UBC Faculty Research and Publications

Gait speed using powered robotic exoskeletons after spinal cord injury: a systematic review and correlational… Louie, Dennis R; Eng, Janice J; Lam, Tania Oct 14, 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12984_2015_Article_74.pdf [ 941.3kB ]
JSON: 52383-1.0361956.json
JSON-LD: 52383-1.0361956-ld.json
RDF/XML (Pretty): 52383-1.0361956-rdf.xml
RDF/JSON: 52383-1.0361956-rdf.json
Turtle: 52383-1.0361956-turtle.txt
N-Triples: 52383-1.0361956-rdf-ntriples.txt
Original Record: 52383-1.0361956-source.json
Full Text

Full Text

REVIEW Open AccessGait speed using powered roboticexoskeletons after spinal cord injury: asystematic review and correlational studyDennis R. Louie1,2,3, Janice J. Eng1,2,3,4,5*, Tania Lam1,4 and Spinal Cord Injury Research Evidence (SCIRE)Research TeamAbstractPowered robotic exoskeletons are an emerging technology of wearable orthoses that can be used as an assistivedevice to enable non-ambulatory individuals with spinal cord injury (SCI) to walk, or as a rehabilitation tool toimprove walking ability in ambulatory individuals with SCI. No studies to date have systematically reviewed theliterature on the efficacy of powered exoskeletons on restoring walking function. Our objective was to systematicallyreview the literature to determine the gait speed attained by individuals with SCI when using a powered exoskeletonto walk, factors influencing this speed, and characteristics of studies involving a powered exoskeleton (e.g. inclusioncriteria, screening, and training processes). A systematic search in computerized databases was conducted to identifyarticles that reported on walking outcomes when using a powered exoskeleton. Individual gait speed data from eachstudy was extracted. Pearson correlations were performed between gait speed and 1) age, 2) years post-injury, 3) injurylevel, and 4) number of training sessions. Fifteen articles met inclusion criteria, 14 of which investigated the poweredexoskeleton as an assistive device for non-ambulatory individuals and one which used it as a training intervention forambulatory individuals with SCI. The mean gait speed attained by non-ambulatory participants (n = 84) while wearing apowered exoskeleton was 0.26 m/s, with the majority having a thoracic-level motor-complete injury. Twelve articlesreported individual data for the non-ambulatory participants, from which a positive correlation was found betweengait speed and 1) age (r = 0.27, 95 % CI 0.02–0.48, p = 0.03, 63 participants), 2) injury level (r = 0.27, 95 % CI 0.02–0.48,p = 0.03, 63 participants), and 3) training sessions (r = 0.41, 95 % CI 0.16–0.61, p = 0.002, 55 participants). In conclusion,powered exoskeletons can provide non-ambulatory individuals with thoracic-level motor-complete SCI the ability towalk at modest speeds. This speed is related to level of injury as well as training time.IntroductionThe inability to walk is arguably one of the most notableimpairments that individuals experience after spinal cordinjury (SCI). Besides leading to physical complicationssuch as skin breakdown, muscle atrophy, reduced car-diorespiratory capacity, and pain [1], being unable towalk also affects psychological well-being and can in-crease the risk of depression and reduce quality of life[2]. For these reasons, recovery of walking consistentlyranks among the top priorities related to mobility forindividuals with SCI [3]. Unfortunately, a large propor-tion of these individuals with complete or incomplete in-jury have limited, if any, recovery of walking function andare thus limited to a wheelchair for their mobility [4]. Evenwith the use of conventional bracing for ambulation, indi-viduals with SCI must expend high levels of energy [5, 6]to achieve modest, non-functional gait speeds [6, 7],dependent on their level of injury [6].Recent developments in gait orthoses have producedthe powered robotic exoskeleton, a rechargeable bi-onic device worn over the lower extremities with mo-torized joints that can provide externally-powered gaitindependent of a treadmill system [8]. Compared totreadmill-based gait orthoses such as the Lokomat(Hocoma, Switzerland) and LOPES (University of Twente,* Correspondence: janice.eng@ubc.ca1University of British Columbia, Vancouver, Canada2Rehabilitation Research Program, 4255 Laurel Street, Vancouver, BC, Canada,V5Z 2G9Full list of author information is available at the end of the articleJ N E R JOURNAL OF NEUROENGINEERINGAND REHABILITATION© 2015 Louie et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Louie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 DOI 10.1186/s12984-015-0074-9Netherlands), these powered robotic exoskeletons arecompact, lightweight, and portable [9]. This new tech-nology has been designed as an assistive device toprovide individuals with complete paralysis the abilityto stand and walk independently over-ground in anatural, full weight-bearing, reciprocal pattern. Theycan also be used in the rehabilitation setting as atraining tool to improve stepping and weight-shiftingfor ambulatory individuals with SCI [9]. Various de-signs have been developed, several of which are com-mercially available and are in the process of beingapproved for use at home and in the community. Aswith any form of gait rehabilitation, walking with apowered exoskeleton requires specialized training andpractice.As a newly developed technology, the current evi-dence base surrounding the use of powered roboticexoskeletons in SCI rehabilitation consists of a num-ber of studies, but the majority