Robot-assisted gait training for patients with hemiparesis due to stroke.

Abstract

Robot-assisted devices are becoming a popular alternative to manual facilitation in stroke rehabilitation. These devices have the potential to reduce therapist burden and treatment costs; however, their effectiveness in terms of functional recovery remains in question. This pilot study compared the outcomes of a stroke rehabilitation program that incorporates robot-assisted gait training (RAGT) with a more traditional therapy program that does not. Twenty hemiparetic stroke patients were recruited at a rehabilitation hospital in Houston, Texas, and were randomly assigned to 2 groups. The control group (n = 10) received 24 1-hour sessions of conventional physical therapy, whereas the RAGT group (n = 10) received 24 1-hour sessions of conventional physical therapy combined with RAGT on a treadmill. Gait function was assessed before and after treatment by an 8-m walk test, a 3-minute walk test, and the Tinetti balance assessment. Both groups showed significant improvement in all 3 outcome measures following treatment (P < .05), but there was no difference between groups. It is concluded that RAGT may provide improvements in balance and gait comparable with conventional physical therapy. A larger multicenter trial is required to investigate the effectiveness of RAGT in hemiparetic stroke.

Full text

Page 1

269
Top Stroke Rehabil 2011;18(3):269–276
© 2011 Thomas Land Publishers, Inc.
www.thomasland.com
doi: 10.1310/tsr1803-269
Robot-Assisted Gait Training for Patients
with Hemiparesis Due to Stroke
Stanley Fisher, MD,1,2 Leah Lucas,2 and T. Adam Thrasher, PhD3
1Methodist Neurological Institute, Houston, Texas; 2HealthSouth Rehabilitation Hospital, Humble, Texas;
3University of Houston, Center for Neuromotor and Biomechanics Research, Houston, Texas
Robot-assisted devices are becoming a popular alternative to manual facilitation in stroke rehabilitation. These devices have
the potential to reduce therapist burden and treatment costs; however, their effectiveness in terms of functional recovery
remains in question. This pilot study compared the outcomes of a stroke rehabilitation program that incorporates robot-
assisted gait training (RAGT) with a more traditional therapy program that does not. Twenty hemiparetic stroke patients
were recruited at a rehabilitation hospital in Houston, Texas, and were randomly assigned to 2 groups. The control group
(n = 10) received 24 1-hour sessions of conventional physical therapy, whereas the RAGT group (n = 10) received 24
1-hour sessions of conventional physical therapy combined with RAGT on a treadmill. Gait function was assessed before
and after treatment by an 8-m walk test, a 3-minute walk test, and the Tinetti balance assessment. Both groups showed
signifi cant improvement in all 3 outcome measures following treatment (P < .05), but there was no difference between
groups. It is concluded that RAGT may provide improvements in balance and gait comparable with conventional physical
therapy. A larger multicenter trial is required to investigate the effectiveness of RAGT in hemiparetic stroke. Key words:
gait, hemiparesis, robot-assisted training, stroke, treadmill training
A
goals for many stroke patients, and signifi cant
improvements are possible through rehabilitation.
The current evidence indicates that intensive,
task-specifi c therapy produces the highest level of
recovery of motor function, even in cases of severe
impairment.1 Locomotor training is the process
of retraining gait through the repetitive execution
of assisted walking movements. The traditional
approach to locomotor training involves manual
assistance from therapists as patients walk on a
treadmill or overground.2 When applied at a high
enough intensity and for a suffi cient duration,
locomotor training has the potential to produce
significant, long-lasting improvements in gait
function.3 In recent years, there has been a lot of
attention focused on the development of advanced
therapeutic interventions involving technologies
such as functional electrical stimulation4 (FES)
and robotics.5 These interventions are designed
to facilitate walking and enable patients to engage
in gait retraining at the earliest point in their
rehabilitation program.
Currently, there are many treatment options
available for gait rehabilitation after stroke.6 Many
large proportion of strokes result in
hemiparesis and gait impairments. Recovery
of balance and walking function are key
of these have proved to be effective in clinical
studies; however, it is not clear that any particular
approach is superior for improving gait speed or
quality.3 The use of partial body weight support
to facilitate locomotor training became popular
in the mid-1990s. Many severely impaired stroke
patients have diffi culty bearing their total body
weight on the affected leg during the swing phase
of gait. This problem can be addressed by partially
relieving the patient’s body weight using an
overhead harness, thus compensating for weakness
of the weight-bearing muscles and allowing the
patient to focus on motor control during walking.
In stroke rehabilitation, body weight–supported
treadmill training (BWSTT) typically involves 1
or 2 therapists who provide manual assistance
to facilitate correct movement. Compared with
traditional locomotor training approaches, BWSTT
enables patients to walk longer and to take more
steps, thus signifi cantly increasing the number of
repetitions and functional gains.7 BWSTT has been
shown to be effective in improving gait function in
Grand Rounds
Elliot J. Roth, MD, Editor

