Original Article

Ann Rehabil Med 2017;41(1):34-41

pISSN: 2234-0645 • eISSN: 2234-0653
https://doi.org/10.5535/arm.2017.41.1.34 Annals of Rehabilitation Medicine

Improved Gait Speed After Robot-Assisted Gait

Training in Patients With Motor Incomplete
Spinal Cord Injury: A Preliminary Study
Seungwon Hwang, MD1, Hye-Ri Kim, MD1, Zee-A Han, MD, PhD1, Bum-Suk Lee, MD, MPH1,
Soojeong Kim, MD1, Hyunsoo Shin, MD1, Jae-Gun Moon, MD1, Sung-Phil Yang, BSPH1,
Mun-Hee Lim, BSPH1, Duk-Youn Cho, MS2, Hayeon Kim, MS2, Hye-Jin Lee, MD1
1
Department of Rehabilitation Medicine, 2Translational Research Center for Rehabilitation Robots,
National Rehabilitation Center & Hospital, Seoul, Korea

Objective To evaluate the clinical features that could serve as predictive factors for improvement in gait speed
after robotic treatment.
Methods A total of 29 patients with motor incomplete spinal cord injury received 4-week robot-assisted gait
training (RAGT) on the Lokomat (Hocoma AG, Volketswil, Switzerland) for 30 minutes, once a day, 5 times a week,
for a total of 20 sessions. All subjects were evaluated for general characteristics, the 10-Meter Walk Test (10MWT),
the Lower Extremity Motor Score (LEMS), the Functional Ambulatory Category (FAC), the Walking Index for Spinal
Cord Injury version II (WISCI-II), the Berg Balance Scale (BBS), and the Spinal Cord Independence Measure
version III (SCIM-III) every 0, and 4 weeks. After all the interventions, subjects were stratified using the 10MWT
score at 4 weeks into improved group and non-improved group for statistical analysis.
Results The improved group had younger age and shorter disease duration than the non-improved group. All
subjects with the American Spinal Injury Association Impairment Scale level C (AIS-C) tetraplegia belonged to
the non-improved group, while most subjects with AIS-C paraplegia, AIS-D tetraplegia, and AIS-D paraplegia
belonged to the improved group. The improved group showed greater baseline lower extremity strength, balance,
and daily living function than the non-improved group.
Conclusion Assessment of SCIM-III, BBS, and trunk control, in addition to LEMS, have potential for predicting the
effects of robotic treatment in patients with motor incomplete spinal cord injury.

Keywords Spinal cord injuries, Locomotion, Robotics, Rehabilitation, Clinical trial

Received March 10, 2016; Accepted May 19, 2016

Corresponding author: Hye-Jin Lee
Department of Rehabilitation Medicine, National Rehabilitation Center & Hospital, 58 Samgaksan-ro, Gangbuk-gu, Seoul 01022, Korea. Tel: +82-2-901-
1634, Fax: +82-2-990-8473, E-mail: Ludin80@naver.com
ORCID: Seungwon Hwang (http://orcid.org/0000-0001-8217-4714); Hye-Ri Kim (http://orcid.org/0000-0003-4431-4549); Zee-A Han (http://orcid.
org/0000-0002-8070-2218); Bum-Suk Lee (http://orcid.org/0000-0002-0506-0403); Soojeong Kim (http://orcid.org/0000-0003-3951-6947); Hyunsoo
Shin (http://orcid.org/0000-0002-9699-9132); Jae-Gun Moon (http://orcid.org/0000-0002-3892-2751); Sung-Phil Yang (http://orcid.org/0000-0002-
1988-8236); Mun-Hee Lim (http://orcid.org/0000-0001-9902-0826); Duk-Youn Cho (http://orcid.org/0000-0002-5170-5030); Hayeon Kim (http://orcid.
org/0000-0002-1251-1102); Hye-Jin Lee (http://orcid.org/0000-0001-9400-6283).
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/
licenses/by-nc/4.0) which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright © 2017 by Korean Academy of Rehabilitation Medicine
 Robot-Assisted Gait Training for Patients With Subacute Motor Incomplete Spinal Cord Injury

