Wright et al. Journal of NeuroEngineering Journal of NeuroEngineering and Rehabilitation (2023) 20:45 https://doi.org/10.1186/s12984-023-01165-0 and Rehabilitation RESEARCH Open Access Multicentric investigation on the safety, feasibility and usability of the ABLE lower‑limb robotic exoskeleton for individuals with spinal cord injury: a framework towards the standardisation of clinical evaluations Mark Andrew Wright1,2,3†, Franziska Herzog4†, Anna Mas‑Vinyals5, Alfons Carnicero‑Carmona5, Joan Lobo‑Prat5, Cornelia Hensel4, Steffen Franz4, Norbert Weidner4, Joan Vidal1,2,3, Eloy Opisso1,2,3 and Rüdiger Rupp4* Abstract Background Robotic lower-limb exoskeletons have the potential to provide additional clinical benefits for persons with spinal cord injury (SCI). However, high variability between protocols does not allow the comparison of study results on safety and feasibility between different exoskeletons. We therefore incorporated key aspects from previous studies into our study protocol and accordingly conducted a multicentre study investigating the safety, feasibility and usability of the ABLE Exoskeleton in clinical settings. Methods In this prospective pretest-posttest quasi-experimental study across two SCI centres in Germany and Spain, in- and outpatients with SCI were recruited into a 12-session training and assessment protocol, utilising the ABLE Exoskeleton. A follow-up visit after 4 weeks was included to assess after-training outcomes. Safety outcomes (device- related adverse events (AEs), number of drop-outs), feasibility and usability measures (level of assistance, donning/ doffing-time) were recorded at every session together with changes in gait parameters and function. Patient-reported outcome measures including the rate of perceived exertion (RPE) and the psychosocial impact of the device were performed. Satisfaction with the device was evaluated in both participants and therapists. Results All 24 participants (45 ± 12 years), with mainly subacute SCI (< 1 year after injury) from C5 to L3, (ASIA Impair‑ ment Scale A to D) completed the follow-up. In 242 training sessions, 8 device-related AEs (pain and skin lesions) were reported. Total time for don and doff was 6:50 ± 2:50 min. Improvements in level of assistance and gait parameters (time, steps, distance and speed, p < 0.05) were observed in all participants. Walking function and RPE improved in participants able to complete walking tests with (n = 9) and without (n = 6) the device at study start (p < 0.05). A † Mark Andrew Wright and Franziska Herzog shared first authorship *Correspondence: Rüdiger Rupp ruediger.rupp@med.uni-heidelberg.de Full list of author information is available at the end of the article © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 2 of 18 positive psychosocial impact of the exoskeleton was reported and the satisfaction with the device was good, with best ratings in safety (participants), weight (therapists), durability and dimensions (both). Conclusions Our study results prove the feasibility of safe gait training with the ABLE Exoskeleton in hospital settings for persons with SCI, with improved clinical outcomes after training. Our study protocol allowed for consistent com‑ parison of the results with other exoskeleton trials and can serve as a future framework towards the standardisation of early clinical evaluations. Trial Registration https://​trial​search.​who.​int/, DRKS00023503, retrospectively registered on November 18, 2020. Keywords Spinal cord injury, Spinal cord disorder, Exoskeleton, Rehabilitation, Robotics, Safety, Feasibility, Usability, Gait, Study protocol, Standardisation Background the development of a transferable study protocol that Spinal cord injury (SCI) is a life-changing condition that takes the key aspects identified in previous literature into affects multiple body systems resulting in sensory and account and therefore allows for the comparison of study motor impairments very often associated with long-term results. immobility [1–3]. Achieving independent ambulation has a high relevance for participation in social and pro- Methods fessional life [1, 4]. In the past 10 years, one of the most Study objectives notable technological developments to support this has The primary objective of this study was to determine the been the creation of lower-limb robotic exoskeletons. safety, feasibility and usability of the ABLE Exoskeleton These devices aim to allow individuals