Professional Documents
Culture Documents
Andersson2015 PDF
Andersson2015 PDF
Karolinska Institutet, Stockholm, Sweden, 3Department of Physical Therapy, Karolinska University Hospital,
Stockholm, Sweden
Background: People with gait difficulties after a stroke usually have an asymmetrical gait and slower gait speed
than age-matched controls. These difficulties restrict people with stroke in their daily life activities.
Objective: This pilot study sought to evaluate the effects of weight-shift training on gait, weight distribution in
standing, and ambulation in people with gait difficulties after a stroke.
Methods: Ten subjects with chronic stroke (3–11 years since insult) and remaining gait difficulties participated in
a 3-week weight-shift training program. Spatial and temporal gait parameters were assessed pre-, post-, and
3-month post-training with a motion analysis system. Weight distribution was assessed with force plates and
ambulation with the Swedish version of the Clinical Outcome Variables Scale (S-COVS). Wilcoxon signed-rank
tests were used to explore differences between test occasions.
Results: Significant changes were seen between pre-and post-tests in decreased stance time on the non-paretic
leg (P = 0.005) and increased score on the S-COVS (P = 0.043). At the 3-month follow-up test, the subjects had
also increased their gait speed significantly (P = 0.037). Standing weight distribution did not change between
pre- and post-tests (P = 0.575), but between the pre-and follow-up tests it shifted from the paretic leg to the
non-paretic (P = 0.007).
Conclusion: Weight-shift training seems to improve gait and ambulation in subjects with chronic stroke, but
not with standing weight distribution. However, this pilot study has several limitations and a larger sample size
with a control group is necessary.
Keywords: Exercise, Gait, Asymmetry, Force Plate, Cerebral lesion
Zero percent indicates symmetry between sides and 100% 2) stepping up and down on a platform with wall-support
indicates maximal asymmetry between sides. Positive val- (Fig. 1B): the subjects stepped, with the paretic side
ues indicate greater loading on paretic side. nearest the wall, up onto a 14- cm high, 90- cm long, and
35- cm wide platform. They had to step up and down
walking forwards and then backwards, always with the
Ambulation paretic side against the wall. To be able to step and lift the
Ambulation was assessed with the Swedish Clinical non-paretic leg, the subjects had to transfer weight to the
Outcome Variables Scale (S-COVS).26 This scale con- paretic leg so that the shoulder of the paretic side touched
sists of 13 motor tasks where the examiner rates from the wall. This task was executed 10 times in three sets with
2 minutes break between each set.
1 (dependent) to 7 (normal) the level of performance in
3) sitting down and rising from a chair with wall-support
different motor tasks. The test permits measurement of (Fig. 1C): the subjects sat on a chair without armrests,
activities regarding transportation, gait, postural control with the paretic side close to a wall. The paretic foot was
in sitting position, handling a wheel-chair, and arm/hand positioned with a fairly large knee angle ( ≤ 90°) closer
function. In this study, only items five to eight, repre- to the chair than the non-paretic foot. The subjects had
senting ambulation variables, were assessed. These were to stand up and sit down while shifting body weight to
the paretic side so that the shoulder of the paretic side
human support, walking aids, endurance, and velocity.
touched the wall. They repeated the task 15 times in three
The individual scores were summarized (maximum 28). sets with 2 minutes break between each set.
The S-COVS was run before, directly after and 3 months
after training. It has been tested for inter-rater reliability In the first session, the subjects were told to lean toward
for the total scores of the test and showed a 0.97 corre- the wall during the stance phase on the paretic leg. This
lation coefficient: 0.93 for the specific items used in this was to make them feel their weight bearing and the con-
study.26 tact of the shoulder with the wall. The wall was used to
create a somatosensory input for the subjects to know
Intervention when they had executed the right weight transfer. Many
The training period consisted of four 1 hour sessions a subjects said they were afraid of falling when shifting their
week for 3 weeks. Every subject trained individually with weight toward the paretic side, so the wall was a safety
the same physical therapist. item preventing them from falling.
The subjects’ symmetry training consisted of three tasks: As soon as the subjects had learned to control the
weigh shift, they were asked to do the exercises without
1) walking with wall-support (Fig. 1A): the subjects walked touching the wall. The goal was to walk without the wall
with the paretic side close to a wall. They were instructed
as a touch reference or safety item. The progression from
to place the foot of the paretic leg on a 3- cm-wide and
10- m-long line running 20 cm from the wall. During having the subjects lean toward the wall to not touch-
the stance phase of the paretic leg, the subjects were ing the wall was decided by the clinician as soon as the
instructed to shift their trunk so that the shoulder of the subjects managed to control their weight shift with the
affected side touched the wall. Wearing regular shoes, right amount of force to touch the wall, not missing it or
they walked along the wall in this manner 10 times in hitting it too hard, and managed this in a 10 m session. The
three sets with 2 minutes rest between each set.
