for function for two reasons. First, the absolute magnitude is more impressive. The change in percentage of stance from phase AI to phase B represents a decrease of 25.8 milliseconds (10.3%) for the right lower extremity and of 47.5 milliseconds (18%) for the left lower extremity. For percentage of swing, the comparable values were a 23.8-millisecond (5%) increase for the right lower extremity and a 45.5-millisecond (10%) increase for the left lower extremity. To achieve the same stance time values as those of an age-matched person without impairments, our subject would have needed a 95-millisecond (38%) decrease in stance time for his right lower extremity and a 110-millisecond (42%) decrease in stance time for his left lower extremity. To achieve the same swing time values as those of an age-matched person without impairments, our subject would have needed a 109-millisecond (23%) increase in swing time for his right lower extremity and a 92-millisecond (20%) increase in swing time for his left lower extremity. In this context, the changes observed in our subject's stance and swing times represent a quarter to one half of the change needed to match the values of a person without impairments.

The second important change in the percentages of stance and swing for running was a move toward right and left symmetry during phase B as compared with phase AI, which can be noted by comparing the values for percentages of stance and swing for right and left lower extremities listed in Table 2. In phase AI, the subject spent more time on his left lower extremity in stance (consistent with right lower-extremity weakness or impairment). In phase B, however, the subject not only reversed that asymmetry, but he actually spent slightly more time on the right lower extremity in stance. Comparable changes were seen for percentage of swing. These changes are notable because improvements in symmetry resulting from gait training are difficult to achieve.[43,44] We believe that a more symmetric gait is more functional because it requires less energy to perform if it is the subject's natural self-selected style. Abnormal gait in persons with hemiplegia, highly characterized by asymmetry, has been reported to require increased energy as compared with the gait of persons without impairments related to ambulation (61% average increase in oxygen consumption at comfortable walking speeds).[45]

Within-phase comparison (phase B). This discussion has focused on the average changes that occurred over the entire 6-week treatment period as compared with the baseline phase. The changes that occurred within the 6-week treatment phase also were of interest to us. The immediate treatment effects (ie, within-session pretreatment and posttreatment values) were, for the most part, negative. That is, for some variables, the subject's performance worsened immediately following treatment with the PBWS and treadmill training. This finding could be the result of two factors. Fatigue might have altered performance (although the subject was given a 10-minute rest in a sitting position to counteract this possibility, and he did not report feeling fatigued). Another factor is a difference in the nature of the motor task in treadmill walking versus level-surface walking. In treadmill walking, the "ground" moves while the subject is stationary in space. In normal walking, however, the subject moves, whereas the ground is stationary. The optic flow patterns received by the subject in normal versus treadmill walking are thus very different. These differences can generate motion aftereffects following treadmill locomotion in which the visual world is perceived to be moving and the subject feels less stable. The subject reportedly experienced these motion aftereffects for durations of 30 to 120 seconds following treadmill locomotion. Again, the rest period was designed to counteract these effects but may have been too brief.

In contrast to the within-session pretreatment and posttreatment values, session-to-session values during the treatment phase showed a pattern of slow, but steady, improvement for comfortable and fast walking (Figs. 2-5). We believe that this outcome is a positive result because it supports the idea that true carryover to function occurred. Comparison of the phase B pretreatment performance values for speed with the training speeds used in phase B shows that the PBWS allowed the subject to train at higher speeds than he naturally selected without the PBWS. For example, during weeks 1 and 2 of training, the subject's pretreatment comfortable walking speeds were 2.98 and 3.06 mph, respectively, and the peak training speeds achieved were 3.0 and 3.5 mph, respectively. These training speeds were still below the subject's self-selected fast walking speed. By the third week, the subject was able to train at speeds that exceeded his self-selected fast walking speeds (eg, in week 4, session 12, his fast walking speed was 3.99 mph and his training speed was 4.5 mph). The training speeds exceeded the average baseline speeds for comfortable and fast walking but not for running.

In humans, stance/swing ratios change dramatically with increased speed of locomotion.[39] Speed changes in human gait are produced predominantly by changes in stance duration time, which shortens with increased speed. Improvements in our subject's gait noted after the combined treadmill-PBWS training may have stemmed from practice at speeds that were higher than his self-selected speeds and from activation of CPGs and the switching mechanism for the stance and swing phases, which is "trained" with repeated activation. Improved function of the switching mechanism then can be facilitated by attempts to ambulate at higher speeds without the treadmill-PBWS system.

Based on the results, it appears that practice speeds do not have to match a normal "switching" speed.[39] As shown in Table 1, our subject trained only at speeds of 4.5 mph or below, yet he demonstrated an improvement in his ability to "switch" to a more normal running pattern (with shorter stance and longer swing phase) at higher speeds following the treadmill-PBWS treatment.