with SCI to per- for individuals with traumatic and non-traumatic SCI in a form motor learning-based training with multiple repeti- hospital setting during a 4–6-week training programme. tions of the locomotor task and minimal physical burden The secondary objectives were to assess the effect of the to therapists, with a view to allowing independent con- ABLE Exoskeleton on gait and functional outcomes, the tinuation of the training in a community setting in the rate of perceived exertion, psychosocial impact and user future [2, 5]. A number of exoskeletons are now certified satisfaction. for use in a clinical setting which previously underwent an initial study as part of the certification process to vali- Study design date their safety and feasibility in the intended popula- We conducted a prospective pretest–posttest quasi- tion [2, 6, 7]. experimental study in two European SCI centres (Institut Despite the widely recognised need for safety and fea- Guttmann, Badalona, Spain and Heidelberg University sibility testing, there is significant variability in the study Hospital, Heidelberg, Germany) from November 2020 protocols used within these studies [8]. While most stud- until October 2021. The individual pre–post training ies include participants 6 months or more after SCI, the period covered 4–6 weeks. The clinical trial was approved participants’ level and severity of injury are often highly by the responsible local ethics committees and com- variable, with vastly different inclusion and exclusion cri- petent authorities of Spain and Germany, respectively teria reported even with the same device [9]. Concern (EUDAMED No.: CIV-20-07-034264). The study was has previously been expressed about the marked vari- conducted in accordance with the requirements of ISO ability in the different methods and levels of detail used 14155:2011 and European Regulation MDR 2017/745 on to report safety aspects, with events such as fractures not medical devices. consistently considered as serious adverse events in pre- vious trials [9, 10]. Accordingly, future study protocols ABLE Exoskeleton should facilitate more meaningful comparisons between The ABLE Exoskeleton (Fig. 1) is a wearable lower-limb studies and should be better structured to identify risk robotic exoskeleton weighing 9.8 kg and consisting of a concerns such as falls, fractures, and long-term adverse rigid brace that attaches to the torso, legs and feet of the effects of using exoskeletons [8, 9]. It has also been high- user via straps and supports. The size of the exoskeleton lighted that future studies should broaden their scope to segments can be adjusted to fit users with a height of include reporting on changes in areas related to quality of 150–190 cm and a maximum weight of 100 kg. This bilat- life (QoL) [8]. eral exoskeleton has two battery-powered motors that The aim of our multicentre investigation was to evalu- assist flexion–extension movements of each knee joint, ate the ABLE Exoskeleton regarding its safe and feasible and two non-motorized, passive hip joints that allow free use in clinical settings. Special emphasis was placed on movement in the sagittal plane. With its knee motors, it Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 3 of 18 walking aid (cane, crutch or walker) for supporting a sta- ble body position of the users who typically have restric- tions to fully stabilize their trunk and always under the supervision of a trained therapist. The device comes with a smartphone with the pre-installed software application ABLE Care (Fig. 1) that allows the therapist to configure and monitor the exoskeleton during the therapy session. Study protocol & framework To address the need for better conformity of exoskeleton trials we based our study protocol on a systematic lit- erature search of studies investigating the safety and/or feasibility and/or usability of robotic lower-limb exoskel- etons. A total of 13 studies were analysed to identify the most common in- and exclusion criteria, training meth- odology and outcome measures (for an overview see Additional file 1). Fig. 1 Overview of the hard- and software of the ABLE Exoskeleton. A The ABLE Exoskeleton with the knee motors and passive hip joints. In‑ and exclusion criteria B The graphical user interface of the ABLE Care mobile app The inclusion and exclusion criteria of the study (see Table 1) were largely identified by using criteria corre- sponding to other