Figure 1 The intervention: symmetry-training in three different tasks. Left panel (A): walking with wall-support. Middle panel (B):
stepping up and down on a platform with wall-support. Right panel (C): rising from and sitting down on a chair with wall-support.
Table 2 Gait parameters for paretic and non-paretic side, Variable Pre- Post- 3-month P-value
median (range) follow-up
T2–T3
Variable Pre- Post- 3-month P-value
P = 0.444
follow-up
Stance 62 (52–76) 59 (43–74) 58 T1–T2
Velocity 0.65 0.67 0.72 T1–T2 phase (49–72) P = 0.037
(m/s) (0.2–1.1) (0.2–1.2) (0.3–1.2) P = 0.203 (% gait
T1–T3 cycle)
P = 0.037 T1–T3
T2–T3 P = 0.218
P = 0.074 T2–T3
Cadence 88.9 88.4 92.6 T1–T2 P = 1.000
(steps/ (47.5– (46.5– (57.0– P = 0.203
min) 107.5) 111.8) 117.8) Bold indicates P < 0.05.
T1–T3
P = 0.074
T2–T3 number of repetitions of each exercise was individually
P = 0.241
Paretic adjusted and ranged between 30 and 45.
Walking with wall-support took approximately
Step 431 452 464 T1–T2
length (240–655) (204–559) (253–609) P = 0.507
15–20 minutes. The subjects walked at about 0.5 m/s for
(mm) 300 m with 2 minutes break after every 100 m. They also
T1–T3 needed about 5 seconds between each 10 m session to
P = 0.575
T2–T3 refocus and to adjust their body position. The stepping up
P = 0.059 and down task took approximately 20 minutes (30 repeti-
Stride 813 (561– 871 (552– 903 (504– T1–T2
tions); here, the subjects needed about 30 seconds to step
length 1263) 1293) 1348) P = 0.139
(mm) up and down with 2 minutes break between each set. The
T1–T3 sitting down and rising from a chair task took approxi-
P = 0.059
T2–T3 mately 15 minutes (45 repetitions) in which the subjects
P = 0.444 needed about 10 seconds for each rising and sitting down
Stance 709 (530– 835 (603– 739 (565– T1–T2 and 2 minutes break between each set. When the subjects
time (ms) 1505) 1785) 1243) P = 0.093
T1–T3 had executed the minimum amount of repetitions for each
P = 0.959 task, they were asked if they could manage to execute more
T2–T3
P = 0.203
repetitions. Likewise, they were also asked to report if they
Stance 57 (43–59) 56 (50–63) 56 T1–T2 felt overtired and the trainer/clinician were observant if the
phase (53–60) P = 0.047 subjects were too tired or trained at a too low intensity level.
(% gait
cycle) The clinician leading the training always stood next
T1–T3 to the participants during the step- and -sit-to-stand exer-
P = 0.169 cises. During the gait training he walked close behind as
T2–T3
P = 0.415 a safety precaution in case participants should lose their
Non-pa- balance.
retic
Step 371 400 428 T1–T2
length (222–620) (284–665) (263–724) P = 0.241 Statistical analyses
(mm) Statistical analyses were performed using the STATISTICA
T1–T3
P = 0.332 computer package (Version 10, Statsoft. Inc., Tulsa, OK,
T2–T3 USA). Descriptive statistics (median and range) were calcu-
P = 0.959
lated for all variables. Non-parametric statistics (Wilcoxon
Stride 829 (585– 883 (495– 885 T1–T2
length 1311) 1214) (263–724) P = 0.721 signed-rank test) was used due to the small number of sub-
(mm) jects and skewed distribution of data. P values less than
T1–T3
P = 0.878 0.05 were considered as statistically significant.
T2–T3
P = 0.074 Results
Stance 878 (643– 795 (448– 734 (498– T1–T2
time (ms) 1935) 1768) 1510) P = 0.005 All 10 subjects attended every training session during the
T1–T3 3 week intervention, i.e., 100% compliance.