are case studies(single-subject reports) or single-intervention trialswith a small number of participants. A recent system-atic review found that energy consumption was reducedwhen walking with powered orthoses compared to con-ventional orthoses in paraplegic SCI [10]. A literature re-view by the same author found that powered gait orthoseshave a beneficial effect on the kinematics and temporal-spatial parameters of gait, but reported minimally on gaitspeed [11]. To our knowledge, no systematic reviews havespecifically determined the gait speed attained by non-ambulatory individuals with SCI while using a poweredrobotic exoskeleton to walk. We defined non-ambulatoryindividuals with SCI as those who do not walk regularly,independently, with or without gait aids or bracing. Gaitspeed is an important indicator and will contribute to theutility of the device; very slow speeds may relegate the de-vice to uses solely for exercise, while faster speeds may en-able community ambulation.The primary objective of this article was to examinethe evidence on the ability of powered robotic exoskele-tons to provide gait, specifically focusing on gait speed,in individuals with SCI by performing a systematic re-view of relevant clinical studies. To provide continuityacross the studies and address the heterogeneity of thepresentation of individuals with SCI, we collected indi-vidual participant data from each study to explore corre-lations between participant characteristics and gaitspeed. We hypothesized that gait speed would be posi-tively correlated with spinal cord preservation (lesionlevel), as well as training time. Before acquiring a pow-ered robotic exoskeleton, clinicians and users alikeshould have an understanding of the feasibility of pow-ered exoskeleton use. Thus, secondary objectives were tosummarize the (1) screening process for determiningsuitability for an exoskeleton and the (2) training processto habituate an individual with SCI to walk with anexoskeleton.ReviewMethodsWe conducted this systematic review according to thePRISMA guidelines and the review protocol is availablefrom the authors [12]. We accessed MEDLINE (1946 toMay 6, 2015), EMBASE (1980 to May 6, 2015), CochraneCentral Register of Controlled Trials (1991 to May 6,2015), and CINAHL (1982 to May 6, 2015). An electronicdatabase search was first conducted using the terms“spinal cord injury” OR “SCI” OR “quadriplegia” OR“tetraplegia” OR “paraplegia” paired using AND with“walk” OR “walking” OR “gait” OR “ambulation”. Thesearch results were then paired using AND with “exoskel-eton” OR “exoskeletal” OR “powered gait orthosis” OR“PGO” OR “ReWalk” OR “Ekso” OR “indego” OR “hybridassistive limb” OR “HAL”. English language and humanstudies were used as restrictions. Hand searches of refer-ence lists from retrieved articles were completed. Titles,abstracts, and full-texts were screened by two independentreviewers; only studies that met inclusion criteria were se-lected and used for further analysis.Eligibility criteria were studies that evaluated walkingoutcomes of individuals with SCI after training with apowered robotic exoskeleton. We defined powered exo-skeleton as a multi-joint orthosis that uses an externalpower source to move at least two joints on each leg,which is portable, and can be used independent of atreadmill or body-weight support. Papers were selectedif they reported gait speed by use of relevant over-groundwalking tests (e.g. 10-Meter Walk Test) or temporal-spatial measures relevant to walking (step length, dis-tance, time walking). Additional inclusion criteria were:(1) adult patients over 18 years of age; and (2) peer-reviewed full articles published or “in press”. Exclusioncriteria were studies that only utilized a combination offunctional electrical stimulation (FES) with the exoskel-eton (hybrid exoskeleton), studies that only reportedjoint angle and muscle moments, or studies that utilizedpopulations with mixed diagnoses (e.g. SCI and stroke)and did not separate the results. Abstracts and confer-ence posters were excluded, as were studies that utilizedan orthosis powered only at one joint or a treadmill fortesting.Design characteristics, inclusion and exclusion criteria,sample demographics, exoskeleton characteristics, train-ing protocol, and adverse events were extracted fromeach study. Individual participant demographics andwalking data were extracted from studies, when avail-able, and compiled for statistical analysis. Gait speed wascalculated, when not directly reported, from walkingoutcomes such as the 10-Meter Walk Test (10MWT) orLouie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 2 of 10other timed measure. We used the Pearson product-moment correlation coefficient to determine relation-ships between common participant variables availablefrom the studies (age, years post-injury, injury level,number training sessions) and independent gait speed(hence, without assistance) while walking within the exo-skeleton device. We omitted individuals with motor-incomplete injuries from these calculations due to theheterogeneous presentation of incomplete SCI. Injurylevel was coded as 0 to 17, representing C4 to L1.ResultsAs illustrated in Fig. 1, our search results yielded 145 re-cords across the MEDLINE, CINAHL, EMBASE, andCochrane databases. After screening for eligibility, 15 ar-ticles [13–27] were included in this review (Table 1);seven eligible records were excluded from the finalcount due to overlapping of participant data. All but tworecords [14, 24] provided individual participant data