INTRODUCTION Most studies considered the Lower Extremity Motor

Score (LEMS) as the primary outcome in examining the
Spinal cord injury (SCI) often results in complete or in- effects of Lokomat, since it is the most important factor in
complete paralysis, affecting the ability to walk and par- predicting future gait function in patients with iSCI [10-
ticipate in physical activity. People with SCI have a sed- 12]. However, several studies reported that LEMS alone
entary lifestyle causing various complications including did not provide sufficient information on gait function
compromised cardiovascular function, pressure ulcers, in patients with iSCI [13,14]. Moreover, there has been
and osteoporosis [1]. These factors increase morbidity no report in Korea on whether patients with iSCI showed
and mortality in SCI. Thus, physicians have focused on improvement in gait quality and speed with Lokomat.
treatments to improve motor function in SCI [2]. Accordingly, the primary outcome was the 10-Meter
In the past, treatment to improve motor function in SCI Walk Test (10MWT), which reflects gait quality, including
was performed by physical therapists using conventional speed and agility. Patients were divided into improved
physical therapy, and focused on stretching, strength- and non-improved groups following robotic treatment.
ening, and manually-assisted gait training. However, in Clinical features that could serve as predictive factors for
recent decades, partial body weight support treadmill improvement in gait speed after robotic treatment were
training (PBWSTT) has been used to improve gait func- investigated.
tion in patients with neurological impairment. Studies
in patients with stroke indicated that PBWSTT could MATERIALS AND METHODS
achieve improved gait speed, balance, and motor recov-
ery [3,4]. PBWSTT also enabled early initiation of gait Subjects
training after injury, consolidation of weight-bearing ac- We evaluated all SCI patients who were admitted to the
tivities, and stepping and balance training by using task- National Rehabilitation Center between March 2013 and
specific approaches [5]. Rehabilitation strategies using February 2015. Patients who were eligible were invited to
conventional PBWSTT offer much benefit, but have some participate in the study. The inclusion criteria were (1)
difficulties because great effort is required from a single iSCI defined as the American Spinal Injury Association
or even two physical therapists for one patient, and result Impairment Scale (AIS) levels C and D [15], (2) subacute
in less reproducible gait patterns and shorter duration of (1 to 6 months since injury) stage patients with iSCI [5],
therapy. Thus, a mechanical system that provides PBW- (3) minimum 19 years of age, and (4) able to walk inde-
STT without the need for a physical therapist, i.e., robot- pendently before injury. The exclusion criteria were prior
assisted gait training (RAGT), was created [6,7]. experience with RAGT; severe skeletal problems such as
Lokomat (Hocoma AG, Volketswil, Switzerland) is one recent fractures, rigidity, or severe spasticity of the lower
of the most widely used RAGT Lokomat systems. It in- limbs; skin problem including pressure ulcers of the low-
cludes a treadmill, a body-weight support system using er limbs or coccyx area; other neurological disorders af-
a harness, and two light-weight robotic arms that attach fecting gait; medical complications such as uncontrolled
to and swing the subject’s legs. The patient puts on the cardiac disorders; pregnant or breast-feeding females;
harness and two robotic arms and walks on the treadmill. severe cognitive and/or communicative disorders; or
The patient can also watch a monitor that provides in- other problems that made it impossible to properly ac-
formation on progress with gait, stride, velocity, weight- complish the training.
bearing, and other parameters. We evaluated general characteristics in all subjects in-
The Lokomat was first introduced in 2003, and has been cluding age, sex, diagnosis, injury type (tetraplegia and
used to improve motor function in patients with SCI [8]. paraplegia), injury cause (traumatic and non-traumatic),
Several studies suggested that RAGT was effective for im- and time since injury. Non-traumatic injury causes in-
proving gait ability in patients with motor incomplete SCI cluded space-occupying lesions, spinal cord infarction,
(iSCI); and one study in Korea also proposed that RAGT spinal cord abscess, and arteriovenous fistula.
combined with conventional physical therapy (PT) could Participants were informed on the purpose, protocol,
improve gait ability in patients with iSCI [5,8,9]. and beneficial or harmful effects of the study and provid-

www.e-arm.org 35
Seungwon Hwang, et al.

