NeuroRehabilitation 20 (2005) 307–322 307

IOS Press

The effects of loading and unloading

treadmill walking on balance, gait, fall risk,
and daily function in Parkinsonism
Tonya Toolea,∗, Charles G. Maitlandb , Earl Warrenc , Monica F. Hubmann d and Lynn Panton a
a
Department of Nutrition, Food & Exercise Sciences, College of Human Sciences, Florida State University,
Tallahassee, FL, USA
b
Department of Clinical Sciences, College of Medicine, Florida State University, Tallahassee, FL, USA
c
Tandem Physical Therapy, Tallahassee, FL, USA
d
Neuroscience Center and Parkinson’s Center, Tallahassee, FL, USA

Abstract. Our study aims were: 1) to determine whether assisted weight bearing or additional weight bearing is more beneficial
to the improvement of function and increased stability in gait and dynamic balance in patients with Parkinsonism, compared with
matched controls (treadmill alone). Twenty-three men and women participants (M ± SD = 74.5 ± 9.7 yrs; Males = 19, Females
= 4) with Parkinsonism were in the study. Participants staged at 1–7 (M ± SD = 3.96 ± 1.07) using the Hoehn & Yahr scale.
All participants were tested before, after the intervention (within one week), and four weeks later on: 1) dynamic posturography,
2) Berg Balance scale, 3) United Parkinson’s Disease Rating Scale (UPDRS), 4) biomechanical assessment of strength and range
of motion, and 5) Gaitrite force sensitive gait mat. Group 1 (treadmill control group), received treadmill training with no loading
or unloading. Group 2 (unweighted group), walked on the treadmill assisted by the Biodex Unweighing System at a 25% body
weight reduction. Group 3 (weighted group), ambulated wearing a weighted scuba-diving belt, which increased their normal
body weight by 5%. All subjects walked on the treadmill for 20 minutes per day for 3 days per week for 6 weeks. Improvements
in dynamic posturography, falls during balance testing, Berg Balance, UPDRS (Motor Exam), and gait for all groups lead us to
believe that neuromuscular regulation can be facilitated in all Parkinson’s individuals no matter what treadmill intervention is
employed.

Keywords: Rehabilitation, intervention, treadmill

1. Introduction to attempt to intervene using physical therapy programs

in an effort to rectify this instability. There is an ex-
Difficulty with balance and equilibrium is a major traordinary lack of research data available on the neuro-
problem experienced by individuals with Parkinson- muscular characteristics of motor control and postural
ism. It is generally held by specialists in the field stability in Parkinsonism, and in particular, as it relates
of movement disorders that postural instability does to therapeutic interventions. We studied the multicom-
not respond adequately to current drug therapies avail- ponent contribution to equilibrium in Parkinsonism and
able [28]. As a consequence, we believe it is important determined that lower limb strength makes a major con-
tribution to upright stability [34]. We then conducted
a series of therapeutic trials utilizing strength and bal-
∗ Address for correspondence: Tonya Toole, PhD, Professor
ance training in patients with advanced disease [18–
Emerita, 436 Sandels Building, Department of Nutrition, Food, and
Exercise Sciences, Florida State University, Tallahassee, FL 32306-
20,33]. Those patients who completed strength and
1493, USA. Tel.: +1 850 508 3319; Fax: +1 850 645 5000; E-mail: balance training demonstrated improvement in equi-
ttoole@mailer.fsu.edu. librium scores and an overall reduction in fall fre-

ISSN 1053-8135/05/$17.00 © 2005 – IOS Press and the authors. All rights reserved
308 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

quency [20,33]. Moreover, these patients sustained im- Utilization of unweighting techniques as specific
provement in equilibrium scores and fall reduction after therapeutic modalities is now under way in patients who
four weeks of abstinence from any subsequent formal have suffered stroke with partial or complete paralysis,
exercise [20]. In another study, we demonstrated that spinal cord injury, knee injury and following hip frac-
aerobic exercise was feasible in patients with Parkin- ture. Visintin et al. [36] treated 100 stroke patients and
sonism and this intervention improved both aerobic ca- produced a significant difference in gait improvement,
pacity and movement initiation times [1]. Despite these motor strength control, over-ground walking speed and
observations, there still exists a very limited research over-ground walking endurance compared with age
base dedicated to determining methods for sustaining match controls. Hesse et al. [16] demonstrated that
improvement in mobility, balance and reducing falls’ treadmill training with partial body weight support in
risk in patients who live with Parkinsonism. Therefore, hemiparesis produces a favorable gait characterized by
the overall goal of this study is to determine the effects prolonged single stance period of the affected limb,
of treadmill walking on motoric function including bal- higher symmetry and more regular activation pattern as
ance, gait, fall risk, and daily function. compared with simple floor walking. Wirz et al. [37]
The ability to produce rapid peak muscular force in studied the long-term effects of body weight support
lower extremities is important for daily function such during locomotor training in patients with spinal cord
as walking, climbing stairs, and braking while driving injury. Leg extensor muscle EMG activity increased
a car. Dietz [7] proposed that such force is driven by significantly over the training period.
proprioceptive reflexes involved in the maintenance of The role of unweighing or assisted weight bearing as
body equilibrium and depends on the presence of con- treatment in Parkinsonism has rarely been studied and
tact forces opposing gravity. Afferent input probably to our knowledge, no study has combined unweigh-
arises from the Golgi tendon organ. Golgi tendon bod- ing with protracted physical therapy. The limited data
ies, serving as extensor load receptors, seem to signifi- available suggest there are no positive relationships be-
cantly contribute to force production, muscle amplitude tween limited weight bearing and treadmill walking and
and gait regulation in experimental models [3,13] and gait patterns in Parkinsonism. Further, to our knowl-
in man [21]. Golgi receptor body input probably is me- edge again, there is no study that has used enhanced
diated by group 1B afferent nerve fibers and is excita- weight bearing to increase body load combined with
tory to leg extensor muscles. Dietz et al. [11] and Dietz automatic treadmill walking in patients with Parkinson-
and Colombo [8] suggest ineffective extensor load re- ism. Fouad et al. [14] studied healthy subjects walk-
ceptor function results in disregulation of gait in Parkin- ing on a split-belt treadmill with increased body load.
sonism. Whereas the amplitude of ankle dorsiflexor Reflex components were larger when the body was
and tibialis anterior muscle activity does not differ from loaded compared with no load. Based on these results
healthy age/match subjects during walking, ankle ex- it would seem that those with Parkinsonism who have
tensor and gastrocnemius muscle amplitude does and is reduced EMG contraction in the gastrocnemius during
clearly different from healthy subjects. The differential gait push-off might benefit from gait training using an
behavior of these antagonistic leg muscles may be due extra body load to enhance gastrocnemius contraction.
to a modulation of gastrocnemius muscle activity at a The power phase of walking includes push-off from
local proprioceptive feedback level; whereas neuronal the support surface and is important in speed regulation.
control of leg flexors may be generated predominantly Leg extensor muscle activity is paramount for a long
through polysynaptic central mechanisms. Such a dis- and brisk stride as it determines the force of push off.
crepancy in feedback and innervation might explain the On the other hand, mechanical contribution from grav-
difference in muscle force between antagonistic muscle ity, leg mass effect, and limb acceleration facilitates the
groups. smooth swing phase during bipedal walking generated
Purely spinal mechanisms, i.e. spinal pattern or step through the leg flexor muscles. As a consequence of
generators, may be responsible for sustained repetitive this mechanical contribution, less neuronal input seems
walking in healthy humans [25]. Spinal pattern gen- required for the flexion or swing phase of the walking
erators may produce automatic walking independent pattern than it does during the force or stance phase.
of higher central nervous system control. Delwaide Dietz et al. [10] demonstrated that the flexion action is
and Gonce [5] have speculated that either a disordered uncompromised in Parkinsonism because mechanical
spinal mechanism or step generator may be responsible factors in that disease appear not to differ from healthy
for the phenomenon of freezing in Parkinson’s disease. control subjects. On the other hand, it appears that ex-
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 309