P = 0.005
Figure 2 Mean gait velocity in m/s, pre-training, post-training and 3 months after training are shown in the upper panel. Upper
left (A): group values and significant (P < 0.05) improvement in velocity between pre-training and follow-up, marked with a
*. Upper right (B): individual trend of gait velocity between different test occasions, all 10 subjects. The lower panel shows
standing weight distribution according to the asymmetry index on the three different test occasions. Lower left (C): weight
distribution for whole group and significant change (P < 0.01), marked with an **. Lower right (D): very diverse individual trends
of weight-bearing between the different test occasions, all 10 subjects.
wall-support. In contrast, the standing weight distribu- This could reflect an increase in dynamic balance con-
tion shifted from the paretic to the non-paretic leg. The trol – which is highly correlated with gait symmetry.33
improvement in gait velocity seen at the follow-up but not At individual level, 7 out of 10 subjects improved their
directly after the training period could be due to motor ambulation scores enough to gain one higher increment,
learning, i.e., that the subjects had had to learn a ‘new’ a significant increase spread through all items. Among
gait pattern by transferring more weight to their paretic the seven subjects who increased their ambulation scores,
leg during the intervention, which in fact slowed their gait three managed a one-increment increase, two a two-incre-
speed down. Later on, with more time to adapt to their ment increase, and two a three-increment increase. One
new gait pattern, they were able to walk faster. This is subject, for example, only managed to walk over 10 m
supported by the model on motor learning by Vereijken before training; while after training, he could walk over
et al.27 where the learner initially simplifies the move- 100 m – which he sustained at follow-up.
ment to reduce the degrees of freedom by constraining One way of interpreting the results in a clinical context
several joints, which he/she gradually releases as the skill would be by comparing them to minimal clinical impor-
is learned. On the other hand, the increase in gait speed tance differences (MCID). MCID has been investigated
which only appeared at 3 months follow-up could arguably on gait speed in individuals with stroke and found to be
be due to a reversion to a more asymmetrical gait pattern 0.16 m/s.34 Only one of our subjects improved gait velocity
since the weight-shift was again more to their non-paretic more than 0.16 m/s. However, the above study took place
leg. This may indicate a learned bad use28 where compen- during the first 60 days post-stroke and is not comparable
satory movements compete with more ‘normal’ move- to our sample of chronic-stroke subjects.34
ments after the training period, when the individuals with This pilot study has several limitations. The sam-
stroke were left on their own to maintain their gait pattern. ple size was small and there was no control group.
An increase in stride length and cadence significantly This may have compromised the generalizability of the
correlates to an increase in gait velocity.29,30 Although results. The lack of control group makes it plausible
there were no significant changes in cadence, stride, or that the results would have appeared even without an
step length between the test occasions, they all showed intervention, although spontaneous improvements must
a tendency to increase and, combined, could explain the be considered less likely in this group. Likewise, this
increase in gait velocity seen at follow-up. The stance study permits no conclusions concerning the benefit
phase decreased in both legs directly after the training of this training in comparison to other interventions.
session, but this did not persist 3 months after the training. The intervention period was quite short although the
However, the decrease led to a similar and more normally number of sessions (n = 12) was similar to those in
distributed stance phase in both legs. Persons with balance other studies.35–37 In addition, weight distribution was
deficits have a longer stance phase than normal subjects.31 assessed during standing instead of during walking,
The normalization of the stance phases seen after the pres- which would have been more specific to the exercises
ent intervention might be due to an increase in dynamic in the intervention.
balance gained from the wall-supported gait training. In conclusion, symmetry training with wall-support
The weight distribution in standing shifted from more seems to improve gait and ambulation in subjects with
weight on the paretic leg before the training period to chronic stroke, but not standing weight distribution. This
more weight on the non-paretic leg at follow-up. This study adds evidence to the fact that individuals with stroke
contradicts the results on stance phase and stance time, can improve their physical performance several years after
indicating a more symmetrical gait after the training their stroke.38 It highlights the importance of rehabilitation
period. However, the weight distribution here was meas- programs in the chronic stage of stroke. However, given
ured during quiet stance and not during gait. Carry-over its several limitations, a larger sample size with a control
effects of weight distribution from standing tasks to gait group is necessary.
are also limited,32 a finding in agreement with another
study by our group23 which also found no improvement Acknowledgements
in standing weight distribution after wall-supported gait We thank all the subjects participating in this study,
training. associate professor Helga Hirschfeld for helping with
Since the S-COVS is an ordinal scale, small changes in the design and acquisition of gait parameters, Eva
motor ability do not alter scores. For example, in item 8 Eriksson, physical therapist student, for analysis of
(walking velocity), a subject had to increase from < 0.3 m/s weight distribution in standing, and Heidi Andersson,
to < 0.6 m/s to score higher. Our subjects with chronic physical therapist, for help with clinical testing. Thanks
stroke had significantly improved their ambulation score also to Ingmarie Apel for technical assistance during
directly after the training session and also at follow-up. the laboratory tests.