ed written consent. The study was approved by the Na- performed at all (0) to perfectly performed (4), for a total
tional Rehabilitation Center Institutional Review Board score of up to 56 [19]. A higher score means better bal-
(IRB No. NRC-2013-02-015). ance.
Activities of daily living were measured using the Spinal
Measurements Cord Independence Measurement version III (SCIM-
All subjects were assessed before (0 week) and after III). SCIM-III is a disability scale for evaluating functional
training (4 weeks). change in patients with SCI [20]. The total SCIM-III score
The 10MWT was measured immediately after comple- ranges from 0 to 100, and includes the following sub-
tion of RAGT (4 weeks). Improvement following RAGT scales: self-care (SCIM-III-S, range 0–20), respiration and
was defined as an increase in gait speed ≥0.13 m/s on sphincter management (SCIM-III-R, 0–40), and mobility
the 10MWT [16]; a patient who was not able to perform (SCIM-III-M, 0–40).
the assessment at 0 weeks, but improved enough to per-
form the assessment at 4 weeks, was also considered Robot-assisted gait training protocol
improved. Otherwise, subjects were classified as non- The Lokomat system includes a treadmill, a support
improved. Lam et al. [16] reported that the standard error system with a harness, two light-weight robotic arms at-
for the measure of the 10MWT was 0.05 m/s; based on tached to the legs, and a monitor showing step length,
this, the calculated smallest real difference of the 10MWT gait velocity, and other parameters.
within the 95% confidence interval was 0.13 m/s. Thus, All subjects received both RAGT on the Lokomat and
0.13 m/s difference in the 10MWT is required for clini- conventional PT for 30 minutes, once a day, 5 times a
cally significant difference between data. In this study, week, for each training method, for a total of 20 sessions
we used this value to determine actual improvement in of RAGT.
the patient’s 10MWT result. At the beginning of treatment, the speed of the tread-
Lower extremity motor function, gait ability, balance, mill was 1.0 km/hr to 1.5 km/hr without incline, and was
and daily living functions were analyzed in order to iden- gradually increased as tolerated based on the rating of
tify the factors associated with the 10MWT. perceived exertion (RPE). The target RPE was 13–15 [21].
LEMS was used to evaluate motor function. LEMS is the Support by the harness began at 50% of the subject’s
sum of bilateral lower extremity key muscle power, rang- weight and was gradually decreased after they could walk
ing from total paralysis (0) to normal active movement safely for 30 minutes.
with full range of motion against gravity and full resis-
tance (5), with a total possible score of 50 [15]. Statistical analysis
To evaluate gait ability, we used the measurement tools Descriptive statistics were used to analyze the demo-
of the Functional Ambulatory Category (FAC) and the graphic and injury data of all subjects. Independent t-test
Walking Index for Spinal Cord Injury version II (WISCI-II). was used to evaluate the age and time after injury, and
The FAC score ranged from 0 to 5—unable to walk (0), de- crosstab analysis was used to evaluate the injury data in
pendency in gait (1 or 2), gait on even and level surfaces each group. We used Fisher exact test since the catego-
without manual contact with another person except for ries with expected frequency below 5 were over 20%. The
safety, requires stand-by guarding, or the need for verbal independent t-test was used for analysis of baseline func-
cuing to complete the task (3), and independent gait over tion in each group. A receiver operating characteristic
15 meters irrespective of aids used (4 or 5) [17]. WISCI-II (ROC) curve was used to deduct the cut-off value of base-
measures gait status based on the requirements for as- line measurements for prediction of improvement after
sistance and/or bracing and/or walking aids [18]. WISCI- RAGT.
II ranges from 0 to 20—grade 0, neither able to stand nor All statistical analyses were conducted using SPSS ver.
walk; and grade 20, able to walk over 10 m without walk- 20 for Windows (IBM, Armonk, NY, USA). Data were
ing aids, brace, or assistance. presented as mean±standard deviation unless otherwise
The Berg Balance Scale (BBS) was used to evaluate bal- stated. A p-value<0.05 was considered statistically signifi-
ance. The BBS consists of 14 items, each ranging from not cant.