tensor muscle regulation is compromised in this condi- the body (score = 3), Stage 2B = add mild balance
tion. Decreased sensitivity of leg extensor regulation difficulties (score = 4), Stage 3 = postural instability or
extensively controlled by reflex mechanisms involving pronounced balance difficulties (score = 5),Stage 3B =
Golgi tendon organs, produces a gait characterized by postural instability and pronounced balance difficulties
reduced force and in turn, by reduced stride height, (score = 6), Stage 4 = patient requires assistance with
length and speed. The consequence is a slow shuffling walking (score = 7). Exclusion criteria included: un-
gait appearance [2]. Dietz et al. [9] concluded that gas- compensated cardiovascular disease, uncontrolled high
trocnemic activity was reduced in patients with Parkin- blood pressure, leg claudication, significant dementia,
sonism as measured by EMG compared with healthy other disorders of comprehension, and/or other medi-
controls under normal body loading conditions. Fur- cal conditions that would interfere with the participants
ther, when body unloading occurs between 25 and 75% safety and comfort during sub maximal exercise.
of total body weight, patients with Parkinsonism ex- All participants took their customary medication
hibit further reduction of gastrocnemic amplitude over within 2-4 hours prior to assessment and/or treatment
each of the load reductions [8,9]. As a consequence of session. No significant change in medication dosing
this reduction of motor activity, patients seem to par- was necessitated throughout the duration of the study
tially compensate by using a greater amount of flexor and all assessment and therapy sessions were sched-
activation, which in turn, leads to a higher degree of uled at the same time of day. Therapy sessions were
coactivation [12]. This muscle antagonism, in turn, held irrespective of the degree of freezing or dyski-
reduces force generation necessary to produce normal nesia movement. Each patient gave informed consent
stride height and length. after medical clearance from his or her primary care
The purpose of the present study was: 1) to demon- physician. Every participant was assessed as at least
strate that patients with advancing Parkinsonism have a level one (1) using the Functional Ambulation Cate-
the capacity to participate in physical therapy programs gory, which denotes that the participant can walk with
utilizing both assisted weight bearing (unweighting) or firm, continuous support of one person to help carry
additional weighted walking on a treadmill, 2) to de- weight or assist with balance.
termine whether assisted weight bearing or additional
weight bearing is more beneficial to the improvement 2.2. Tests
of function and increased stability in gait and dynamic
balance in patients with Parkinsonism, compared with All participants were tested before, after the inter-
controls walking on a treadmill without assisted or ad- vention (within 1 week), and 4 weeks later on: 1) dy-
ditional weight bearing loads, 3) to demonstrate a re- namic posturography (Equitest Sensory Organization
duction in falls’ frequency in testing conditions that Test, Neurocom, Inc.), 2) Berg Balance scale, 3) United
predictably result in disequilibrium, 4) to demonstrate Parkinson’s Disease Rating Scale (UPDRS), 4) biome-
that treadmill conditioning improves balance, and 5) chanical assessment of strength and range of motion
to demonstrate that patients undergoing such training (Hanoun Medical Evaluation System), and 5) Gaitrite
would demonstrate retention benefits. force sensitive gait mat. These tests were used to assess
equilibrium and falls during destabilizating conditions
(Equitest) and balance during dynamic and static con-
2. Methods ditions (Berg Balance), motor function and activities
of daily living (UPDRS), step length, velocity, and sta-
2.1. Participants bility (GaitRite mat, CIR Systems), and strength and
range of motion (Physical Therapy assessment).
Twenty-three men and women participants (M ±
SD = 74.5 ± 9.7 yrs; Males = 19, Females = 4) with 2.3. Procedures/intervention
parkinsonism were in the study. Participants staged at
1–7 (M ±SD = 3.96±1.07) using the Hoehn and Yahr Participants were randomly assigned by using a table
scale [17]. This scale has the following criteria: Stage of random numbers to one of the following groups:
1A = symptoms confined to one side of the body (score Group 1, (treadmill control group, M stage = 4.88
= 1), Stage 1B = add a midline symptom, such as ± 1.25, M age = 75.37 ± 7.99), received treadmill
speech and/or swallowing difficulty (score = 2), Stage training with no loading or unloading for 20 minutes
2A = Parkinson’s symptoms present on both sides of per day for 3 days per week for 6 weeks. For the first
310 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