exoskeleton safety and feasibility trials. Most studies use the American Spinal Injury Association actively assists a person to stand up, walk and sit down. Impairment Scale (AIS) of the International Standards Steps with the associated knee flexion–extension move- for Neurological Classification of SCI (ISNCSCI) [12] for ments during swing phase are triggered either manually inclusion, however, in order to not limit the recruitment by the therapist or automatically by the user. The man- process we decided to add the criterion “sufficient arm ual step trigger is activated by means of pressing either strength to support body weight on a walking frame” of the pushbuttons (left for triggering the left knee flex- as in previous studies [4, 13, 14]. Screening evaluations ion–extension movement and right for the right one) that for fracture risks in previous studies ranged from Dual are located on the lumbar segment of the exoskeleton. An Energy X-Ray Absorptiometry (DXA) bone density scans exoskeleton user is encouraged to maintain an upright [15, 16] to simply a medical history of osteoporosis [13, posture during gait and to utilise momentum inducing 14, 17]. Considering the variability in measures used, we weight shifts synchronously to the triggering of the step- decided to exclude any individual with a history of lower- ping action on the respective side. In the automatic step limb fragility fractures in the last 2 years [4], and/or, who initiation mode, the exoskeleton autonomously seeks to had 5 or more risk factors present for fragility fractures determine the time instant when the user wants to take as cited by Craven et al. [18]. It was considered that these a step, detected as a change in the thigh angular veloc- criteria were extensive enough to cover risks, whilst also ity measured with an inertial measurement unit on the being accessible to all SCI centres without immediate leg brace of the exoskeleton. When this signal change access to DXA scans. is detected, a step is triggered on the leg that is behind. Due to the passive hip joints of the exoskeleton, in peo- Training methodology ple with no or very limited voluntary hip movements, Sampling was completed by the pre-screening of all in- hip flexion is generated by forward-rotational movement and outpatients. All patients who met the inclusion cri- of the pelvis. An analysis of the gait pattern used whilst teria and did not present with criteria for exclusion were walking with the ABLE exoskeleton has been analysed asked to participate in the clinical trial. After screening in a previous study [11]. Note that the aim of having the for in- and exclusion criteria, a baseline visit was com- different walking modes is to adapt the exoskeleton to pleted, followed by the training programme and a post- the users’ needs. Once participants are familiar with the training visit. Four weeks after the final training session a device and have learned how to move to initiate steps, it follow-up visit was performed by phone interview due to is possible to change the step initiation mode from man- the COVID-19 restrictions. ual to automatic walking. The device is to be used with a Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 4 of 18 Table 1 Inclusion and exclusion criteria. Inclusion Criteria Exclusion Criteria ∙ 18–70 years of age ∙ WISCI II [19] without exoskeleton of > 16 ∙ Traumatic and non-traumatic SCI ∙ History of lower-limb fragility fractures in the last 2 years ∙ AIS grade A to–D with sufficient arm strength to support body weight on ∙ 5 or more risk factors present for fragility fractures as stated by Craven a walking frame et al. [18] ∙ Currently receiving treatment as an inpatient or outpatient at one of the ∙ Deterioration > 3 points of the total ISNCSCI motor score within the last investigational sites 4 weeks ∙ Ability to give informed consent ∙ Spinal instability ∙ Modified Ashworth scale [20]  > 3 in lower limbs ∙ Unable to tolerate 30 min standing without clinical symptoms of orthos‑ tatic hypotension ∙ Unable to perform a sit-to-stand transfer or stand in the device with assistance ∙ Psychological or cognitive issues that do not allow the participant to fol‑ low the study procedures ∙ Known pregnancy or breastfeeding ∙ Any neurological condition other than SCI ∙ Medically unstable (Unstable CVS, hemodynamic instability, untreated hypertension (SBP > 140, DBP > 90 mmHg), unresolved DVT, uncontrolled AD) ∙ Severe comorbidities (any condition that a physician considers to not be appropriate to complete participation in the study) ∙ Ongoing skin issues (Grade I or higher on the European Pressure Ulcer Advisory Panel (EPUAP) scale [23] on areas that will be in contact with the exoskeleton) ∙ Height, width, weight or other anatomical constraints (such as leg length differences) incompatible with the device ∙ Insufficient ROM for device preventing a participant to achieve a normal gait pattern or to complete a sit-to-stand/stand-to-sit transition (hip joint: insufficient ROM for step length, knee joint: insufficient ROM for standing, ankle: insufficient ROM to reach neutral position) Abbreviations are defined in the Abbreviations section Screening Baseline Session 1 Session 2-11 Session 12 Post-Training Follow-up ISNCSCI SCIM III Skin Check (NPUAP/EPUAP) ISNCSCI SCIM III WISCI II TUG, Pain Check (VAS) WISCI II PIADS 10MWT, Time & LoA don/doff SCIM III QUEST 6MWT 2.0 without Exo LoA activity tasks PIADS Therapy metrics TUG, BORG 10MWT, TUG, TUG, 6MWT 10MWT, 10MWT, without Exo 6MWT 6MWT with Exo with Exo BORG BORG BORG QUEST Session 6: QUEST 2.0 QUEST 2.0 2.0 <1 week 3x/week for 4-6 weeks 4 weeks Body function & structure Activities Participation Personal & environmental factors Fig. 2 Overview of all study visits and measurements based on the International Classification of Functioning, Disability and Health (ICF) domains. Abbreviations are defined in the Abbreviation section Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 5 of 18 The training programme with the ABLE Exoskeleton for AEs; skin checks for signs of skin lesions on parts of (Fig. 2) consisted of three sessions a week over 4 weeks the body in contact with the exoskeleton graded as I or to achieve a total of 12 sessions. Missed sessions could be higher with the European Pressure Ulcer Advisory Panel made up within 6 weeks after baseline. The training ses- (EPUAP) scale [23]; and, recording of the presence, loca- sions were scheduled for 60 min to include adjustments, tion and severity (VAS) of pain. Any falls during train- donning and doffing time and data collection time. Ther- ing sessions were also documented. Furthermore, any apy time (time spent standing, walking or sitting in the requirement to have a medical review after the last ses- exoskeleton) was intended to be at least 30 min. Each ses- sion and its relation to the device was registered. sion was carried out by at least one trained therapist plus For feasibility and usability, the ease of donning and an additional therapist or assistant if required. During doffing the device was measured by the time and the level the first training sessions, participants were educated on of assistance (LoA) the participants needed to conduct the operating mechanisms of the exoskeleton and guided these activities, with the aim to conduct donning and through the basic use of the device: sit-to-stand, standing, doffing as independently and as time-efficiently as pos- weight shifting, stand-to-sit and walking, using crutches sible. Several studies have previously charted LoA, how- or frame as deemed appropriate for each individual. The ever defining this can be a very subjective process, with step initiation mode was individually chosen accord- variabilities between grading [6, 13, 15, 16, 21, 24]. As a ing to the participant’s ability and may have changed as result, we combined the best practices identified in previ- the study progressed. If the therapist considered it to be ous studies to produce a LoA scale (see Table 2). The LoA safe, the participant was required to complete the follow- was rated by the therapists in the following stages: total, ing activity tasks during every session with as little assis- maximum, moderate and minimal assistance, supervi- tance as possible from the therapist: sit-to-stand, walk sion and independence. These stages were also used to 10 metre, turn 180° and stand-to-sit. These predefined assess the LoA to perform the predefined activity tasks. activity tasks were implemented to achieve a better com- The study participants’ ability to use the device was also parability of the feasibility in different individual users of assessed by the therapy metrics recorded by the device: the exoskeleton. number of steps, standing and walking time, speed and distance. Outcome measures Secondary endpoints aimed to identify the effect of the The primary endpoints for safety were the