thatcould be extracted for statistical analysis.Study designThe 15 studies ranged from single-subject case studies toprospective intervention trials comparing other types oforthoses within the study. Thirteen studies [14, 16–27]used the powered exoskeleton as an assistive device forambulation, and thus were post-test studies; in thesestudies, outcomes were only measured while wearingthe device, after a period of training, since individualsdid not have the ability to walk without the device(mostly participants with complete injuries). One study[13] used the powered exoskeleton as a training inter-vention to improve ambulation, assessing walking with-out the device in a pre-post study design in participantswith incomplete and low-complete ambulatory SCI.One study [15] used the powered exoskeleton as bothan assistive device as well as training intervention, asthey included motor complete and ambulatory incom-plete SCI. No studies included a control group and allparticipants received the powered exoskeleton as theirintervention. Two studies [14, 18] compared a poweredexoskeleton to standard rigid orthoses, where the sameparticipants trialed multiple orthoses.ExoskeletonsOf the 15 studies included, 12 studies [13, 15–23, 26, 27]used commercially developed exoskeletons, such asthe ReWalk (ReWalk Robotics, Israel), Ekso (EksoBionics, USA), and the Indego (Parker Hannifin Cor-poration, USA). Two studies investigated exoskeletonsdeveloped for research purposes: the Wearable PowerAssist Locomotor (WPAL) [25] and Mina [24]. Onestudy [14] utilized a custom device designed by theauthors in a previous study [28], an isocentric recipro-cating gait orthosis (IRGO) combined with electricallyactuated motors. All exoskeletons in the includedstudies were actuated at the hip and knee joints. Thecontrol of walking while wearing a powered exoskel-eton varies, with some exoskeletons having multiplecontrol options (Table 2).Fig. 1 Study results during PRISMA phases: a flowchart of selection process based on inclusion/exclusion criteriaLouie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 3 of 10Inclusion/exclusion criteriaTen of the 14 studies using the powered exoskeleton as anassistive device for non-ambulatory individuals with SCIincluded motor complete or complete SCI (AIS A/B),while three [15, 20, 26] included incomplete SCI (AIS C).One study [22] allowed any participant with lower extrem-ity weakness or paralysis to be eligible, and provided SCI-specific data. Seven studies [15–17, 20, 22, 26, 27]allowed cervical-level injuries to be eligible; the restof the studies either listed thoracic-level injury orbelow T1 to qualify for the inclusion criteria. Thir-teen studies [13, 15–23, 25–27] specified height andweight restrictions, generally within the range of 1.45 m to2.0 m and less than 113 kg. Time post-injury for inclusionvaried as well, when mentioned, with one study [21] set-ting a maximum time of two years post-injury, and fivestudies setting a minimum time post-injury of six months[16, 26, 27] or one year [15, 23]. Three studies [16, 24, 27]required participants to be a regular RGO user in order tobe included, and two [15, 22] required participants to beable to maintain an upright posture with or without astanding device.The study [13] using the powered exoskeleton solelyas a training tool for ambulatory individuals includedonly those with traumatic SCI to the conus medullaris/cauda equina with chronic incomplete or completeTable 1 Characteristics of studies included in the reviewAuthors Exoskeleton Use of theexoskeletonParticipants Walking outcomemeasuresTraining periodAach et al. (2014) [13] HAL Training tool 8 (AIS A to D, T8 to L2) 6MWT, 10MWT, TUG 5d/wk for 90 days,90 min per sessionArazpour et al. (2013) [14] Custom poweredIRGOAssistive device 5 (AIS A/B, T6 to T12) Gait speed, distance 3d/wk for 8 weeks,2 h per sessionBenson et al. (2015) [15] ReWalk Assistive device/Training Tool5 (AIS A/C), C7 to L1 6MWT, 10MWT, TUG 2d/wk for 10 weeks,2 h per sessionEsquenazi et al. (2012) [16] ReWalk Assistive device 12 (AIS A/B, T3 to T12) 6MWT, 10MWT 3d/wk for 8 weeks,75–90 min per sessionEvans et al. (2015) [17] Indego Assistive device 5 (AIS A, T6 to T12) 6MWT (self-selectedpace, fast pace)At least 5 sessionsFarris et al. (2014) [18] Indego Assistive device 1 (AIS A, T10) 6MWT, 10MWT, TUG 20 sessions in one yearFineberg et al. (2013) [19] ReWalk Assistive device 6 (AIS A/B, T1 to T11) Gait speed 3d/wk for up to 6 months,1–2 h per sessionHartigan et al. (2015) [20] Indego Assistive device 16 (AIS A to C, C5 to L1) 6MWT, 10MWT 5 sessions,90 min per sessionKolakowsky-Hayner et al.(2013) [21]Ekso Assistive device 7 (AIS A, T4 to T12) Walking distance, time 6d/wk for 1 week,up to 60 min per sessionKozlowski et al. (2015) [22] Ekso Assistive Device 7 (AIS A to C, C4 to L1) 2MWT, longest walk Up to 24 sessions, up to2 h per sessionKressler et al. (2013) [23] Ekso Assistive device 3 (AIS A, T1/2 to T9/10) Gait speed, distance 3d/wk for 6 weeks,60 min per sessionNeuhaus et al. (2011) [24] Mina Assistive device 2 (AIS A, T10 and T12) Gait speed 9 sessionsTanabe et al. (2013) [25] WPAL Assistive device 7 (AIS A/B, T6 to T12) Walking distance, time 2–11 sessions,60 min per sessionYang et al. (2015) [26] ReWalk Assistive Device 12 (AIS A to C, C8 to T11) 6MWT, 10MWT Up to 102 sessions,1–2 h per sessionZeilig et al. (2012) [27] ReWalk Assistive device 6 (AIS A/B, T5 to T12) 6MWT, 10MWT, TUG Until able to walk100 m unassistedHAL Hybrid Assistive Limb; 6MWT Six Minute Walk Test; 10MWT Ten Meter Walk