Table 1.
Treadmill (TM) Training Protocol(a) With TM Speed
(in Miles per Hour)

TM                              Time (min)
Session
No.         0-4    5-8    9-12    13-16   17-20

1           1.5    2.0     2.5     3.0      3.0
2           1.5    2.0     2.5     3.0      3.0
3           1.5    2.0     2.5     3.0      3.0
4           1.5    2.0     2.5     3.0      3.0
5           1.5    2.0     2.5     3.0      3.5
6           1.5    2.0     2.5     3.0      3.5
7           1.5    2.0     2.5     3.0      3.5
8           2.0    2.5     3.0     3.5      4.0
9           2.0    2.5     3.0     3.5      4.0
10          2.0    2.5     3.0     3.5      4.0
11          2.5    3.0     3.5     4.0      4.5
12          3.0    3.5     4.0     4.5      4.5
13          2.5    3.0     3.5     4.0      4.0
14          2.5    3.0     3.0     3.5      3.5
15          3.0    3.5     4.0     4.0      4.0
16          3.0    3.5     4.0     4.0      4.0

(a) This training protocol was modified from the original Bruce protocol.(35)

Heart Rate

The subject demonstrated improvements in cardiovascular status, as evidenced by (1) a steady progression of heart rate increase at greater treadmill intensities, (2) increased intensity and duration of training, and (3) evidence of a steady state at higher constant workloads than the subject was able to achieve earlier.

A training effect may be evident based on a decline in heart rate at a constant workload over several training sessions. For example, the subject's heart rate was 118 beats per minute (bpm) at 3.0 mph in session 1 and declined to 96 bpm at 3.0 mph by session 15. An overall increase in the intensity and duration of training was evident over time. Intensity increased from 3.0 mph in session 1 to 4.5 mph in session 11. The duration of training at a higher intensity also increased, as indicated by the subject's increased duration of training at 4.0 mph (ie, 4 minutes in session 8 versus 12 minutes in sessions 15 and 16). The subject achieved 60% to 65% of maximum predicted heart rate (115-124 bpm) after approximately 15 minutes in most sessions (Tab. 3).

Use of the treadmill-PBWS system allowed the subject to train at higher speeds than was possible without the system, thus, theoretically, improving cardiovascular conditioning. A better assessment of cardiovascular effects would have been to use maximal oxygen uptake, which has been shown to correlate well with physical training, had this been available.[35] Oxygen uptake, however, has been found to increase almost linearly with heart rate,[35] which, for our study, provided a reasonable and clinically feasible substitute measure.

Conclusions

The treadmill-PBWS training used in this study produced improvements in the subject's gait performance off the treadmill. Effects were small for comfortable and fast walking speeds but larger for running speed. Improvements in gait variables may have been due to neural mechanisms influenced by retraining the stepping mechanism through CPG activation. Cardiovascular changes were noted by achievement of a steady state, with a corresponding decrease in heart rate at higher exercise intensities. The results of this single-subject experimental design study suggest that the treadmill-PBWS gait training technique holds sufficient promise for functional benefit to justify testing its efficacy in a larger group of subjects with neurological impairments.

Acknowledgments

We thank Dr Bjorn Svendsen for his kindness in lending us the weight relief apparatus used in the study and for his helpful advice on the practical aspects of using the vest and weight relief system; Almas Dossa, PT, for her assistance in setting up the heart rate monitoring equipment; and Terry Michel, PT, CCS, and Cynthia Zadai, PT, CCS, for their helpful advice and comments regarding the cardiovascular aspects of the study. We thank Leslie Portney, PhD, PT, and David Uddin, PhD, for their assistance with the statistics. We also thank Robin Cole for her computer assistance and Sheila Roush, PT, and Robin Corey, PT, for their clinical input.

(*) Svendsen Consultants Inc, 416 N Homer St, Lansing, MI 48912.

([dagger]) Challenger Fitness Equipment Inc, 8201 Sovereign Row, Dallas, TX 75247.

([double dagger]) Camp International Inc, PO Box 84, Jackson, MI 49204.

([sections]) B&L Engineering, PO Box 3905, 8807 Pioneer Blvd, Unit C, Santa Fe Springs, CA 90670.

([parallel]) International Business Machines Inc, 1000 NW 51st St, Boca Raton, FL 33431.

(#) Transkinetics, 110 Shawmut Rd, Canton, MA 02021.