36 www.e-arm.org
 Robot-Assisted Gait Training for Patients With Subacute Motor Incomplete Spinal Cord Injury

RESULTS (p=0.032). However, paralysis type showed no statistical

difference between both groups. All subjects with AIS-
Subject characteristics C tetraplegia belonged to the non-improved group. No
Between March 2013 and February 2015, a total of 82 significant difference was found in injury cause in both
patients with subacute iSCI were screened and 32 were groups (Table 2).
eligible for the study. Three dropped out after initiation
of the study: one subject wanted to quit the study, an- Baseline function
other dropped out because of poor health not associated For analysis of the difference in baseline functions,
with RAGT, and the other dropped out because of loss of we compared baseline assessments (0 week) between
contact. A total 29 subjects were finally included (Fig. 1). the improved and non-improved groups. The improved
Demographic and injury data of all subjects were pre- group showed greater basal lower extremity strength
sented in Table 1. The average age was 49.9 years and
all subjects were males. Nine subjects had AIS-C and 20
had AIS-D. Nineteen subjects had tetraplegia and 10 had Table 1. General characteristics (n=29)
paraplegia. There were 22 subjects with traumatic and 7 Value
with non-traumatic SCI. Average time since injury were Age (yr) 49.9±11.4
15.2 weeks (Table 1). Sex (male) 29 (100)
AIS
Demographic data Level C 9 (31.0)
After training (4 weeks), subjects were divided into im- Level D 20 (69.0)
proved group (18 subjects) and non-improved group (11 Paralysis type
subjects) based on the 10MWT (Fig. 1). In the improved Tetraplegia 19 (65.5)
group, 9 of 18 subjects were patients who had not accom- Paraplegia 10 (34.5)
plished 10MWT during the pre-test but carried out the Cause of injury
10MWT at post-test. The improved group had younger Trauma 22 (75.9)
average age and shorter average time since injury than Non-trauma 7 (24.1)
the non-improved group (46.7 years, 13.8 weeks and Time after injury (wk) 15.2±4.6
55.2 years, 17.4 weeks, respectively). Result from the Values are presented as mean±standard deviation or num­
Fisher exact test indicated a significant difference in AIS ber (%).
scale between the improved and non-improved groups AIS, American Spinal Injury Association impairment scale.

Screened
(n=82) Fig. 1. Between March 2013 and
Excluded out for February 2015, a total of 82 pa-
inclusion/exclusion criteria
(n=50) tients with subacute iSCI (incom-
Entered into study plete spinal cord injury) were
(n=32)
Discontinued intervention
screened, of which 32 were eli-
(n=3) gible for the study. Three dropped
Reasons: out after initiation of the study:
Refused to participating, medical problem
or loss of contact one subject voluntarily quit the
Stratified into a 4-week
10MWT score for analysis study, another due to poor health
(n=29) not associated with robot-assisted
gait training, and the other due to
loss of contact. A total 29 subjects
Assigned to improved group Assigned to non-improved group
were finally included and divided
(10MWT improved >0.13 m/s or first (10MWT improved <0.13 m/s or not into improved (n=18) and non-
achieved during 4 weeks) performed during 4 weeks) improved groups (n=11). 10MWT,
(n=18) (n=11)
10-Meter Walk Test.

www.e-arm.org 37
Seungwon Hwang, et al.

Table 2. General characteristics of the experimental groups

Improved group (n=18) Non-improved group (n=11) p-value
Age (yr) 46.7±11.3 55.2±9.8 0.048*
AIS 0.032*
Level C 3 (16.7) 6 (54.5)
Level D 15 (83.3) 5 (45.5)
Paralysis type 0.149
Tetraplegia 10 (55.6) 9 (81.8)
Paraplegia 8 (44.4) 2 (18.2)
Cause of injury 0.139
Trauma 12 (66.7) 10 (90.9)
Non-trauma 6 (33.3) 1 (9.1)
Time after injury (wk) 13.8±4.4 17.4±4.4 0.044*
Values are presented as mean±standard deviation or number (%).
AIS, American Spinal Injury Association impairment scale.
*p<0.05.