3 weeks of the study, the treadmill was maintained in a After the intervention, participants were tested im-
horizontal incline for ease of training and to accustom mediately (within 1 week on all measures) and after a
participants using the treadmill. Treadmill velocity was one-month retention interval. This retention test helped
set at a level that was comfortable for each participant to determine the longevity of hypothesized gains during
with a target heart rate not to exceed 60% of theoret- treatment.
ical heart rate capacity. Rest periods were provided
as needed. For the final 3 weeks of intervention, the 2.4. Procedures/tests
treadmill was level of the first 4 minutes and raised to a
2% grade for 8 minutes and lowered to a 2% grade for 2.4.1. Gait
8 minutes. Grade was elevated in order to increase the Gait characteristics were assessed as participants
difficulty. walked down an instrumented GaitRite (CIR Systems,
Group 2 (unweighted group, M stage = 3.57 ± 0.53, Inc.) mat. This electronic walkway contains six sensor
M age = 76.42 ± 10.24), walked on the treadmill as- pads embedded within a carpet to produce an active
sisted by the Biodex Unweighing System at a 25% body area 24 inches wide and 144 inches long. As a par-
weight reduction for 15 minutes three days per week ticipant ambulates down the walkway, the system cap-
for 6 weeks. During each session and at the conclu- tures the relative arrangement, geometry and applied
sion of the 15-minute walk, the Unweighing System pressure of each footfall as a function of time. The
was removed and the participant walked for 5 minutes application software, integrated with an IBM compati-
more unassisted on the treadmill (the hands could hold ble computer, operates the walkway, processes the raw
a walking bar for stabilization as needed). For the first data into footfall patterns, and computes the temporal
3 weeks of the study, the treadmill was maintained in a and spatial parameters. Participants walked down the
horizontal incline for ease of training and to enable par- gait mat two times at normal pace and these two trials
ticipants to become accustomed to the treadmill. The
were averaged for data analysis.
treadmill velocity was set at a level that was comfort-
able for each participant not to exceed 60% of theoret-
2.4.2. Dynamic Posturography (Sensory Organization
ical heart rate capacity. Rest periods were provided as
Test)
needed on a per participant basis. For the final 3 weeks
Equilibrium was measured using the computerized
of intervention, the treadmill was raised to a 2% grade
dynamic posturography (CDP) [27]. CDP is a reli-
for 8 minutes and lowered to horizontal for 8 minutes
able [27] and valid [15] method of assessing balance.
while the person was wearing the 25% body weight
unloading harness. These conditions are hypothesized The measurements followed protocol and safety regu-
to promote motor recruitment for greater step thrust lations. The subject stood with feet centered facing a
(increase step length) and control (stability) and stimu- visual surround on a forceplate that measured anterior
late proper sequencing of muscle timing (deregulate co- to posterior center of gravity displacements on six sen-
activation) during gait. The unweighting harness was sory conditions. Each condition of the Sensory Orga-
removed for the last 4 minutes of horizontal treadmill nization Test (SOT) was tested in a 20 sec trial. When a
walking. fall occurred, however, subjects were tested on another
Group 3 (weighted group, M stage = 3.38 + 0.52, 20 sec trial until they were able to maintain balance or
M age = 72.0 ± 11.47), ambulated wearing a weighted until three trials were tested in that condition. When
scuba-diving belt, which increased their normal body subjects lost their balance, they were protected from
weight by 5%. Participants ambulated on a treadmill injury by a safety harness. Two support surface condi-
20 minutes per day 3 days per week for 6 weeks. Dur- tions, support sway-referenced (SSR), and support sta-
ing each session and at the conclusion of the first 15- ble (SS), and three visual conditions, eyes open (EO),
minutes, the additional 5% body weight was removed eyes closed (EC), or sway-referenced surround (SRS)
and the participant walked for 5 minutes more on the were used. During SSR, the floor and visual surround
treadmill. The treadmill was maintained at horizontal were controlled by a servomechanism (Gain = 1.0) to
incline for the first 3 weeks and then the incline was in- follow movements of the subjects’ center of gravity.
creased/decreased to a 2% grade as in the Group 1 and Sway-referencing eliminates accurate somatosensory
2 intervention for 8 minutes each. Participants were and/or visual information used to orient the body. Prior
allowed rest periods as needed, and treadmill velocity to each testing condition subjects were instructed that
was comfortable for each participant, not exceeding during testing the forceplate, visual surround, or both
60% of theoretical heart rate capacity. would move. For each condition an equilibrium score
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 311

(0–100 points) was calculated by the Equitest system. foot agility, arising from a chair, posture, gait, postural
The equilibrium score represents how much a subject stability, and bradykinesia. Scores on the 41 items are
sways in each trial. Scores closer to 100 indicate very summed within each category per individual. A trained
little excursion or sway while scores closer to 0 in- and experienced examiner rated each individual.
dicate increased anterior/posterior displacement. The
equilibrium score compares the subject’s sway during
each trial to a theoretical sway stability limit of 12.5 2.4.5. Strength
degrees [27]. A total composite score for all six con- The Hanoun Medical Evaluation system is a com-
ditions is used as an indicator of the subjects over- puter-integrated apparatus permitting reliable and ob-
all ability to adapt to altered visual and support en- jective measurement of isometric strength of joint mus-
vironments [27]. The equilibrium score is ecologi- culature and range of motion (ROM). There were eight
cally valid based on tests of environmental conditions strength variables measured on each leg: average ankle
that induce falls. When the cone of stability (theo- dorsiflexion, peak ankle dorsiflexion, average knee ex-
retical 12.5 degrees) is threatened the proprioceptive, tension, peak knee extension, average ankle plantarflex-
vestibular, and visual systems provide feedback to ef- ion, peak ankle plantar flexion, average knee flexion,
fect musculoskeletal corrections. The Equitest system and peak knee flexion. An average was calculated from
provides challenges to these systems and measures a 3 trials of each joint action.
body’s biomechanical response.
The Composite Score from the SOT is an average
of all 18 trials. This score is calculated by averaging 2.4.6. Range of motion (ROM)
condition 1 trials (1–3 trials), averaging condition 2 There were five ROM variables measured on each
trials (1–3 trials), and then adding all of the remaining leg by the Hanoun system. The five variables were:
12 trials. Lastly, this product is divided by 14. ankle dorsiflexion ROM, ankle plantarflexion ROM,
knee extension ROM, knee flexion ROM, and straight
2.4.3. Berg balance leg raise. Since each measure was taken on each leg
The Berg Functional Balance Assessment is a stan- there were 10 variables for ROM. Again, 3 trials were
dardized, reliable assessment tool for measuring a per- averaged for each ROM.
son’s balance during transfers and stance with imposed
restraints and conditions, with an interrater reliability
of 0.98 [4]. It contains 14 items rated by the experi- 2.5. Data analysis
menter on a scale of 0 to 4 for each item. Total scores
range from 0 to 56 indicating levels of assist required
for “safe” mobility. Items include, for example, sit to For each of the test variables, equilibrium, falls, Berg
stand, standing unsupported, standing with eyes closed, balance, gait, UPDRS, strength, and ROM, the analy-
standing with feet together, turning 360 degrees, and sis of variance (ANOVA) technique with repeated mea-
standing on one foot. sures was performed (SPSS, Statistical Package for the
Social Sciences 11.0). The statistical design was a
2.4.4. United Parkinson’s Disease Rating Scale mixed model 3 × 3 (group by pre/post/follow-up test)
There are six categories in the UPDRS: 1) menta- with repeated measures on the last factor. When signif-
tion, behavior, and mood, 2) activities of daily liv- icant differences were found, a follow-up Tukey Hon-
ing, 3) motor examination, 4) complications of ther- estly Significant Difference Test (HSD) was used to
apy (in the past 7 days), 5) clinical fluctuations, and compare individual means. For the fall numbers, a non-
6) “other complications”. Items are rated on a scale parametric Kruskal-Wallis Test was performed. All
of 0 (no impairment) to 4 (sustained or severe impair- tests were performed at the 0.05 level of significance.
ment). There are four items in the category menta-
tion, behavior, and mood; 13 items in activities of daily
living; 14 items in the motor examination; four items
3. Results
of complications of therapy; three clinical fluctuations
items; and three other complications items, for a to-
tal 41 items. For example the motor exam includes: There was 99% adherence to the program. Partici-
rigidity, finger taps time, rapid alternating movements, pants attended even when they did not feel their best.
312 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