Test; TUG Timed Up and Go Test; IRGO Isocentric Reciprocal Gait Orthosis; 2MWTTwo Meter Walk Test; WPAL Wearable Power-Assist LocomotorTable 2 Control options to initiate stepping for poweredexoskeletons included in this reviewExoskeleton ExternaloperatorUser-operatedvia buttonsUser-operatedvia weightshiftsUser-operatedvia bioelectricsignal detectionReWalk • •Ekso • • •Indego • • •HAL • •Mina •WPAL •CustomIRGO•HAL Hybrid Assistive Limb, WPAL Wearable Power-Assist Locomotor,IRGO Isocentric Reciprocal Gait OrthosisLouie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 4 of 10paraplegia with extensive zones of partial preservation.Regardless of completeness, they required participants tohave some volitional motor function of the hip and kneeextensor and flexor groups in order to use the HybridAssistive Limb (HAL) exoskeleton (Cyberdyne, Japan),which detects the user’s bioelectrical signals to generatestepping.Of all the studies included in this review, 11 of thestudies had stated exclusion criteria. The exclusion cri-teria generally consisted of severe comorbidities thatwould make it unsafe for the participant to use the pow-ered exoskeleton: concurrent neurological or other pro-gressive disease [14–16, 21, 25–27]; unstable spine,fracture risk or osteoporosis [13, 15–17, 21–23, 25–27];cardiorespiratory limitations to exercise such as auto-nomic dysreflexia [13, 16, 17, 21–23, 25, 26]; pressuresores at point of contact [13, 16, 17, 21–23, 25–27]; se-vere limitations in range of motion due to contracture,heterotypic ossification, or spasticity [13–17, 21–23, 25, 26];or cognitive deficits [13, 15, 16, 21, 25, 27]. Other exclu-sion criteria were pregnancy [17, 21, 23], asymmetric hippositions [14, 23], surgery in the last three months [23],participation in lower extremity conditioning in last threemonths [23], previous use of any robotic exoskeletal de-vice [15], Type I or II Diabetes [23], and pain limiting fore-arm crutch use [23]. Only one study [13] listed non-traumatic SCI as an exclusion criteria for their study.Powered exoskeleton as an assistive device for ambulationParticipants and level of impairmentThere were 92 participants (74 males) across the 14studies that utilized a powered exoskeleton as an assist-ive device for ambulation. Of these participants, the ma-jority were motor complete (AIS A or B) thoracic-levelSCI (Table 1); six participants had incomplete SCI. Thehighest level of injury included was C4 and the lowestwas L1 with a mean injury level of T7. Participantsranged from two months post-injury to 24 years, with amean of 5.8 years (SD: 5.6 years) after injury. The meanage of participants across all studies was 37.5 years (SD:12.3 years).Gait speedOf the 14 studies utilizing the powered exoskeleton as anassistive device, eight studies [15, 16, 18–20, 23, 26, 27]assessed gait speed by means of the 10MWT, while twostudies [14, 24] simply reported gait speed. Two studies[21, 25] reported walking parameters (time and distance)recorded during a session that could be used to calculate agait speed; these session durations were generally quitelong, ranging from 4.5 to 54 min. Two studies [17, 22] cal-culated gait speed from measures of endurance: the 2-Minute Walk Test (2MWT) and 6-Minute Walk Test(6MWT). Twelve studies reported individual participantgait speed, which ranged ranged from 0.031 m/s to0.71 m/s. The mean gait speed attained by the 84 partici-pants in these 12 studies was 0.26 m/s (SD: 0.15 m/s)(Table 3).Gait aid at assessmentAt this time, powered exoskeletons require the use of agait aid for support during stepping. The general expect-ation is for exoskeleton users to eventually progress toforearm crutches, which provide less stability than walk-ing frames but are less bulky and thus more portable.One study [20] which included individuals with cervical-level SCI, allowed participants to use a platform walkerif needed. Seven studies [14, 17, 18, 20, 21, 23, 26]allowed participants to use a 2-wheeled walker for as-sessment; 10 studies [15–17, 19–22, 24, 26, 27] had par-ticipants who achieved exoskeletal walking with forearmcrutches by the end of the training period.Control of exoskeleton and independenceThe control of walking while wearing a powered exo-skeleton varies (Table 2). In two studies [23, 25], partici-pants ambulated by controlling stepping with buttons ontheir walker. In 10 studies [15–20, 22, 23, 26, 27], partic-ipants generated stepping by shifting their own weightwithin the exoskeleton; the exoskeleton is able to detectchanges in centre of mass over one limb and in responsegenerates a step contralaterally. In another three studies[14, 21, 24], exoskeletal stepping was initiated by an exter-nal operator using a control interface. While all studies re-ported on participants that did not require assistance, fourstudies [19, 20, 22, 26] also reported on several partici-pants requiring minimal to moderate hands-on assistancewith the exoskeleton during the gait assessment. TheTable 3 Mean gait speed of non-ambulatory participants whileusing exoskeleton at end of training periodGait speed (m/s)Mean (SD)Participants with individual data (n = 84) 0.26 (0.15)Incomplete SCI participants (n = 6) 0.32 (0.25)Complete SCI participants (n = 78) 0.25 (0.14)By deviceReWalk (n = 37) 0.28 (0.15)Ekso (n = 18) 0.14 (0.07)Indego (n = 20) 0.31 (0.11)WPAL (n = 7) 0.16 (0.06)By assistanceaNo hands-on assistance (n = 63) 0.26 (0.15)Hands-on assistance (n = 15) 0.21 (0.07)SD Standard Deviation, WPAL Wearable Power-Assist LocomotoraHands-on physical assistance provided