Table 3. Baseline functional abilities at start of RAGT

Improved group Non-improved group p-value
LEMS 30.1±6.3 22.6±9.6 0.018*
WISCI-II 10.2±7.6 5.7±8.3 0.150
FAC 2.3±1.4 1.2±1.5 0.060
BBS 26.6±17.0 12.5±15.1 0.033*
SCIM-III 62.3±25.1 34.7±22.4 0.006**
SCIM-III-S 11.1±8.0 3.2±7.0 0.011*
SCIM-III-R 33.3±6.8 26.2±7.2 0.013*
SCIM-III-M 17.9±13.4 5.4±10.4 0.014*
RAGT, robot-assisted gait training; LEMS, Lower Extremity Motor Score; WISCI-II, Walking Index for Spinal Cord In-
jury version II; FAC, Functional Ambulatory Category; BBS, Berg Balance Scale; SCIM-III, Spinal Cord Independence
Measurement version III; SCIM-III-S, self-care subscale; SCIM-III-R, respiration and sphincter control subscale;
SCIM-III-M, mobility subscale.
*p<0.05, **p<0.01.

with a mean LEMS of 30.1, as compared to 22.6 in the ROC curve

non-improved group (p=0.018); the improved group also Based on the results in Table 3, ROC curve analysis for
showed greater change in baseline balance with a mean LEMS, BBS, SCIM-III, SCIM-III-S, SCIM-III-R, and SCIM-
BBS of 26.6, as compared to 12.5 in the non-improved III-M showed significant findings (p<0.05).
group (p=0.033). The improved group showed higher The values that had both high sensitivity and high
total SCIM-III score than the non-improved group (62.3 specificity were 22.5 in LEMS (sensitivity 88.9%, specific-
vs. 34.7; p=0.006); even when divided into subscales, the ity 45.5%), 7 in BBS (sensitivity 94.4%, specificity 63.6%),
improved group showed higher scores than the non-im- and 37 in SCIM-III (sensitivity 83.3%, specificity 81.8%).
proved group with respect to SCIM-III-S, SCIM-III-R, and In the SCIM-III subscales, the values that had both high
SCIM-III-M (p<0.05). No significant differences in FAC sensitivity and high specificity were 1.5 in SCIM-III-S
and WISCI-II were found between groups (Table 3). (sensitivity 88.9%, specificity 81.8%), 27.5 in SCIM-III-R
(sensitivity 88.9%, specificity 63.6%), and 5.5 in SCIM-III-
M (sensitivity 83.3%, specificity 81.8%).

38 www.e-arm.org
 Robot-Assisted Gait Training for Patients With Subacute Motor Incomplete Spinal Cord Injury