Sensory Organization Test/Composite Score

Condition Main Effect
Mean Sensory Organization Composite Score

80

70

60

50
treadmill alone unloaded loaded

INTERVENTION GROUPS

Fig. 1. The intervention condition main effect for the sensory organization test composite score.

3.1. Balance (Kruskal Wallis = 9.154, p = 0.01) but the treadmill

alone group still had a significantly higher number of
3.1.1. NeuroCom Sensory Organization Test falls compared to the unloaded group on the posttest.
There was a significant intervention group main ef- There were no group differences on the follow-up test.
fect, F (2, 14) = 4.27, p = 0.03, effect size = 0.38,
power = 0.65, see Fig. 1. The Tukey HSD post hoc 3.2. Berg balance test
test (HSD = 19.09) determined that the unloaded group
was significantly higher than the treadmill alone group The main effect of pretest, posttest, and follow-up
and no other comparisons were significant. What is not test was significant, F (2, 36) = 12.37, p = 0.000,
evident from this result is that the treadmill alone group effect size = 0.41, power = 0.99. A Tukey (HSD =
had a beginning pretest score that was much lower 1.643) test showed that the posttest was significantly
better than the pretest Berg scores and that subjects
(mean = 48.83) than the other groups (unloaded pretest
did not get significantly worse at follow-up, see Fig. 3.
mean = 75; loaded pretest mean = 69.3) and that they
Also, the follow-up scores were significantly better than
improved considerably due to the intervention (posttest
the pretest scores. The group main effect was not sig-
mean = 54.83). While this comparison is evident from
nificant and neither was the interaction of group with
observation, it was not statistically significant.
pretest, posttest, follow-up.
3.1.2. Number of falls on SOT 3.3. United Parkinson’s Disease Rating Scale
The number of falls that subjects experienced on the (UPDRS)
SOT protocol was also analyzed using a Kruskal Wal-
lis non-parametric procedure, see Fig. 2. There were 3.3.1. Mentation, behavior, and mood
significant differences between groups on the pretest The 4 items rated on mentation, behavior, and
showing that the treadmill alone group had a larger mood improved significantly from pretest to posttest,
number of falls than the other two groups (Kruskal Wal- F (2, 38) = 5.74, p = 0.007, effect size = 0.23, power
lis = 6.042, p = 0.049). There were also significant = 0.84, Tukey HSD = 0.414, see Fig. 4. Scores con-
differences between groups on the posttest showing tinued to improve somewhat after the walking program
that the treadmill alone group and the unloaded group ended but they did not change significantly from the
decreased their number of falls after the intervention posttest scores.
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 313

Number of Falls on SOT

5

4
Mean Number of Falls

Intervention Groups
1
treadmill alone

0
unloaded

-1 loaded
Pretest Posttest Follow-up

TESTS

Fig. 2. Mean number of falls on the Sensory Organization Test (SOT) for each of the intervention groups over pre, post, and follow-up tests.
= pre and posttest differences, = treadmill alone group is significantly higher than the other two groups on the posttest.

Berg Test Scores

54

53
Mean Berg Scores

52

51

50

49
Pretest Posttest Follow-up

Fig. 3. Mean Berg Balance Test scores for each of the pretests, posttests, and follow-up tests.

3.3.2. Motor examination ter the walking program ended for four weeks. An im-
The pre to posttest scores improved significantly, provement of 9 percent in motor function due to walk-
F (2, 38) = 4.07, p = 0.025, effect size = 0.18, power ing 3 days per week for 6 weeks is meaningful. The
= 0.69, Tukey = 1.07. Figure 5 shows the significant Motor Examination scores also have a significant differ-
improvement in Motor Examination scores for all sub- ence between the loading group and the treadmill alone
jects from pretest to posttest with no further change af- group, F (2, 19) = 3.75, p = 0.042, effect size = 0.28,
314 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

Mentation, Behavior, & Mood

UPDRS
3.1

3.0
Mean Rating (0-4 scale)

2.9

2.8

2.7

2.6

2.5

2.4
Pretest Posttest Follow-up test

Fig. 4. United Parkinson’s Disease Rating Scale (UPDRS) mentation, behavior, and mood significant improvement from pretest to posttest, and
follow-up test.

Motor Examination
UPDRS
14.6

14.4
Mean Rating (0-4 scale)

14.2

14.0

13.8

13.6

13.4

13.2

13.0
Pretest Posttest Follow-up test

Fig. 5. Motor examination scores improved significantly from pretest to posttest for all subjects on the United Parkinson’s Disease Rating Scale
(UPDRS).

power = 0.61, Tukey HSD = 5.80. Figure 6 shows the compared with the treadmill alone group, they were
significantly lower (better) Motor Examination scores not significantly better. This comparison averages all
for those who were in the loading condition compared scores on the pre/post/follow-up and compares groups.
with the treadmill alone subjects. While the unloading It could be asked if the treadmill alone group was just
condition was better on the motor examination scores worse or higher in motor exam scores initially than the
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 315

Motor Examination
UPDRS
18

17
Mean Rating (0-4 scale)

16

15

14

13

12

11

10
control unloading loading

Treatment Conditions

Fig. 6. Motor examination scores are significantly better (lower) over all tests in those who were in the loading condition than for those in the
treadmill alone condition.