number of exoskeleton training on participants’ gait and function, serious adverse events (SAEs), adverse events (AEs) and rate of perceived exertion (RPE), user satisfaction and study drop-outs related to the use of the device. SAEs psychosocial impact. Gait was assessed by conducting and AEs were classified systematically according to ISO the following standardised walking assessments: Walking 14155:2011, MEDDEV 2.7/3 and MDCG 2020-10/1. AEs Index for Spinal Cord Injury (WISCI) II, Timed Up and were defined in four categories: (1) device-related (AEs Go (TUG), 10 Metre Walk Test (10MWT) and 6 Minute that have occurred as a direct result of the device itself ), Walk Test (6MWT) as performed in numerous previous (2) procedure-related (AEs that occurred as a result of exoskeleton trials [1, 2, 4, 14–17, 21, 22, 24, 25]. If deemed the activities performed in training, but were not caused safe by the therapist, walking assessments were completed by the device), (3) disease-related (AEs that occurred without the device at baseline and post-training, and with as a result of the underlying health conditions), and (4) the device during sessions 1 and 12. Standardised rest other causes or undetermined relation. In addition to breaks were included in these sessions to ensure results these definitions, we considered device deficiencies as were reliable and valid. If any training sessions between AEs without a medical occurrence. Within these catego- 2 and 11 were missed, the participant was still given the ries the degree of relation was rated by the investigators opportunity to complete walking tests at session 12 with as ‘related’, ‘probably related’, ‘possibly related’, ‘unlikely the exoskeleton before the end of the 6-week training pro- related’ and ‘not related’. Only those AEs rated as ‘related’ gramme. For the 6MWT, a course of 50 m was marked were defined as device/procedure/disease-related AEs. out, with participants turning 180° once arriving at 50 m. Specific AE monitoring that was also noted in previous As in previous studies, the RPE was measured with the studies involving exoskeleton devices included checks for BORG-scale after the 6MWT [14]. Function in respect skin lesions, assessment of pain levels using a visual ana- to independence was evaluated by the Spinal Cord Inde- logue scale (VAS) pre- and post-each session [1, 2, 4, 6, pendence Measure (SCIM) III as in previous studies [22]. 13, 15, 16, 21, 22], frequency of falls and any event that Most of the assessments used in prior exoskeleton trials required medical intervention [6, 21, 22]. The following covered the function and activity domain of the Inter- assessments were performed before and after each train- national Classification of Functioning, Disability and ing session in this study: screening of medical records Health (ICF) (see Fig. 2). To provide a meaningful insight Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 6 of 18 Table 2 Level of Assistance definitions Level of Assistance Don/Doff Therapy activity task Total assistance Participant performs 0–25% of the effort to don/doff the Participant performs 0–25% of the effort to use the exoskeleton exoskeleton Two therapists are required to support the participant in the Participant is essentially reliant on the trainer to perform all device at all times aspects of the donning/doffing Maximum assistance Participant performs 25–50% of the effort to don/doff the Participant performs 25–50% of the effort to use the device device Participant needs maximum assistance from the therapist to Participant needs maximum assistance to transfer to device remain balanced and position legs but may be able to adjust the thigh straps Moderate assistance Participant performs 50–75% of the effort to don/doff the Participant performs 50–75% of the effort to use the device device The therapist has both hands on the participant or device at all Participant needs moderate assistance to transfer to device times to provide occasional guidance or balance support and position legs but may be able to adjust the thigh and shin straps Minimal assistance Participant performs > 75% of the effort to don/doff the device Participant performs > 75% or more of the effort to use the Participant can transfer to device and adjust straps but may device need help to position legs The therapist has one hand on the participant or device for infrequent guidance or balance support Supervision The therapist is not