during evaluation of gait speedLouie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 5 of 10mean gait speeds attained by participants using the exo-skeleton as an assistive device, grouped by exoskeleton,level of assistance, and completeness of injury, are shownin Table 3.Training protocolAs seen in Table 1, training period varied significantlyacross the studies included in this review. Some studies in-volved a shorter training period [17, 20, 21, 24, 25, 27],often ending when the exoskeleton user achieved inde-pendence or the ability to walk a set distance; other stud-ies [14–16, 19, 22, 23] utilized a set training protocollasting several weeks to months, not based on participantprogress. One study [26] did not have a set training proto-col or end-point, with participants undergoing between 12and 102 training sessions to achieve their best perform-ance with the exoskeleton. An aggregate mean of 19.8(SD = 18.6, n = 79) training sessions was calculated acrossall studies; training sessions were 60 to 120 min in dur-ation. In all studies, participants were generally progressedfrom standing in the exoskeleton to weight shifting andstepping exercises to walking either within parallel bars orusing a gait aid. In three studies [15, 19, 22], participantswere progressed to training on different surfaces includingsidewalk, grass, or stairs. Tanabe et al. [25] incorporated atreadmill as part of the training protocol to improve userconfidence and speed. Only one study included upper ex-tremity strengthening and lower extremity stretching aspart of the intervention protocol [14].Powered exoskeleton as a training tool to improveambulationAs an intervention for ambulatory individuals with SCI,eight participants trained with the HAL in the Aach etal. [13] study for five days a week over a 90-day period(mean of 51.75 sessions). Participants ambulated on abody weight-supported treadmill while wearing the HAL;speed and body weight-support were adjusted individually.At the end of the intervention period, the participants im-proved their mean gait speed without the exoskeletonfrom 0.28 m/s to 0.50 m/s (p < 0.05, n = 8, effect size =0.71). They also demonstrated an improvement in mean6MWT distance from 70.1 m to 163.3 m (p < 0.05, n = 8,effect size = 0.64). On the other hand, the two participantswith incomplete SCI in the Benson et al. [15] study didnot show clear improvements in mean gait speed (0.26 m/sto 0.27 m/s) or 6MWT distance. In contrast to the Aachet al. [13] study, these two participants underwent only 20training sessions over 10 weeks, which did not includecontinuous treadmill training.Adverse eventsAcross all 15 articles, five [14, 17–19, 25] did not reporton whether any adverse events occurred with use of apowered exoskeleton. Of the 10 studies that reported onadverse events, five [13, 22–24, 27] reported no skinchanges, while five [15, 16, 20, 21, 26] reported mild skineffects (redness or superficial abrasions). Four articles[16, 21, 23, 24] addressed and reported no change inspasticity, and five [16, 21, 23, 24, 27] which addressedpain reported no change or a slight decrease in usualpain. Safety precautions, such as overhead tether orclose guarding, were taken in all studies to ensure par-ticipant safety, though loss of balance was used for somearticles as an outcome measure. In one study [21], fallsengaging the overhead tether were reported for threeparticipants over six days of training; two of these partic-ipants experienced a combined three falls due to mech-anical programming errors of the exoskeleton, while thethird participant experienced over 10 falls due to mal-functioning of specialized forearm crutches which werelater discontinued. Two studies [16, 24] reported somelower extremity edema due to prolonged standing. Onestudy [15] removed a participant for safety reasons dueto a “near-serious” device-related adverse event involvinga hairline fracture of the talus that did not requiretreatment.Factors influencing exoskeletal gait speed in non-ambulatoryindividuals with SCIFour variables were found in the majority of studies whichmight influence gait speed in non-ambulatory individualsusing the exoskeleton device to walk: age, injury duration,injury level, and number of training sessions. As the num-ber of incomplete participants across all the studies wassmall (n = 6), they were not included in the correlationalanalyses. We also removed the participants who requiredhands-on assistance to ambulate with the exoskeletonfrom the correlation calculations.All 12 studies reporting individual data provided infor-mation on participant age; in some cases, a narrow agerange (e.g. 20–24) was provided, and the midpoint of therange was used for that individual. A significant correl-ation was found between increasing age and faster gaitspeed (r = 0.27, 95 % CI 0.02–0.48, p = 0.03, n = 63)(Fig. 2). However, no relationship was found betweeninjury duration and gait speed (r = 0.19, 95 % CI −0.09–0.44, p = 0.18, n = 53) from 10 studies. From the 12 stud-ies, we found a significant correlation between injurylevel and gait speed (r = 0.27, 95 % CI 0.02–0.48, p =0.03, n = 63). Higher speeds were associated with a lowerlevel of injury when walking with an exoskeleton as anassistive device (Fig. 3).Eleven studies reported the number of training ses-sions for individual participants. Those who were able topractice longer with the powered exoskeleton achievedfaster gait speeds (r = 0.27, 95 % CI 0.003–0.49, p = 0.048,n = 56). One individual in the Yang et