other two groups. A one-way ANOVA was used to the supporting leg held the weight from pre to post test
make this comparison and answer the question. There and that decrease was maintained over the time until
were no significant differences between the treadmill the follow-up test, see Fig. 7. If velocity of walking
alone, unloading, and loading groups on their pretest had increased due to the interventions, then one would
motor examination scores. Therefore, Fig. 6 shows us have an explanation for a decrease in the percent of
that those in the loading condition were better overall, time spent in left foot single support. While there were
for the pre/post/follow-up combined, than the treadmill no significant differences in velocity, it is apparent that
alone group. There were no other significant differ- the treadmill alone group walked faster from a pretest
ences for the other categories (activities of daily living, of 111 cm/sec to a posttest of 118 cm/sec. The loaded
complications of therapy (in the past week), clinical group also increased their velocity from 127 cm/sec to
fluctuations, and other complications) of the UPDRS. 130.9 cm/sec. The unloaded group, however, decreased
in velocity from 131.7 cm/sec to 128.2 cm/sec from pre
3.4. Gait to posttest. This averages to an increase of 7.4 cm/sec
for all groups from pre to posttest. It may be that
The ten gait variables were stride length left and right this overall increase in gait velocity resulting from the
feet, heel to heel base of support left and right feet, interventions, while not significant, is the reason why
single support time left and right feet, double support there is a significant decrease in percent time spent in
time left and right feet, velocity and cadence. For left leg single support from pre to posttest.
single support time on the left foot (in percent of total There was one significant effect for cadence (steps
gait cycle), there was a test main effect, F (2, 32) = per minute) and that was a group by test interaction,
3.44, p = 0.04, effect size = 0.18, power = 0.60, F (4, 36) = 2.72, p = 0.04, effect size = 0.23, power
n = 19, see Fig. 7. There were two outliers from = 0.69, n = 21, see Fig. 8. The figure shows that
the treadmill alone group identified by SPSS and those all three groups were similar in cadence on the pretest
were eliminated from this one analysis. The means and but that after the intervention the unloaded and loaded
standard deviations were pre = 40.9 % of gait cycle groups became slightly faster. On the follow-up both
(2.03), post = 40.07% of gait cycle (1.99), follow-up = unloaded and loaded groups produced a remarkably
40.03 % of gait cycle (2.86). For all groups there was a faster cadence. Conversely, the treadmill alone group
significant decrease in percent time of the gait cycle that slowed their cadence considerably on the follow-up in
316 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

GAIT--Test Main Effect

Single Support Time (% of Gait Cycle)
Mean Single Support Time Lf. (% gait cycle)

Left Foot
41.0

40.8

40.6

40.4

40.2

40.0

39.8

39.6
Pretest Posttest Follow-up Test

TESTS

Fig. 7. Gait test main effect for single support time (percent of gait cycle) on the left leg.

Group by Test Interaction 3.4.1. Gait analyses with all groups combined
Gait Cadence Since there were no consistently significant differ-
ences among the groups on many of the gait variables,
125 all groups were combined for these next analyses. After
Mean Cadence Time (sec)

comparing only the tests for all subjects (no interven-

120 tion group separation), there is a significant increase in
Treadmill
stride length for the left and right leg (Fig. 9). Only
115
Alone one figure for the right leg is presented because both
Unloaded
110 legs mirror one another. The 6-week interventions of
Loaded treadmill alone, unloading, and loading combined pro-
105 duced a significant increase in stride length for all sub-
jects. This is the most promising of all the gait effects
100 found for this study. There is a significant main effect
Pretest Posttest Follow- due to all three interventions from pre to posttest for
up Test stride length on the right leg and on the left leg. There
is a significant main effect of test from pre to posttest
Fig. 8. The interaction of group with pre/post/follow-up tests for gait
cadence. with a decrease in percent time the left leg holds the
weight (F (1, 22) = 4.06, p = 0.05, effect size = 0.16,
comparison to the other two groups. It is this opposite power = 0.49) see Fig. 10. The intervention has served
effect of the treadmill alone group on the follow-up to decrease percent time in single support for all sub-
that created the significant interaction. When referring jects (pretest M = 40.85%, SD = 2.35; posttest M =
back to the single support right foot result, it may be 40.15%, SD = 1.99). From post to follow-up test,
that the treadmill alone group had lower percent right there was no significant decline in any of the groups.
single support because they had slowed their cadence
on the follow-up test. When this happened, they were 3.4.2. Strength
in double support for a longer period of time in the Since strength training was not a component of
gait cycle and this helped them to maintain stability. this study, there were no expectations of changes in
There were no significant effects for the other eight gait strength. We used the strength measures to try to de-
variables. termine the cause of potential changes in balance and
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 317

Pre and Post Test

Stride Length Right Leg
132
Mean Stride Length Rt.Leg (cm)

131

130

129

128

127

126

125
Pretest Posttest

Pre and Post Test

Fig. 9. Pre and posttest for all subjects on stride length right leg.

Pre and Post Test for All Subjects

Single Support Left Leg
41.0
Mean Single Support Lf. Leg (%)

40.8

40.6

40.4

40.2

40.0
Pretest Posttest

PREPOST

Fig. 10. Single support left leg pre and posttest for all subjects.

gait. When an ANOVA was used to test each of the F (2, 32) = 7.30, p = 0.002, effect size = 0.31,
eight strength variables on each leg (16 total analyses), power = 0.91. Tukey’s HSD post hoc test (HSD =
there were no significant changes in any of the vari- 2.64) showed that the posttest was significantly higher
ables. This result rules out any significant changes in (greater ROM, M = 7.5 deg) than the pretest (M = 3.5
strength as potential causal mechanisms for change in deg). While participants decreased some in ROM after
balance and gait. the four-week detraining period, they did not decrease
significantly (M = 6.7 deg).
3.4.3. Range of Motion (ROM) The ROM for the left leg knee extension was signifi-
The ROM for the left ankle dorsiflexion produced cant for the pre/post/follow-up main effect, F (2, 34) =
a significant main effect of pre/post/follow-up test, 3.70, p = 0.035, effect size = 0.18, power = 0.64.
318 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