DISCUSSION endurance, and use of assistive devices in patients with

iSCI [25], and was highly associated with gait ability in
In this study, improvement in gait speed with 10MWT patients with chronic SCI [32]. Also, patients with iSCI
as the primary outcome were used to evaluate the effects have slower gait speed in order to maintain the center of
of robotic treatment from a new perspective as compared mass within the base of support [33]. Balance thus has a
to prior reports [5,8,9]. In addition to LEMS and WISCI- significant influence on gait speed in patients with iSCI.
II, various assessment tools such as FAC, BBS, and SCIM- The improved group had greater improvement in gait
III were used for the ROC curve. To our best knowledge, speed possibly due to better baseline balance than the
this is a novel approach to identifythe cut-off values for non-improved group.
baseline functions that can predict the effects of robotic When measuring the ROC curve for BBS, the cut-off val-
treatment. ue was 7 points. The total possible score of the BBS test is
The 10MWT, Timed Up and Go test (TUG), and 6-Min- 56, and the tasks become more difficult and challenging
ute Walk Test (6MWT) are used to evaluate gait ability in for patients with iSCI in the latter parts of the test [25].
subjects with iSCI quantitatively [22]. The 10MWT and Consequently, the BBS score of 7 points was achieved
TUG are used more than the 6MWT because they are from the categories of sit to stand, stand to sit, and static
quick and easy to perform [23,24]. We used the 10MWT standing balance or transfer [19].
as the primary outcome because it is safer and easier Baseline SCIM-III score was higher in the improved
than the TUG, and does not have a ceiling effect [22,25]. group than the non-improved group (p=0.006); more-
Age, time since injury and AIS scale were previously over, significant differences in SCIM-III-S, SCIM-III-R,
identified as factors that significantly influence the recov- and SCIM-III-M were observed (p<0.05).
ery of gait ability following iSCI [26-30]. Our study con- When the ROC curve was measured for SCIM-III-M
curred with the above findings, which showed that the alone, 5.5 was the cut-off value. SCIM-III-M measures
members of the improved group were younger and had categories of bed mobility, transfer, and indoor/outdoor
shorter time since injury than those in the non-improved gait, and low scores for this assessment were seen for bed
group (p<0.05). Additionally, subjects with AIS-D showed mobility and transfers [20]. Consequently, a BBS higher
greater improvements than those with AIS-C. However, than 7 and SCIM-III-M higher than 6 means that patients
there was no statistically significant difference in paraly- with some degree of static standing balance and transfer,
sis type between the two groups. Interestingly, 5 patients or at least some trunk control, showed improvement in
with AIS-C tetraplegia showed no improvement, which gait speed after robotic treatment.
can be explained in terms of trunk control, which is an Balance is also essential in daily life. SCI patients need
important factor that affects gait [31]. The patients with static or dynamic balance for sitting, wheelchair manipu-
AIS-C tetraplegia showed no improvement after robotic lation, and daily living skills [34]. Thus, the difference
treatment, possibly because they had poorer trunk con- in balance contributed to the difference in SCIM-III-S
trol as compared to other patients with iSCI. between the improved and non-improved groups. Ulti-
LEMS is the most important factor in predicting gait mately, patients who had better baseline balance showed
ability in patients with iSCI, as confirmed in our study greater improvement in gait speed.
[10-12]. A previous study indicated that the prospects for In this study, the baseline score of SCIM-III-R also
gait improvement are poor if LEMS is ≤20 and good if ≥30 showed significant differences in each group. However, it
[10]. In this study, ROC curve analysis showed that LEMS is difficult to interpret this result. Respiration components
of 23 was the cut-off value to predict the effects of RAGT, in SCIM-III-R scored maximum for all subjects. Thus,
which corresponded to previous studies. Thus, LEMS ≥23 other components of self-care such as intermittent cath-
can predict improvement in gait speed after RAGT. eterization in urinary sphincter management, suppository
In our study, the baseline SCIM-III score was higher use in anal sphincter management, and toilet use that are
in the improved group than in the non-improved group. associated with balance including trunk control could
Balance was also a very important factor for gait ability in have made a meaningful difference in each group.
patients with iSCI. Balance affected gait speed, posture, Unlike LEMS baseline scores, there were no differences

www.e-arm.org 39
Seungwon Hwang, et al.