touching the participant but may provide The therapist is not touching the participant but is close verbal prompts or contact guarding to ensure safety enough to provide support for balance or guidance as needed Independent Participant is fully independent donning/doffing device The participant is fully independent while using the device and the therapist does not provide any assistance Level of Assistance was defined separately for Donning/Doffing of the device and for the therapy activity tasks of exoskeleton use for all domains of the ICF we placed Results a special emphasis on the inclusion of Patient-Reported Twenty-four individuals with SCI with neurological lev- Outcome Measures (PROMs). Two additional question- els of injury ranging from C5 to L3 participated in the naires were therefore included, the Quebec User Evalu- study (see Table 3). For complete injuries (AIS A) only ation of Satisfaction with assistive Technology (QUEST thoracic lesions from T5 to T12 were present in our 2.0) and the Psychosocial Impact of Assistive Devices study population. Motor complete injuries (AIS grades Scale (PIADS). The QUEST 2.0 measures the satisfaction A and B) were present in 16 participants and 8 were clas- with an assistive device and has been used in previous sified as motor incomplete (AIS grades C and D). Twenty studies with exoskeletons [13, 26, 27]. The PIADS assesses individuals sustained the injury within the last year (acute the psychosocial impact of a device on a sum scale from or subacute), while 4 individuals were in the chronic -78 (negative impact) to + 78 (positive impact) [28]. To stage (onset of paralysis injury more than 1 year). On the best of our knowledge, the PIADS has not been imple- average, participants were 45 ± 12 years old and mostly mented in previous studies with mobile exoskeletons used male (70.8%). 58.3% of participants had a traumatic SCI in a clinical setting by people with SCI [29–31]. Thera- while 41.7% had a non-traumatic SCI. pists’ satisfaction was measured once with the QUEST 2.0 On average 10.1 ± 3.5 training sessions were com- at the end of the study period. pleted. Sixteen participants (66.7%) completed all 12 training sessions, 2 participants missed one session (8%), Statistical analysis 2 participants missed three and five sessions respectively Quantitative variables were summarised using stand- (8%), and 4 participants completed less than six training ard descriptive statistics (median, interquartile range sessions (16.6%). Nearly all missed sessions were due to (IQR), mean, standard deviation (SD), minimum and health reasons unrelated to training, along with required maximum). Qualitative variables including (S)AEs were maintenance periods for the device. One participant described using group sizes and frequencies. The dif- could not complete session 12 due to training-related ferences between pre–post training outcome measures back pain, which recovered after 1 week without the need were analysed using the non-parametric Wilcoxon-test. for medical treatment. All participants remained in the For multiple measures, the Friedman test and Wilcoxon study until the follow-up visit, thus there were no drop- post-hoc test with adjusted Bonferroni–Holm correction outs (see Fig. 3). were calculated with IBM SPSS Statistics 28.0. P-values p ≤ 0.05 were considered statistically significant. Table 3 A detailed overview of the characteristics of each study participant Participant Sex Age range Height ranges Weight ranges Days from Injury NLI AIS LEMS LEMS WISCI WISCI SCIM SCIM injury at day aetiology screening post- screening post- mobility mobility of IC training training baseline follow-up 1 Male 61–70 176–180 96–100 22 NT L3 B 20 21 1 6 17 19 2 Female 61–70 161–165 66–70 15 NT T10 D 39 44 0 8 15 19 3 Female 61–70 166–170 56–60 36 NT T11 A 0 0 0 0 16 17 4 Female 18–30 166–170 46–50 2184 T T7 A 0 0 6 5 19 18 5 Female 31–45 171–175 56–60 670 NT T12 B 7 7 13 13 21 21 6 Male 46–60 176–180 86–90 2941 T T6 A 0 0 0 0 14 18 7 Female 31–45 161–165 86–90 6202 T T9 A 0 0 0 0 14 13 Wright et al. Journal of NeuroEngineering and Rehabilitation 8 Male 31–45 176–180 71–75 115 T T12 A 0 0 6 6 15 17 9 Male 46–60 176–180 66–70 168 T T10 A 0 0 0 3 15 17 10 Male 31–45 181–185 71–75 54 T T10 A 0 0 1 4 17 19 11 Male 31–45 171–175 51–55 55 NT T1 D 46 – 8 – 18 22 12 Male 46–60 176–180 76–80 104 NT T7 D 46 – 4 – 18 20 (2023) 20:45 13 Male 31–45 186–190 71–75 92 T L3 B 21 21 9 9 18 