Tukey’s HSD post hoc test (HSD = 0.550) showed that 3.5.4. UPDRS
there was a significantly lower ROM in left knee ex- The 4 items rated on mentation, behavior, and mood
tension (participants could not fully extend their left significantly improved from pretest to posttest for all
knees) during the pretest (M = −0.55 deg), but at subjects. The Motor Examination scores also im-
posttest they could fully extend their left knees (M = 0 proved significantly for all subjects with no further
deg). There was no significant change at the follow-up change after 4 weeks of no intervention. There also
test (M = 0 deg). were significantly lower (better) averaged scores over
pre/post/follow-up tests for the Motor Examination cat-
egory for those who were in the loading condition com-
3.5. Summary of results pared with the treadmill alone subjects. While tread-
mill alone subjects were higher on the pretest Motor
3.5.1. Balance Sensory Organization Test (SOT) Examination scores, they were not significantly higher
The composite score for the SOT was significantly than the loading group. Therefore, it can be summa-
better for the unloaded group than the treadmill alone rized that the loading condition seemed to lead to better
group and no other comparisons were significant. What Motor Examination scores on posttest and follow-up
is not evident from this result is that the treadmill alone tests compared with the treadmill alone condition.
group had a beginning pretest score that was much
lower (M = 48.83) than the other groups and that they 3.5.5. Strength
improved considerably due to the intervention (posttest There were no significant changes in any of the eight
M = 54.83). On the other hand, the unloaded group strength measures for each leg.
started at M = 75.00 and they did not improve (posttest
M = 75.2). This interaction was not statistically sig- 3.5.6. ROM
nificant. While the unloaded group was higher overall, Of the five ROM measures on each leg, left ankle
they had high scores before the intervention and made dorsiflexion and left leg knee extension improved sig-
no improvement due to the intervention. The treadmill nificantly from pre to post test with no significant de-
alone group, however, had a much lower composite cline to baseline values after the detraining period.
balance score and they did improve due to the inter-
vention. While there is no significance between condi- 3.5.7. Gait
tions and pre/post/follow-up times, it is quite apparent On the posttest all intervention groups spent signif-
that the treadmill alone subjects benefited considerably icantly less percent time of the gait cycle supporting
from the intervention whereas the unloaded subjects their body weight on the left leg. In addition, the av-
did not. erage increase in gait velocity was 2.5 cm/sec for all
groups from pre to posttest. Walking on the treadmill
was beneficial for all subjects in that they spent less time
3.5.2. SOT falls supporting their weight and gained in step velocity.
Both the unloaded and treadmill alone groups de- All three groups were similar in cadence on the
creased their falls on the SOT conditions after the inter- pretest but after the intervention the unloaded and
vention, but the treadmill alone group remained signif- loaded groups became slightly faster. On the follow-up
icantly higher in number of falls (posttest M = 3.5 for both unloaded and loaded groups produced a remark-
treadmill alone vs 0 for the unloaded group). Again, ably faster cadence. Conversely, the treadmill alone
one needs to consider that the unloaded group only had group slowed their cadence considerably on the follow-
a mean of 0.6 falls on the pretest. Thus, both groups up in comparison to the other two groups. It is this op-
improved equally (M improvement = 0.6 falls). posite effect of the treadmill alone group on the follow-
up that created a significant interaction. When referring
back to the single support right foot result, it may be
3.5.3. Berg balance test that the treadmill alone group had lower percent right
The posttest was significantly better than the pretest single support because they had slowed their cadence
Berg scores for all subjects and they did not get signif- on the follow-up test. When this happened, they were
icantly worse at follow-up. Also, the follow-up scores in double support for a longer period of time in the gait
were significantly better than the pretest scores. cycle and this helped them to maintain stability.
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 319

There were no significant effects for the other eight treadmill walking was beneficial for all subjects to the
gait variables. However, when groups were col- extent that they spent less time supporting their weight
lapsed and another analysis was done on just the on one leg during gait and gained in step velocity.
pre/post/follow-up data, very important results oc- Stride length also significantly increased from pre to
curred. Stride length significantly increased from pre posttest. The six-week interventions of treadmill alone,
to posttest for all groups. The six-week interventions unloading, and loading combined to produce signifi-
of treadmill alone, unloading, and loading combined to cant stride length increases of 4.5 cm. Lastly, left leg
produce significant stride length increases of 4.5 cm. single support of the gait cycle was also reduced from
Since one of the major functional problems in Parkin- pre to posttest. Subjects were standing on this leg for
sonism is shuffling gait, it is an excellent result to see a shorter period of time, thus balance on this leg was
important increases in stride length in just 6 weeks time. improved, stride length improved, and gait velocity in-
Furthermore, stride length decreased over the 4 weeks creased. Overall, there were a number of important
time after the intervention, but these were not signifi- benefits to walking and balance gained from each of
cant decreases. Lastly, left leg single support of the gait the interventions.
cycle was also reduced from pre to posttest. Subjects The above changes cannot be explained by strength
were standing on this leg for a shorter period of time, since strength of lower limbs did not change signifi-
thus balance on this leg was improved, stride length cantly. This is consistent because of the previous ob-
improved, and gait velocity increased. In conclusion servation that without resistance training individuals
for gait, there were many important benefits to walking with Parkinsonism cannot expect to see improvements
gained from each of the interventions. in lower limb strength [20]. In addition, only two of
the 10 ROM variables (five on each leg), left ankle dor-
siflexion and left leg knee extension, improved for all
4. Discussion participants. While these two changes may have served
to improve balance and gait, we would expect both legs
Our study aims were: 1) to demonstrate that patients to have ROM changes in order to see the magnitude of
with advancing Parkinsonism have the capacity to par- balance and gait improvements seen in this study. One
ticipate in physical therapy programs utilizing both as- might speculate that all gait and balance improvements
sisted and weight bearing (unweighting) and/or addi- were simply a result of forcing better attention to gait
tional weight weighting, 2) to determine whether as- and balance by walking on the treadmill no matter what
sisted weight bearing or additional weight bearing is condition the subject used. However, this rationale is
more beneficial to the improvement of function and not a logical explanation for the improvements seen be-
increased stability in gait and dynamic balance in pa- cause subjects did not treadmill walk for one month af-
tients with parkinsonism, compared with matched con- ter the intervention, thus they did not attend to treadmill
trols (treadmill alone), 3) to demonstrate a reduction in gait parameters for one month. Nevertheless, improve-
the fall frequency in testing conditions that have sway ments were still maintained. A more likely reason for
referenced support that predictably result in disequi- improvement appears to be related to aspects of motor
librium, 4) to determine if treadmill conditioning im- control that are not attention based.
proves both strengthening of the lower limb and con- We suggest that treadmill walking has effected
sequently balance in patients, and 5) to demonstrate change in the efferent commands and/or function of
whether patients undergoing such training would show spinal pattern generators rather than due to attention or
capacity for retention benefits. any peripheral sensory signal either at a local (i.e. spinal
The composite score for the SOT was significantly cord) or higher level. We attempted to manipulate sen-
better for the unloaded group than the treadmill alone sory signals by loading and unloading the body during
group. The treadmill alone group had a beginning walking. One group wore a weighted belt that was 5%
pretest score that was much lower than the other groups of body weight, which we hypothesized to enhance the
and they improved considerably due to the interven- sensory signal from Golgi tendon organs during weight
tion. Overall, the study demonstrated significant im- bearing, thus affecting the efferent signal for agonist
provement in several parameters. The Berg balance contraction. The other group, the unloaded group, was
test improved significantly for all groups and falls were hypothesized to improve walking due to repetitive and
reduced in the treadmill alone and unloaded conditions improved function of spinal pattern generators while at
on destabilizing SOT balance conditions. In addition, the same time, decreasing the sensory signal from Golgi
320 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