al. [26] studyLouie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 6 of 10underwent 102 training sessions to achieve their bestwalking outcome, compared to the group mean of 19.8sessions. When we removed this outlier, the correlationcoefficient increased to 0.41 (95 % CI 0.16–0.61; p = 0.002,n = 55) (Fig. 4).DiscussionThe advent of the powered exoskeleton in rehabilitationhas many implications for individuals with SCI with lim-ited or no walking ability. It allows wheelchair-users tostand and ambulate, which may influence communitymobility and social participation. Powered exoskeletonsalso require less energy to use than standard rigid orth-oses [10] and are becoming lighter and more accessible.Use of powered exoskeletons without overhead body-weight support for over-ground ambulation is a new re-habilitation strategy, and to our knowledge our review isthe first to examine their ability for promoting gait speedfor individuals with SCI.The relationship between level of injury and gait speedsuggests that proficiency of powered exoskeletal walkingis linked to the functional presentation of the user. Indi-viduals with more neurological preservation of theirspinal cord are more likely to achieve greater speedsFig. 2 Gait speed plotted against age using individual participant data, excluding those with incomplete injuries or requiring assistance toambulate (n = 63 from 12 studies)Fig. 3 Gait speed plotted against injury level using individual participant data, excluding those with incomplete injuries or requiring assistance toambulate (n = 63 from 12 studies)Louie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 7 of 10with a powered exoskeleton. Though the upper extrem-ities are considered spared in all thoracic-level SCI, indi-viduals with high thoracic injuries include pectoralismajor and latissimus dorsi in their postural controlmuscle synergies [29]; individuals with greater preserva-tion have more trunk musculature activation and cancontrol their centre of mass with less dependence onthe arms. Currently, all powered exoskeletons requirethe use of an additional gait aid, and some generatestepping in response to lateral shifts of centre of mass.An individual with less reliance on the upper extremitiesfor maintaining postural stability will be more able to liftor push their gait aid and to navigate their centre ofmass.There was an unexpected relationship found betweenage and gait speed, with older participants achievinggreater speeds than younger participants. One possibleexplanation for this relationship may lie in the epidemi-ology of SCI. Younger individuals with SCI tend to sus-tain a traumatic SCI, while older individuals with SCItend to have a non-traumatic SCI [30, 31]. Further tothis, traumatic SCI tends to result in a higher level of in-jury and more neurological impairment than non-traumatic SCI [32, 33]. Many of the studies included inthis review did not indicate whether participants had atraumatic or non-traumatic injury, so we could not con-firm this hypothesis. However, a post-hoc analysis founda non-significant trend between increasing age and lowerlevels of injury (i.e. less neurological impairment) (r =0.20, 95 % CI −0.05–0.43, p = 0.11, n = 63). Without con-trolling for injury level, we would then expect the olderindividuals in our included studies to walk faster thanyounger individuals.Participants were able to ambulate independently withina reasonable training time, with some subjects doing sowithin the first training session. However, those whowere able to train for several weeks to months weregenerally able to achieve ambulation at faster speedswith a powered exoskeleton. Repetitive task practice isa requirement for improved speed and accuracy of anew skill [34], and is a possible explanation for this re-lationship. As exoskeletons are beginning to be ap-proved for personal and home use, daily use may helpexoskeleton-users attain higher gait speeds quickly.Our findings showed that use of a powered exoskeletonallowed non-ambulatory individuals with SCI to ambulateat a mean speed of 0.26 m/s, despite the maximum speedof commercial powered exoskeletons such as the ReWalkbeing 0.55 m/s (ReWalk™ Personal System User Guide,ReWalk Robotics, Israel). A gait speed of 0.26 m/s is notconsidered sufficient for community ambulation; Forrestet al. [35] found a threshold of 0.44 m/s for limited com-munity ambulation after incomplete SCI while Andrewset al. [36] determined the mean speed necessary to crossan intersection as set by traffic signals to be 0.49 m/s.However, 0.26 m/s is within a range comparable to indi-viduals with incomplete SCI who are able to walk with orwithout supervision indoors [37]. In our included studies,one individual with a motor-incomplete C8 SCI using aReWalk was able to ambulate at 0.71 m/s, higher than thedevice’s reported maximum speed of 0.55 m/s.As a training intervention for ambulatory individualswith SCI, participants in the Aach et al. [13] study dem-onstrated significant improvements in gait speed and en-durance with use of a powered exoskeleton. These largeimprovements may be due in part to the principles ofFig. 