in FAC and WISCI-II between the two groups. This is Center for Rehabilitation Robots (Grant No. NRCTR-
contradictory to the concept that gait speed should be IN16001).
the primary outcome. However, FAC was scored using
only 6 levels, and did not adjust for assistance or braces REFERENCES
and walking aids; therefore, FAC did not reflect subtle
changes in gait ability [17]. In contrast, WISCI-II classi- 1. McKinley WO, Jackson AB, Cardenas DD, DeVivo MJ.
fied 21 levels, with more detail than FAC. However, ac- Long-term medical complications after traumatic
cording to Lam et al. [16], 17% of patients experienced a spinal cord injury: a regional model systems analysis.
ceiling effect and 53% of patients showed flooring effects Arch Phys Med Rehabil 1999;80:1402-10.
in WISCI-II. In this study, 4 of 29 subjects had 0 point and 2. Ditunno JF Jr, Burns AS, Marino RJ. Neurological and
2 of 29 had 20 points in both pre- and post-tests. In addi- functional capacity outcome measures: essential to
tion, WISCI-II score was biased because of a predisposi- spinal cord injury clinical trials. J Rehabil Res Dev
tion toward setting the scores around a specific score. 2005;42:35-41.
According to Morganti et al. [35], score of WISCI-II was 3. Finch L, Barbeau H, Arsenault B. Influence of body
frequently concentrated on scores of 13, 16 and 20, as weight support on normal human gait: development
seen in our study. These factors could have resulted in no of a gait retraining strategy. Phys Ther 1991;71:842-55.
significant difference in WISCI-II. 4. Visintin M, Barbeau H, Korner-Bitensky N, Mayo
Our study had several limitations as follows: (1) the NE. A new approach to retrain gait in stroke patients
small number of subjects, (2) all subjects received both through body weight support and treadmill stimula-
conventional PT and RAGT, which makes it difficult to tion. Stroke 1998;29:1122-8.
determine whether the improvement is solely from RAGT 5. Schwartz I, Sajina A, Neeb M, Fisher I, Katz-Luerer M,
or also from the conventional PT. Meiner Z. Locomotor training using a robotic device
Recently, studies on robotic treatment for patients with in patients with subacute spinal cord injury. Spinal
iSCI have been actively conducted and effects have been Cord 2011;49:1062-7.
demonstrated. LEMS is an important factor for indepen- 6. Jezernik S, Colombo G, Keller T, Frueh H, Morari M.
dent gait for patients with iSCI. However, balance is also Robotic orthosis lokomat: a rehabilitation and re-
an important factor because it reflects not only LEMS search tool. Neuromodulation 2003;6:108-15.
but also trunk control ability and proprioception. As 7. Hornby TG, Zemon DH, Campbell D. Robotic-as-
seen in this study, patients with better balance showed sisted, body-weight-supported treadmill training in
greater improvements in gait speed. Our study is the first individuals following motor incomplete spinal cord
to show that patients with BBS scores above 7 had higher injury. Phys Ther 2005;85:52-66.
improvement ratio in gait speed. Therefore, baseline bal- 8. Alcobendas-Maestro M, Esclarin-Ruz A, Casado-
ance can be viewed as an important factor in predicting Lopez RM, Munoz-Gonzalez A, Perez-Mateos G,
improvement in gait speed after robotic treatment. In the Gonzalez-Valdizan E, et al. Lokomat robotic-assisted
future, assessment of SCIM-III, BBS, and trunk control, in versus overground training within 3 to 6 months of
addition to LEMS, will be helpful in predicting the effects incomplete spinal cord lesion: randomized controlled
of robotic treatment in patients with iSCI. trial. Neurorehabil Neural Repair 2012;26:1058-63.
9. Shin JC, Kim JY, Park HK, Kim NY. Effect of robotic-as-
CONFLICT OF INTEREST sisted gait training in patients with incomplete spinal
cord injury. Ann Rehabil Med 2014;38:719-25.
No potential conflict of interest relevant to this article 10. Waters RL, Adkins R, Yakura J, Vigil D. Prediction of
was reported. ambulatory performance based on motor scores de-
rived from standards of the American Spinal Injury
ACKNOWLEDGMENTS Association. Arch Phys Med Rehabil 1994;75:756-60.
11. Zörner B, Blanckenhorn WU, Dietz V; EM-SCI Study
This study was supported by the Translational Research Group, Curt A. Clinical algorithm for improved pre-

40 www.e-arm.org
 Robot-Assisted Gait Training for Patients With Subacute Motor Incomplete Spinal Cord Injury