tendon organ due to the unloading. This hypothesized rotransmitters, neurogenesis within the basal ganglia,
reduction of sensory signal was suggested by Dietz et and/or elevated function of spinal pattern generators it
al. [9] and Dietz and Colombo [8] who showed reduc- can be said that those more affected by Parkinsonism
tion of gastrocnemic amplitude over each of the 25% can produce the greatest neural and functional balance
to 75% body load reductions. Neither of our loaded or and gait changes.
unloaded groups improved in SOT balance or gait to Lastly, the UPDRS measures of mentation, behav-
the extent of the treadmill alone group. Therefore, it ior, and mood as well as the motor examination im-
appears that sensory manipulation is not the cause of proved significantly from pre to posttest for all subjects.
improvement. The items that assess mentation, behavior, and mood
Improvements in SOT Balance, Berg Balance, UP- include intellectual impairment, thought disorder, de-
DRS (Motor Exam), and gait for all groups leads us pression, and motivation/initiative. The mechanisms
to believe that neuromuscular regulation can be facil- of change for some of these behaviors are, most cer-
itated in all Parkinson’s individuals no matter what tainly, benefits from neurotransmitters including sero-
treadmill intervention is employed. It may be that tonin [22,23,29].
the efferent commands are better able to reduce co- What about the retention of these important behav-
contraction through the mechanisms of inhibition of ioral functions? All subjects were asked to refrain from
antagonist muscle groups and increased contraction or treadmill walking at home or in fitness facilities for 4
facilitation of the agonist muscles. How can reduction weeks after the interventions stopped. They were then
in co-contraction, greater antagonist inhibition, and tested after the 4-week lay-off. Results showed that im-
increased agonist firing occur due to treadmill walk- provements were either maintained fully or there was
ing? Three mechanisms seem possible: first, sensori- minimal decline of function in all intervention groups.
motor training may have been neuroprotective as has All measured variables of SOT balance, Berg balance,
been shown during slow degeneration of nigrostriatal gait, ROM, and UPDRS that showed significant im-
dopaminergic neurons [32]. Second, neurogenesis may provements were maintained or declined minimally
occur in basal ganglia cells leading to stronger signals over the 4 weeks.
from basal ganglia, to thalamus, and then to motor Suggestions for clinical management include: 1) in-
cortex where inhibition and activation commands are corporating treadmill walking for at least 3 times per
generated. Animal studies support neurogenesis where week, 20 minutes per day, for patients with Parkinson-
brain damage has occurred followed by forced use in ism, and 2) treadmill walking alone without loading or
the affected limb [24,30,35]. Three, spinal pattern unloading is equally beneficial so it is recommended to
generators (SPGs) may have been positively affected provide treadmill walking alone since less equipment is
by treadmill walking. Purely spinal mechanisms, i.e. needed and it is less effortful than wearing a weighted
spinal pattern or step generators, may be responsible belt.
for support in sustained repetitive walking in healthy In conclusion, it is clear that all of the study aims
humans [25]. The spinal pattern generators may pro- were supported with modification to #2. We: 1) demon-
duce automatic walking independent of higher central strated that patients with advancing Parkinsonism have
nervous system control. The basic ingredient for al- the capacity to participate in physical therapy pro-
ternate discharge of two SPGs is their tendency to fire grams utilizing both assisted weight bearing (unload-
in bursts and to inhibit each other so that only one is ing) and/or additional weight bearing (loading), 2) de-
active at a time. This produces contraction/inhibition termined that assisted weight bearing or additional
patterns within muscle groups [25]. Through repetitive weight bearing was not more beneficial to the improve-
use the SPGs may have been facilitated in our subjects. ment of function and increased stability in gait and
While the unloading group was significantly better dynamic balance in patients with Parkinsonism, com-
in SOT balance overall, it was the treadmill alone group pared with matched controls (treadmill alone) but that
that improved more than either two groups. Those sub- treadmill walking, in general, is beneficial to improve-
jects in the treadmill alone group were considerably ment of function and increased stability in gait and dy-
compromised on the pretest compared to the other two namic balance, 3) demonstrated a reduction in the falls’
groups. From this we might infer those most severely frequency in testing conditions that predictably result
affected by Parkinsonism benefited the most on SOT in disequilibrium, 4) demonstrated that treadmill con-
balance from treadmill walking. If the mechanism of ditioning did not improve strengthening of the lower
change induced by walking is increased dopamine neu- limbs and consequently the changes in balance and gait
 T. Toole et al. / The effects of loading and unloading treadmill walking on balance 321