4 Gait speed plotted against number of training sessions using individual participant data, excluding those with incomplete injuries orrequiring assistance to ambulate (n = 55 from 11 studies, one outlier removed)Louie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 8 of 10motor learning and neuroplasticity. The high level ofrepetition on the treadmill and feedback of successfulactive stepping with combined use of the bioelectricsignal-dependent HAL exoskeleton may have helped tostrengthen the intact neural pathways in incomplete SCI[38]. On the other hand, the two participants in theBenson et al. [15] study with incomplete SCI did notshow any improvement in gait speed or endurance. Simi-larly, a systematic review of treadmill-based robotics-assisted locomotor training found reduced walkingendurance and no difference in gait speed after robotics-assisted locomotor training using the Lokomat (Hocoma,Switzerland) compared to other forms of gait training[39]. Due to these mixed findings, further research in thispopulation is required to investigate the potential of pow-ered exoskeletons as a training tool.The training protocol was similar across all the stud-ies, progressing from becoming familiar with standingand balancing in the exoskeleton to stepping and walk-ing within the exoskeleton. This progression of confi-dence is similar to training with other lower limborthoses, with repetition being a key principle for train-ing. All studies employed safety precautions (spottingand overhead tether) to ensure safety and confidencewhile learning to use a new assistive device.This systematic review has some limitations. The levelof evidence in the current literature is limited to studieswith a small number of participants. In addition, a truecontrol group (without a device to walk) is not relevantas most participants would not have been able to walkwithout the exoskeleton; however, future studies couldcompare different orthotic, FES, or exoskeleton systems.There was heterogeneity in the study characteristics (de-vice, control of stepping, training duration, outcomemeasurement), which made it challenging to compareresults and reduces the ability to generalize results.However, we attempted to overcome this by aggregatingparticipant data to allow statistical analysis to explorecorrelations between participant characteristics and out-comes. In the future, it would be useful for studies to re-port on the exact intensity of training, using suchmeasures as number of steps or walking time.ConclusionIn conclusion, powered exoskeletons can provide individualswith thoracic-level motor-complete SCI the ability to walkat modest speeds. Exoskeletal gait speed is related to theamount of time spent practicing as well as level of injury.Competing interestsThe authors declare that they have no competing interests.Authors’ contributionsJJE conceived the review paper idea. DRL conducted the literature search, aswell as drafting the manuscript. JJE and TL guided the literature and draftingof the manuscript. All authors read and approved the final manuscript priorto submission.AcknowledgementsWe acknowledge support from the Rick Hansen Institute, OntarioNeurotrauma Foundation and Canadian Institutes of Health Research.Author details1University of British Columbia, Vancouver, Canada. 2Rehabilitation ResearchProgram, 4255 Laurel Street, Vancouver, BC, Canada, V5Z 2G9. 3VancouverCoastal Health Research Institute, Vancouver, Canada. 4InternationalCollaboration on Repair Discoveries, 818 West 10th Avenue, Vancouver, BC,Canada, V5Z 1M9. 5Department of Physical Therapy, University of BritishColumbia, 212-2177 Wesbrook Mall, Vancouver, BC, Canada, V6T 1Z3.Received: 4 June 2015 Accepted: 4 September 2015References1. Hitzig SL, Tonack M, Campbell KA, McGillivray CF, Boschen KA, Richards K, etal. Secondary health complications in an aging Canadian spinal cord injurysample. Am J Phys Med Rehabil. 2008;87:545–55.2. Post M, Noreau L. Quality of life after spinal cord injury. J Neurol Phys Ther.2005;29:139–46.3. Ditunno PL, Patrick M, Stineman M, Ditunno JF. Who wants to walk?Preferences for recovery after SCI: a longitudinal and cross-sectional study.Spinal cord. 2008;46:500–6.4. Saunders LL, Krause JS, DiPiro ND, Kraft S, Brotherton S. Ambulation andcomplications related to assistive devices after spinal cord injury. J SpinalCord Med. 2013;36:652–9.5. Massucci M, Brunetti G, Piperno R, Betti L, Franceschini M. Walking with theAdvanced Reciprocating Gait Orthosis (ARGO) in thoracic paraplegicpatients: energy expenditure and cardiorespiratory performance. SpinalCord. 1998;36:223–7.6. Kawashima N, Taguchi D, Nakazawa K, Akai M. Effect of lesion level on theorthotic gait performance in individuals with complete paraplegia. SpinalCord. 2006;44:487–94.7. Scivoletto G, Petrelli A, Lucente LD, Giannantoni A, Fuoco U, D’Ambrosio F,et al. One year follow up of spinal cord injury patients using a reciprocatinggait orthosis: Preliminary report. Spinal Cord. 2000;38:555–8.8. Sale P, Franceschini M, Waldner A, Hesse S. Use of the robot assisted gaittherapy in rehabilitation of patients with stroke and spinal cord injury. Eur JPhys Rehabil Med. 2012;48:111–21.9. Chen G, Chan CK, Guo Z, Yu H. A review of lower extremity assistive roboticexoskeletons in rehabilitation therapy. Crit Rev Biomed Eng. 2013;41:343–63.10. Arazpour M, Samadian M, Bahramizadeh M, Joghtaei M, Maleki M, Bani MA,et al. The efficiency of orthotic interventions on energy consumption inparaplegic patients: a literature review. Spinal Cord. 2015;53:168–75.11. Arazpour M, Hutchins SW, Bani MA. The efficacy of powered orthoses onwalking in persons with paraplegia. Prosthet Orthot Int. 2015;39:90–9.12. Moher D, Liberati A, Tetzlaff J, Altman DJ. The Prisma Group. Preferredreporting items for systematic reviews and meta-analysis: the PRISMAstatement. BMJ. 2009;339:b2535.13. Aach M, Cruciger O, Sczesny-Kaiser M, Höffken O, Meindl RC, Tegenthoff M,et al. Voluntary driven exoskeleton as a new tool for rehabilitation inchronic spinal cord injury: a pilot study. Spine J. 2014;14:2847–53.14. Arazpour M, Bani MA, Hutchins SW, Jones RK. The physiological cost indexof walking with mechanical and powered gait orthosis in patients withspinal cord injury. Spinal Cord. 2013;51:356–9.15. Benson I, Hart K, Tussler D, van Middendorp JJ. Lower-limbexoskeletons for individuals with chronic spinal cord injury: Findingsfrom a feasibility study. Clin Rehabil. 