Page 2

270 TOPICS IN STROKE REHABILITATION/MAY-JUNE 2011
study involving only 16 participants found no
signifi cant differences.10 In all 3 cases, the Lokomat
was the RAGT device in question.
In the present study, our goal was to compare the
outcomes of conventional physical therapy with
the outcomes of RAGT using the AutoAmbulator.
The personnel needs of a conventional physical
therapy program are signifi cantly higher than
for RAGT; however, it is still not clear to what
extent RAGT can improve functional recovery.
The purpose of this study was to determine the
feasibility of using the AutoAmbulator as a form
of RAGT and compare the functional outcome
measures with conventional physical therapy. We
hypothesized that patients who received RAGT as
part of their rehabilitation program would recover
more gait and balance function than patients
who received an equal dosage of conventional
physical therapy.
Methods
Participants
Participants were recruited from the inpatient
program at the HealthSouth Rehabilitation
Hospital in Humble, Texas. All new patients who
were admitted with hemiparesis due to a stroke
less than 12 months prior were considered eligible
and were referred to the investigators. Eligible
patients were then screened according to the
following criteria.
Inclusion criteria:
1. Age between 18 and 80 years
Exclusion criteria:
1. Orthopedic problems or muscle diseases that
impair mobility
2. Severe congestive heart failure with ejection
fraction of less than 30%
3. Unstable angina requiring medications
4. Intrathecal baclofen pump implantation
within 6 weeks
5. Evidence of active infection
6. Severe dementia or other cognitive impairment
preventing meaningful communication (as
determined by Mini-Mental State Examination
[MMSE])
7. Body weight more than 300 pounds (135 kg)
8. Pregnancy
both mild and severe hemiparesis due to stroke,
even in nonambulatory patients.1,8
BWSTT has also been shown to be effective in
the recovery of gait function. However, BWSTT
can require a large amount of effort from therapists;
in certain situations, 2 therapists are required per
patient to perform the BWSTT intervention.9,10
Rehabilitation robots are electromechanical
devices that provide external forces to the limbs
of patients to create normal kinematics during
the performance of tasks such as reaching11 and
walking.12,13 Robot-assisted gait training (RAGT)
is an alternative to therapist-assisted BWSTT that
is becoming increasingly popular. In RAGT, the
assistance provided by the robot is intended to
serve as an analog to manual assistance provided by
therapists. However, most of the robotic-assisted
devices discussed in the literature are designed to
provide constant guiding forces resulting in near-
normal kinematics and do not mimic the variable
assistance that a human therapist applies. Whether
the clinical benefi ts of RAGT are comparable
with therapist-assisted BWSTT is a vibrant topic
of debate among rehabilitation practitioners and
researchers.
RAGT devices currently come in 2 distinct
forms. First, there are the powered exoskeletons,
such as the Lokomat (Hocoma SA, Switzerland),14
the AutoAmbulator (HealthSouth, Houston, Texas,
USA),15 and the LOPES exoskeleton robot,16 which
attach in parallel to the lower limb segments and
move in unison with the patient. These devices
are typically mounted over treadmills and include
some form of body weight support. An alternative
to powered exoskeletons are RAGT devices that
use movable footplates to which the patient’s feet
are attached. Two examples are the Gait Trainer I5,17
and the HapticWalker.
Despite the number of RAGT devices that
have been developed in the past 10 years, there
have been only 3 randomized controlled studies
comparing RAGT with neurofacilitative physical
therapy or BSWTT. No signifi cant differences
were found in terms of walking function when
RAGT was compared with conventional physical
therapy.18 When RAGT was compared with
manually assisted BWSTT, 1 study involving 48
participants found that BWSTT resulted in greater
improvements in gait function,19 whereas another