References

[1] Fung J, Stewart JE, Barbeau H. The combined effects of clonidine and cyproheptadine with interactive training on the modulation of locomotion in spinal cord injured subjects. J Neurol Sci. 1990;100:85-93.

[2] Stewart JE, Barbeau H, Gauthier S. Modulation of locomotor patterns and spasticity with clonidine in spinal cord injured patients. Can Neurol Sci. 1991;18:321-332.

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

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

[5] Wall JC, Turnbull GI. Gait asymmetries in residual hemiplegia. Arch Phys Med Rehabil. 1986;67:550-553.

[6] Brandstater ME, de Bruin H, Gowland C, Clark BM. Hemiplegic gait: analysis of temporal variables. Arch Phys Med Rehabil. 1983;64:583-587.

[7] Mizrahi J, Susak Z, Heller L, Najenson T. Objective expression of gait improvement of hemiplegics during rehabilitation by time-distance parameters of the stride. Med Biol Eng Comput. 1982;20:628-634.

[8] Richards CL, Malouin F, Dumas F, et al. Gait velocity as an outcome measure of locomotor recovery after stroke. In: Craik RL, Oatis CA, eds. Gait Analysis: Theory and Applications. St Louis, Mo: Mosby; 1995:355-364.

[9] Finch L, Barbeau H. Hemiplegic gait: new treatment strategies. Physiotherapy Canada. 1986;38:36-41.

[10] Delcomyn F. Neural basis of rhythmic behavior in animals. Science. 1980;210:492-498.

[11] Grillner S. Control of locomotion in bipeds, tetrapods, and fish. In: Brookhart JM, Mountcastle VB, eds. Handbook of Physiology: The Nervous System, Volume II: Motor Control, Part 2 Washington, DC: American Physiological Society; 1981:1179-1236.

[12] Grillner S. Locomotion in the spinal cat. In: Stein RB, Pearson KG, Smith RS, Redford JB, eds. Advances in Behavioral Biology: Control of Posture and Locomotion. New York, NY: Plenum Publishing Corp; 1973;7:515-535.

[13] Smith JL. Programming of stereotyped limb movements by spinal generators. In: Stelmach GE, Requin J, eds. Tutorials in Motor Behavior. New York, NY: Elsevier Science Inc; 1980:95-115.

[14] Rossignol S, Barbeau H, Julien C. Locomotion of the adult chronic spinal cat and its modifications by monoaminergic agonists and antagonists. In: Goldberger ME, Gorio A, Murray M, eds. Development and Plasticity of the Mammalian Spinal Cord. Padua, Italy: Liviana Editrice SpA; 1986:323-345. Fidia Research Series.

[15] Barbeau H. Rossignol S. Recovery of locomotion after chronic spinalization in the adult cat. Brain Res. 1987;412:84-95.

[16] Pillar T, Dickstein R, Smolinski Z. Walking reeducation with partial relief of body weight in rehabilitation of patients with locomotor disabilities. J Rehabil Res Dev. 1991;28:47-52.

[17] 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.

[18] Malouin F, Richards CL. Menier C, et al. Supported Walker Ambulation Performance Scale (SWAPS): development of an outcome measure of locomotor status in children with cerebral palsy. Pediatric Physical Therapy. 1997;9(2):48-53.

[19] Kelsey DD, Tyson E. A new method of training for the lower extremity using unloading. J Orthop Sports Phys Ther. 1994;19:218-223.

[20] Kline Mangione K, Axen K, Haas F. Mechanical unweighting effects on treadmill exercise and pain in elderly people with osteoarthritis of the knee. Phys Ther. 1996;76:387-394.

[21] Hunter D, Cole EZ, Murray JM, Murray TD. Energy expenditure of below-knee amputees during harness supported treadmill ambulation. J Orthop Sports Phys Ther. 1995;21:268-276.

[22] Van Faassen F, Molen NH. A new method in the treatment of gait disorders. In: Cerquiglini S, Venerando A, Wartenweiler J, eds. Biomechanics III. Baltimore, Md: University Park Press; 1973:489-491.

[23] Grillner S, Rossignol S. On the initiation of the swing phase of locomotion in chronic spinal cats. Brain Res. 1978;146:269-277.

[24] Hersen M, Barlow D. Single-Case Experimental Designs: Strategies for Studying Behavior Change. New York, NY: Pergamon Press; 1984.

[25] Kendall FP, McCreary E. Muscles: Testing and Function. 3rd ed. Baltimore, Md: Williams & Wilkins; 1983.

[26] Lawson A, Calderon L. Interexaminer reliability of applied kinesiology manual muscle testing. Percept Mot Skills. In press.