2015. doi:10.1177/0269215515575166.16. Esquenazi A, Talaty M, Packel A, Saulino M. The ReWalk poweredexoskeleton to restore ambulatory function to individuals with thoracic-levelmotor-complete spinal cord injury. Am J Phys Med Rehabil. 2012;91:911–21.17. Evans N, Hartigan C, Kandilakis C, Pharo E, Clesson I. Acute cardiorespiratoryand metabolic responses during exoskeleton-assisted walking overgroundamong persons with chronic spinal cord injury. Top Spinal Cord Inj Rehabil.2015;21:122–32.Louie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 9 of 1018. Farris RJ, Quintero HA, Murray SA, Ha KH, Hartigan C, Goldfarb M. Apreliminary assessment of legged mobility provided by a lower limbexoskeleton for persons with paraplegia. IEEE Trans Neural Syst Rehabil Eng.2014;22:482–90.19. Fineberg DB, Asselin P, Harel NY, Agranova-Breyter I, Kornfeld SD, BaumanWA, et al. Vertical ground reaction force-based analysis of poweredexoskeleton-assisted walking in persons with motor-complete paraplegia.J Spinal Cord Med. 2013;36:313–21.20. Hartigan C, Kandilakis C, Dalley S, Clausen M, Wilson E, Morrison S, et al.Mobility outcomes following five training sessions with a poweredexoskeleton. Top Spinal Cord Inj Rehabil. 2015;21:93–9.21. Kolakowsky-Hayner SA, Crew J, Moran S, Shah A. Safety and feasibility ofusing the Ekso™ bionic exoskeleton to aid ambulation after spinal cordinjury. J Spine. 2013. doi:10.4172/2165-7939.S4-003.22. Kozlowski AJ, Bryce TN, Dijkers MP. Time and effort required by personswith spinal cord injury to learn to use a powered exoskeleton for assistedwalking. Top Spinal Cord Inj Rehabil. 2015;21:110–21.23. Kressler J, Thomas CK, Field-Fote EC, Sanchez J, Widerström-Noga E, CilienDC, et al. Understanding therapeutic benefits of overground bionicambulation: Exploratory case series in persons with chronic, complete spinalcord injury. Arch Phys Med Rehabil. 2014;95:1878–87.24. Neuhaus PD, Noorden JH, Craig TJ, Torres T, Kirschbaum J, Pratt JE. Designand evaluation of Mina: A robotic orthosis for paraplegics. IEEE Int ConfRehabil Robot. 2011. doi:10.1109/ICORR.2011.5975468.25. Tanabe S, Hirano S, Saitoh E. Wearable Power-Assist Locomotor (WPAL) forsupporting upright walking in persons with paraplegia. NeuroRehabil.2013;33:99–106.26. Yang A, Asselin P, Knezevic S, Kornfeld S, Spungen AM. Assessment ofin-hospital walking velocity and level of assistance in a poweredexoskeleton in persons with spinal cord injury. Top Spinal Cord InjRehabil. 2015;21:100–9.27. Zeilig G, Weingarden H, Zwecker M, Dudkiewicz I, Bloch A, Esquenazi A.Safety and tolerance of the ReWalk™ exoskeleton suit for ambulation bypeople with complete spinal cord injury: A pilot study. J Spinal Cord Med.2012;35:96–101.28. Arazpour M, Chitsazan A, Hutchins SW, Mousavi ME, Takamjani EE, GhomsheFT, et al. Evaluation of a novel powered gait orthosis for walking by a spinalcord injury patient. Prosthet Orthot Int. 2012;36:239–46.29. Seelen HAM, Potten YJM, Drukker J, Reulen JPH, Pons C. Development ofnew muscle synergies in postural control in spinal cord injured subjects.J Electromiogr Kinesiol. 1998;8:23–34.30. McKinley WO, Seel RT, Hardman JT. Nontraumatic spinal cord injury:Incidence, epidemiology, and functional outcome. Arch Phys Med Rehabil.1999;80:619–23.31. Güzelküçük Ü, Demir Y, Kesikburun S, Yaşar E, Yılmaz B. Spinal cord injury inolder population in Turkey. Spinal Cord. 2014;52:850–4.32. McCammon JR, Ethans K. Spinal cord injury in Manitoba: a provincialepidemiological study. J Spinal Cord Med. 2011;34:6–10.33. New PW, Simmonds F, Stevermuer T. A population-based study comparingtraumatic spinal cord injury and non-traumatic spinal cord injury using anational rehabilitation database. Spinal Cord. 2011;49:397–403.34. Shmuelof L, Krakauer JW, Mazzoni P. How is a motor skill learned? Changeand invariance at the levels of task success and trajectory control.J Neurophysiol. 2012;108:578–94.35. Forrest GF, Hutchinson K, Lorenz DJ, Buehner JJ, VanHiel LR, Sisto SA, et al.Are the 10 meter and 6 minute walk tests redundant in patients with spinalcord injury? PLoS One. 2014;9: doi:10.1371/journal.pone.0094108.36. Andrews AW, Chinworth SA, Bourassa M, Garvin M, Benton D, Tanner S.Update on distance and velocity requirements for community ambulation.J Geriatr Phys Ther. 2010;33:128–34.37. van Hedel HJ. Gait speed in relation to categories of functional ambulationafter spinal cord injury. Neurorehabil Neural Repair. 2009;23:343–50.38. Dobkin BH, Apple D, Barbeau H, Basso M, Behrman A, Deforge D, et al.Methods for a randomized trial of weight-supported treadmill trainingversus conventional training for walking during inpatient rehabilitation afterincomplete traumatic spinal cord injury. Neurorehabil Neural Repair.2003;17:153–67.39. Mehrholz J, Kugler J, Pohl M. Locomotor training for walking after spinalcord injury. Cochrane Database Syst Rev. 2008. doi:10.1002/14651858.CD006676.pub2.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitLouie et al. Journal of NeuroEngineering and Rehabilitation  (2015) 12:82 Page 10 of 10


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items