Page 3

Robot-Assisted Gait Training for Stroke
271
Figure 1, the Autoambulator is a device consisting
of a treadmill, an overhead lift, a pair of articulated
arms, and 2 upright structures housing the
computer controls and parts of the mechanism.14
Individuals, when placed in the machine and
fi tted with a special harness, are attached to the
overhead lift and raised to a standing position
over the treadmill where weight-bearing can be
assessed. The articulated robot arms mounted to
the upright structures are hinged outward and are
mechanically driven for vertical adjustment.
The gait drive components are computer
controlled through position, time, and distance
to provide a smooth, accurate, and coordinated
movement of the legs and treadmill through
variable speeds. The control interface is through
a touch screen computer display that can collect,
process, display, and archive pertinent session
data. The device is equipped with safety interlocks,
redundant travel limits, variable torque limits, and
other means to detect and clear inordinate safety
situations.
The robot arms move with 4 degrees of freedom
corresponding to hip fl exion/extension and knee
flexion/extension bilaterally. The robot arms
Patients who satisfi ed all of the screening criteria
were invited to join the study and were required to
sign informed consent documents. All procedures
followed in this study were approved by the
Institutional Review Board of the HealthSouth
Rehabilitation Hospital.
Protocol
After consenting to enter the study, each
participant underwent a walking assessment
consisting of 3 components. First, participants
performed an 8-m walk test by walking for a distance
of 8 m in a wide, empty hallway. Participants were
allowed to use canes or walkers as necessary.
The time to complete this task was recorded. If
a participant could not complete the task in 4
minutes, the test was terminated and a score of
240 seconds was recorded. Next, in a 3-minute
walk test, participants were instructed to walk as
far as they could for 3 minutes, and the distance
traveled was recorded. Later in the day, participants
completed the Tinetti balance assessment20; their
total score (out of 28) was recorded.
After the baseline walking assessment, each
participant was randomly assigned to 1 of 2 groups.
A sequence of 20 one-digit numbers was selected
from a random number table by the research
coordinator, and participants were assigned in the
order of their enrollment. If the digit corresponding
to their enrollment was between 0 and 4, they
were assigned to the fi rst group, referred to as the
control group, and received 24 one-hour sessions
of goal-oriented physical therapy administered by a
highly trained neurorehabilitation team consisting
of a physical therapist and 2 physical therapy
assistants. These sessions consisted of stretching
and strengthening exercises of the affected lower
extremity, as well as overground walking exercises
using durable medical equipment such as canes
and walkers. During these exercises, therapists
used neurofacilitation techniques by applying
appropriate manual assistance when needed.
The second group, referred to as the RAGT
group, also received 24 one-hour sessions of
therapy. These sessions consisted of 30 minutes of
goal-oriented physical therapy (as per the control
group) followed by 30 minutes of RAGT using
the HealthSouth Autoambulator. Illustrated in
Figure 1. Autoambulator.