[27] Lilienfeld AM, Jacobs M, Willis M. A study of the reproducibility of muscle testing and certain other aspects of muscle scoring. Phys Ther Rev. 1954;34:282.

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

[29] Goodkin R, Diller L. Reliability among physical therapists in diagnosis and treatment of gait deviations in hemiplegics. Percept Mot Skills. 1973;37:727-734.

[30] Robinson JL, Smidt GL. Quantitative gait evaluation in the clinic. Phys Ther. 1981;61:351-353.

[31] Holden MK, Gill KM, Magliozzi MR, et al. Clinical gait assessment in the neurologically impaired: reliability and meaningfulness. Phys Ther. 1984;64:35-40.

[32] Wall JC, Ashburn A. Assessment of gait disability in hemiplegia. Phys Ther. 1965;45:17-29.

[33] Holden MK, Gill KM, Magliozzi MR. Gait assessment for neurologically impaired patients: standards for outcome assessment. Phys Ther. 1986;66:1530-1539.

[34] Dubin D. Rapid Interpretation of EKGs. 3rd ed. Tampa, Fla: Cover Publishing Co; 1986.

[35] Ellestad MH. Stress Testing: Principles and Practice. 3rd ed. Philadelphia, Pa: FA Davis Co; 1986:164-165.

[36] Ottenbacher KJ. Evaluating Clinical Change: Strategies for Occupational and Physical Therapists. Baltimore, Md: Williams & Wilkins; 1986.

[37] Portney LG, Watkins M. Foundations of Clinical Research: Applications to Clinical Practice East Norwalk, Conn: Appleton & Lange; 1992.

[38] Payton O. Research: The Validation of Clinical Practice. Philadelphia, Pa: FA Davis Co; 1979:108.

[39] Corcoran PJ. Energy expenditure during ambulation. In: Downey JA, Darling RC, eds. Physiological Basis of Rehabilitation. Philadelphia, Pa: WB Saunders Co; 1971:185-198.

[40] Imms FJ, Edholm OG. Studies of gait and mobility in the elderly. Age Ageing. 1981;10:147-156.

[41] McArdle WD, Katch FI, Katch VL. Exercise Physiology: Energy, Nutrition, and Human Performance. 2nd ed. Philadelphia, Pa: Lea & Febiger, 1986.

[42] Astrand P, Rodahl K. Textbook of Work Physiology: Physiological Bases of Exercise. 2nd ed. New York, NY: McGraw-Hill Book Co; 1977.

[43] Hesse SA, Jahnke MT, Schreinder C, Mauritz KH. Gait symmetry and functional walking performance in hemiparetic patients prior to and after a 4 week rehabilitation program. Gait and Posture. 1993;1:166-171.

[44] Winstein CJ. Balance retraining: Does it transfer? In: Duncan PW, ed. Proceedings of the APTA Forum on Balance; June 13-15, 1989;.Nashville, Tenn. Alexandria, Va: American Physical Therapy Association; 1989: 95-104.

[45] Corcoran PJ, Jebsen RH, Brengelmann GL, Simons BC. Effects of plastic and metal leg braces on speed and energy cost of hemiparetic ambulation Arch Phys Med Rehabil. 1970;51:69-77.

MB Gardner, PT, is Physical Therapist, Physical Therapy Department, Miami Valley Hospital, One Wyoming St, Dayton, OH 45409 (USA) ([e-mail]). She was Staff Physical Therapist, Spaulding Rehabilitation Hospital, Boston, Mass, and a graduate student at MGH Institute of Health Professions, Boston, Mass, when this study was conducted in partial fulfillment of the requirements for her Master of Science degree. Address all correspondence to Ms Gardner.

MK Holden, PhD, PT, is Assistant Professor and Coordinator, Neurology Specialization, MGH Institute of Health Professions. Dr Holden was also

Senior Research Therapist and Biofeedback Consultant, Physical Therapy Department, Spaulding Rehabilitation Hospital, when this study was conducted.

JM Leikauskas, PT, is Physical Therapist, North Shore Visiting Nurse Association, Danvers, Mass. She was Assistant Director, Physical Therapy Department, Spaulding Rehabilitation Hospital, when this study was conducted.

RL Richard, PT, is Burn Clinical Specialist, Miami Valley Hospital, Dayton, Ohio.

This article was adapted from a presentation at the Joint Congress of the Canadian Physiotherapy Association and the American Therapy Association, June 4-8, 1994, Toronto, Ontario, Canada.

This work was supported in part by a grant from the Ionta Research Fund.

The study was reviewed and approved by the human subjects committees of Spaulding Rehabilitation Hospital and Massachusetts General Hospital.

This article was submitted March 17, 1995, and was accepted June 24, 1997.