had to be the result of other causal mechanisms, and [15] J.M.R. Furman, R.W. Baloh, K. Barin, T.C. Hain, S. Herdman,
5) demonstrated that patients undergoing such training H.R. Konrad and S.W. Parker, Assesment: Posturography,
Neurology 43 (1993), 1261–1264.
showed capacity for retention benefits. [16] S. Hesse, M. Konrad and D. Uhlenbrock, Treadmill walking
with partial body-weight support versus floor walking in hemi-
paretic subjects, Archives of Physical Medicine Rehabilitation
80 (1999), 421–427.
Acknowledgement [17] M.M. Hoehn and M.D. Yahr, Parkinsonism: Onset, progres-
sion, and mortality, Neurology 17 (1967), 427–442.
This research was funded by The Tallahassee Memo- [18] M.A. Hirsch, R.A. Rider, T. Toole and H.V.B. Hirsch, Falls
prevention in individuals with Parkinson’s Disease: Part I: The
rial Health Care Foundation, 2001–2003. importance of balance and resistance training, Forum of Sport,
Physical Education & Recreation for Those with Disabilities
14(3) (1998), 15–20.
[19] M.A. Hirsch, R.A. Rider, T. Toole and H.V.B. Hirsch, Falls
References prevention in individuals with Parkinson’s Disease: Part II:
Developing a balance and resistance training program, Forum
[1] J.B. Bergen, T. Toole, R.G. Elliott, B. Wallace, K. Robinson of Sport, Physical Education & Recreation for Those with
and C.G. Maitland, Aerobic exercise intervention improves Disabilities 14(4) (1998), 21–26.
aerobic capacity and movement initiation in Parkinson’s Dis- [20] M.A. Hirsch, T. Toole, C.G. Maitland and R.A. Rider, The ef-
ease patients, NeuroRehabilitation 17(2) (2002), 161–168. fects of balance training and high-intensity resistance training
[2] D. Brunt, D.W. Vander Linden and A.L. Behrman, The relation on persons with idiopathic Parkinson’s Disease, Archives of
between limb loading and control parameters of gait initiation Physical Medicine and Rehabilitation 84 (2003), 1109–1117.
in persons with stroke, Archives of Physical Medicine and [21] G.A. Horstmann and V. Dietz, A basic posture control mech-
Rehabilitation 76 (1995), 627–634. anism: The stabilization of the centre of gravity, Electroen-
[3] B.A. Conway, H. Hultborn and O. Kiehn, Proprioceptive input cephalography and Clinical Neurophysiology 76 (1990), 165–
resets central locomotor rhythm in the spinal cat, Experimental 176.
Brain Research 68 (1987), 643–656. [22] B.L. Jacobs and C.A. Fornal, Serotonin and motor activity,
[4] E. Daubney and E.G. Culham, Lower-extremity muscle force Current Opinions in Neurobiology 7(6) (1997), 820–825.
and balance performance in adults aged 65 years and older, [23] B.L. Jacobs, H. Praag and F.H. Gage, Adult brain neurogen-
Physical Therapy 79 (1999), 1177–1185. esis and psychiatry: a novel theory of depression, Molecular
[5] P.J. Delwaide and M. Gonce, Pathophysiology of Parkin- Psychiatry 5(3) (2000), 262–269.
son’s signs, in: Parkinson’s Disease and Movement Disorders, [24] T.A. Jones and T. Schallert, Use-dependent growth of pyrami-
J. Jankovic and E. Tolosa, eds, Williams & Wilkins, 1998, dal neurons after neocortical damage, Journal of Neuroscience
pp. 159–175. 14 (1994), 2140–2150.
[6] P.J. Delwaide, J.L. Pepin and A. Martens de Noordhout, Short- [25] M.L. Latash, Neurophysiological Basis of Movement, Human
latency autogenic inhibition in patients with parkinsonian Kinetics, Champaign, IL 1998.
rigidity, Annals of Neurology 30 (1991), 83–89. [26] M.E. Morris, R. Iansek, T.A. Matyas and J.J. Summers, Stride
[7] V. Dietz, Interaction between central programs and afferent length regulation in Parkinson’s disease: Normalization strate-
input in the control of posture and locomotion, Journal of gies and underlying mechanisms, Brain 119 (1996), 551–568.
Biomechanics 29 (1996), 841–844. [27] NeuroCom International, Inc., Equitest System Data Interpre-
[8] V. Dietz and G. Colombo, Influence of body load on the gait tation Manual, Clackamas, Oregon, 1992.
pattern in Parkinson’s Disease, Movement Disorders 13(2) [28] R. Pahwa and W.C. Koller, Gait disorders in Parkinsonism and
(1998), 255–261. other movement disorders, in: Gait Disorders of Aging: Falls
[9] V. Dietz, K.L. Leenders and G. Colombo, Leg muscle acti- and Therapeutic Strategies, J.C. Masdeu, L. Sudarsky and L.
vation during gait in Parkinson’s disease: influence of body Wolfson, eds, Lippincott-Raven, 1997, pp. 209–220.
unloading, Electroencephalography and Clinical Neurophys- [29] L.E. Rueter, C.A. Fornal and B.L. Jacobs, A critical review
iology 105 (1997), 400–405. of 5-HT brain microdialysis and behavior, Reviews in Neuro-
[10] V. Dietz, M. Schubert and M. Trippel, Visually induced desta- science 8(2) (1997), 117–137.
bilization of human stance: neuronal control of leg muscles, [30] T. Schallert and T.A. Jones, Exhuberant neuronal growth after
NeuroReport 3 (1992), 449–452. brain damage in adult rats: the essential role of behavioral
[11] V. Dietz, W. Zijlstra, C. Assaiante, M. Trippel and W. Berger, experience, Journal of Neural Transplant Plasticity 4 (1993),
Balance control in Parkinson’s disease, Gait and Posture 2 193–198.
(1993), 77–84. [31] G.E. Stelmach, Basal ganglia impairment and force control, in:
[12] V. Dietz, W. Zijlstra, T. Prokop and W. Berger, Leg muscle Tutorials in Motor Neuroscience, J. Requin and G.E. Stelmach,
activation during gait in Parkinson’s disease: adaptation and eds, Kluwer, Dordrecht, 1991, pp. 137–148.
interlimb coordination, Electroencephalography and Clinical [32] J.J. Tillerson, S.L. Castro, M.J. Zigmond and T. Schallert,
Neurophysiology 97 (1995), 408–415. Motor rehabilitation of forelimb use in a unilateral 6-OHDA
[13] J. Duysens and K.G. Pearson, Inhibition of flexor burst gener- rat model of Parkinson’s disease, Society of Neuroscience Ab-
ation by loading ankle extensor muscles in walking cats, Brain stracts 672 (1998), 1720.
Research 187 (1980), 321–332. [33] T. Toole, M.A. Hirsch, A. Forkink, D.A. Lehman and C.G.
[14] K. Fouad, C.M. Bastiaanse and V. Dietz, Reflex adaptations Maitland, The effects of a balance and strength training pro-
during treadmill walking with increased body load, Experi- gram on equilibrium in Parkinsonism: A preliminary study,
mental Brain Research 137(2) (2001), 133–140. NeuroRehabilitation 14 (2000), 165–174.
322 T. Toole et al. / The effects of loading and unloading treadmill walking on balance

[34] T. Toole, S.B. Park, M.A. Hirsch, D.A. Lehman and C.G. Mait- [36] M. Visintin, H. Barbeau, N. Korner-Bitensky and N.E. Mayo,
land, The multicomponent nature of equilibrium in persons A new approach to retrain gait in stroke patients through body
with Parkinson’s disease: A regression approach, Journal of weight support and treadmill stimulation, Stroke 28(6) (1998),
Neural Transmission 103 (1996), 561–580. 1122–1128.
[35] H. Van Prag, G. Kempermann and F.H. Gage, Running in- [37] M. Wirz, G. Colombo and V. Dietz, Long term effects of
creases cell proliferation and neurogenesis in the adult mouse locomotor training in spinal humans, Journal of Neurology,
dentate gyrus, Natural Neuroscience 2 (1999), 266–270. Neurosurgery & Psychiatry 71(1) (2001), 93–96.