Page 4

272 TOPICS IN STROKE REHABILITATION/MAY-JUNE 2011
Results
Twenty stroke patients participated in this
study. Ten were assigned to the control group
and 10 to the RAGT group. There were no
statistical differences between groups in gender
(P = 1.000), age (P = .154), stroke duration (P =
.261), or baseline 8-m walk (P = .652), baseline
3-minute walk (P = .472), or baseline Tinetti
balance score (P = .409), indicating an equitable
randomization. Demographic and treatment data
for all participants are shown in Table 1.
All participants successfully completed 24
therapy sessions. Signifi cant improvements over
time were seen in both groups for the 8-m walk
(F1,18 = 14.02; P = .0015), the 3-minute walk
(F1,18 = 18.66; P = .0004), and the Tinetti score
(F1,18 = 53.55; P < .0001). However, the difference
between groups was not signifi cant for the 8-m
walk (F1,18 = 0.29; P = .5967), the 3-minute walk
(F1,18 = 0.63; P = .4385), or the Tinetti score
(F1,18 = 0.40; P = .5354). A summary of the changes
in outcome measures are given in Table 2.
provide assistance in the sagittal plane only. The
patient is physically connected to the robot via a
set of cuffs worn at the mid-thigh and calves. The
device produces symmetrical reciprocal gait by
providing forces throughout the entire gait cycle,
including swing phase. A handlebar is fi xed to the
chassis of the device in front of the patient, which
he or she can hold onto as though pushing a cart.
Treatment began immediately following
group assignment. All participants started the
study as inpatients and were scheduled for 5
therapy sessions per week with at least a 1-day
interval between sessions. Most participants were
discharged from inpatient care before completing
all 24 sessions. At discharge, participants were
moved to a residence outside of the hospital and
transferred immediately to an outpatient schedule
of 3 sessions per week. Participants completed the
study in roughly 6 to 8 weeks, according to the
duration of their inpatient stay, which depended
on factors not controlled by this study.
Following the completion of all treatment
sessions, participants repeated the 8-m walk test,
3-minute walk test, and Tinetti balance assessment.
Analysis
To ensure balanced randomization, we tested
differences between the control group and RAGT
group at baseline using a chi-square test for gender
and Mann-Whitney U tests for age, days post
stroke, and all baseline assessments (Tinetti balance
score, 3-minute walk test, and 8-m walk test). A
2 × 2 repeated measures analysis of covariance with
factors of treatment (RAGT vs conventional PT)
and time was used to identify effect of treatment
with respect to the outcome measures. Stroke
duration at the time of enrollment into the study
was used as a covariate. The level of signifi cance
for all statistical tests was set to ? = 0.05.
Table 1. Participant data
Characteristics
Control group
(n = 10)
RAGT group
(n = 10)
Gender
Male
Female
7
3
7
3
Age, mean (±SD),
years
60 (14) 60 (14)
Stroke duration,
mean (±SD), days
81 (106) 57 (73)
Side of hemiparesis 4L /6R 3L /7R
No. of sessions,
mean (±SD), as:
Inpatient
Outpatient
11 (5.0)
13 (5.0)
11 (6.5)
13 (6.5)
Note: L = left; R = right; RAGT = robot-assisted gait training.
Table 2. Outcome measures, mean (±SD)
Measure
RAGT groupControl group
BaselinePosttreatmentBaselinePosttreatment
8-m walk, s
3-minute walk, ft
Tinetti score
143 (105)
120 (157)
8.3 (8.4)
56 (73)
290 (340)
17.7 (9.9)
105 (95)
62 (82)
7.4 (8.5)
57 (65)
213 (166)
13.9 (7.7)
Note: RAGT = robot-assisted gait training.

Page 5

Robot-Assisted Gait Training for Stroke
273
the 8-m walk test after treatment. The general
trend of improvement following treatment can
be seen clearly in Figure 3, which illustrates the
baseline and posttreatment scores from the Tinetti
balance assessment. All participants improved
during the treatment period; however, there was
The baseline and posttreatment scores for the
8-m walk test are shown in Figure 2. At baseline, 5
participants in the RAGT group and 3 participants
in the control group were unable to walk 8 m
in 4 minutes. All of these participants except
for 1 in the RAGT group were able to complete
8m walk test
Baseline Post-treatmentBaseline Post-treatment
Time to complete (s)
0
50
100
150
200
250
300
RAGT groupControl group
Figure 2. Individual participant scores on the 8-m walk test for the
robot-assisted gait training (RAGT) group and control group.
Tinetti Balance Assessment
BaselinePost-treatmentBaselinePost-treatment
Tinetti Score
0
5
10
15
20
25
30
RAGT group Control group
Figure 3. Individual participant scores on the Tinetti balance
assessment for the robot-assisted gait training (RAGT) group and
control group.