diction of ambulation and patient stratification af- sons. J Am Geriatr Soc 1991;39:142-8.
ter incomplete spinal cord injury. J Neurotrauma 24. Graham JE, Ostir GV, Fisher SR, Ottenbacher KJ. As-
2010;27:241-52. sessing walking speed in clinical research: a system-
12. Curt A, Dietz V. Ambulatory capacity in spinal cord in- atic review. J Eval Clin Pract 2008;14:552-62.
jury: significance of somatosensory evoked potentials 25. Lemay JF, Nadeau S. Standing balance assessment in
and ASIA protocol in predicting outcome. Arch Phys ASIA D paraplegic and tetraplegic participants: con-
Med Rehabil 1997;78:39-43. current validity of the Berg Balance Scale. Spinal Cord
13. Wirz M, van Hedel HJ, Rupp R, Curt A, Dietz V. Muscle 2010;48:245-50.
force and gait performance: relationships after spinal 26. Geisler FH, Dorsey FC, Coleman WP. Recovery of mo-
cord injury. Arch Phys Med Rehabil 2006;87:1218-22. tor function after spinal-cord injury: a randomized,
14. Shin JC, Yoo JH, Jung TH, Goo HR. Comparison of placebo-controlled trial with GM-1 ganglioside. N
lower extremity motor score parameters for patients Engl J Med 1991;324:1829-38.
with motor incomplete spinal cord injury using gait 27. Burns SP, Golding DG, Rolle WA Jr, Graziani V, Di-
parameters. Spinal Cord 2011;49:529-33. tunno JF Jr. Recovery of ambulation in motor-
15. Kirshblum SC, Burns SP, Biering-Sorensen F, Donovan incomplete tetraplegia. Arch Phys Med Rehabil
W, Graves DE, Jha A, et al. International standards for 1997;78:1169-72.
neurological classification of spinal cord injury (re- 28. Scivoletto G, Morganti B, Ditunno P, Ditunno JF, Mo-
vised 2011). J Spinal Cord Med 2011;34:535-46. linari M. Effects on age on spinal cord lesion patients’
16. Lam T, Noonan VK, Eng JJ; SCIRE Research Team. rehabilitation. Spinal Cord 2003;41:457-64.
A systematic review of functional ambulation out- 29. Scivoletto G, Tamburella F, Laurenza L, Torre M, Mo-
come measures in spinal cord injury. Spinal Cord linari M. Who is going to walk? A review of the factors
2008;46:246-54. influencing walking recovery after spinal cord injury.
17. Holden MK, Gill KM, Magliozzi MR. Gait assessment Front Hum Neurosci 2014;8:141.
for neurologically impaired patients. Standards for 30. Piepmeier JM, Jenkins NR. Late neurological changes
outcome assessment. Phys Ther 1986;66:1530-9. following traumatic spinal cord injury. J Neurosurg
18. Dittuno PL, Ditunno JF Jr. Walking index for spinal 1988;69:399-402.
cord injury (WISCI II): scale revision. Spinal Cord 31. Campagnolo DI, Kirshblum S, Nash MS, Heary RF,
2001;39:654-6. Gorman PH. Spinal cord medicine. 2nd ed. Philadel-
19. Berg KO, Wood-Dauphinee SL, Williams JI, Maki B. phia: Lippincott Williams & Wilkins; 2011. p. 123-4.
Measuring balance in the elderly: validation of an in- 32. Scivoletto G, Romanelli A, Mariotti A, Marinucci D,
strument. Can J Public Health 1992;83 Suppl 2:S7-11. Tamburella F, Mammone A, et al. Clinical factors
20. Catz A, Itzkovich M, Tesio L, Biering-Sorensen F, that affect walking level and performance in chronic
Weeks C, Laramee MT, et al. A multicenter interna- spinal cord lesion patients. Spine (Phila Pa 1976)
tional study on the Spinal Cord Independence Mea- 2008;33:259-64.
sure, version III: Rasch psychometric validation. Spi- 33. Lemay JF, Duclos C, Nadeau S, Gagnon D, Desrosiers
nal Cord 2007;45:275-91. E. Postural and dynamic balance while walking in
21. Braddom RL. Physical medicine and rehabilitation. adults with incomplete spinal cord injury. J Electro-
4th ed. Philadelphia: Elsevier/Saunders; 2010. p. 408- myogr Kinesiol 2014;24:739-46.
9. 34. Chen CL, Yeung KT, Bih LI, Wang CH, Chen MI, Chien
22. van Hedel HJ, Wirz M, Dietz V. Assessing walking abil- JC. The relationship between sitting stability and
ity in subjects with spinal cord injury: validity and functional performance in patients with paraplegia.
reliability of 3 walking tests. Arch Phys Med Rehabil Arch Phys Med Rehabil 2003;84:1276-81.
2005;86:190-6. 35. Morganti B, Scivoletto G, Ditunno P, Ditunno JF, Mo-
23. Podsiadlo D, Richardson S. The timed “Up & Go”: a linari M. Walking index for spinal cord injury (WISCI):
test of basic functional mobility for frail elderly per- criterion validation. Spinal Cord 2005;43:27-33.

www.e-arm.org 41