23 14 Male 31–45 161–165 51–55 120 T C5 B 0 0 3 3 13 12 15 Male 31–45 176–180 81–85 123 T T5 A 0 0 1 3 14 15 16 Female 31–45 161–165 56–60 64 T T11 D 24 27 5 9 18 21 17 Male 31–45 176–180 71–75 71 T L3 B 18 28 9 12 18 24 18 Male 46–60 181–185 71–75 179 NT T5 D 38 38 13 13 21 21 19 Male 31–45 176–180 76–80 79 NT T7 D 40 46 13 16 16 31 20 Male 31–45 181–185 76–80 55 T T11 A 0 0 3 3 18 19 21 Male 46–60 181–185 71–75 107 T T6 C 19 22 3 6 16 19 22 Male 31–45 181–185 71–75 94 T T7 B 0 0 3 5 18 19 23 Male 31–45 186–190 76–80 111 NT L1 D 36 40 4 8 17 21 24 Female 31–45 161–165 76–80 185 NT T11 B 0 1 1 1 15 16 Ranges of age, height and weight are given to ensure anonymity of the study participants. The neurological level and the severity of the lesion are given together with the lower extremity motor scores, the dependency on walking aids, and the functional status before and after the training IC informed consent; NLI neurological level of injury; AIS American Spinal Injury Association (ASIA) Impairment Scale; T traumatic; NT non-traumatic; LEMS Lower Extremity Motor Score according to the International Standards for Neurological Classification of SCI (ISNCSCI) [12]; WISCI Walking Index for Spinal Cord Injury, measured without the ABLE Exoskeleton; SCIM Spinal Cord Independence Measure Page 7 of 18 Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 8 of 18 Primary study outcome Pre-Screening Safety outcome measures Germany: 176 Spain: 262 A total of 81 AEs were reported during the 242 sessions Total: 438 performed throughout the study across the two sites. Eight (9.9%) of these events, occurring in 6 participants, were classified as device-related AEs, i.e. rated as ‘related’ Screening to the device by the assessors. None of the AEs were Germany: 7 rated as serious. Other reported events without a medical Spain: 17 occurrence were classified as device deficiencies accord- Total: 24 ing to ISO 14155:2011. No falls or fractures occurred. Table 4 summarises the types of AEs that occurred during the study, as well as the percentage of partici- Enrolment pants affected. The most reported AEs corresponded to Germany: 7 skin lesions (33.3%) and pain (27.2%) in 13 and 10 par- Spain: 17 Total: 24 ticipants, respectively. Three participants experienced isolated episodes of orthostatic hypotension either at the start of a session before standing up, or upon standing Training completed up, with one of the participants unable to continue with the remainder of that session. Medication was changed 12 Sessions < 12 Sessions Germany: 4 Germany: 3 for two of these participants prior to the subsequent ses- Spain: 12 Spain: 5 sion, whilst the third participant remained on the same Total: 16 Total: 8 medication management. All three participants were able to continue with the training sessions as planned without further problems. Drop-outs Most of the device-related skin lesions were mild pres- Total: 0 sure injuries (grade I according to the EPUAP scale) (3/5) and were caused by the contact between the par- Follow-up ticipants’ body and the exoskeleton structure. These Germany: 7 were located at the lower back at the level of the already Spain: 17 padded lumbar module and at the instep area under the Total: 24 foot straps. All skin injuries were resolved without fur- Fig. 3 Study flow chart of recruited participants in line with the ther complications (average recovery time of 5 days) by STROBE statement (http://​www.​strob​estat​ement.​org) re-adjusting the fitting of the exoskeleton segments and, Table 4 Overview of adverse events (AEs) Type of AE Number (% out Device-related (% Procedure-related Underlying disease- Other causes or of total AEs) out of AE-type) (% out of AE-type) related (% out of undetermined relation (% out AE-type) of AE-type) Skin lesion 27 (33.3) 5 (18.5) 0 (0) 1 (3.7) 21 (77.8) Pain 22 (27.2) 3 (13.6) 6 (27.3) 4 (18.2) 9 (40.9) Neuropathic pain 9 (11.1) 0 (0) 0 (0) 9 (100) 0 (0) Urinary tract/gastro‑ 8 (9.9) 0 (0) 0 (0) 5 (62.5) 3 (37.5) intestinal infection Fatigue 3 (3.7) 0 (0) 3 (100) 0 (0) 0 (0) Spasticity 3 (3.7) 0 (0) 0 (0) 3 (100) 0 (0) Hypotension 5 (6.2) 0 (0) 3 (60.0) 2 (40.0) 0 (0) Inflammation 2 (2.5) 0 (0) 0 (0) 0 (0) 2 (100) Other 2 (2.5) 0 (0) 1 (50) 0 (0) 1 (50) Total 81 (100) 8 (9.9) 13 (16.0) 24 (29.6) 36 (44.4) Type and number of AEs that occurred during the clinical investigation Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 9 of 18 when needed, adding extra padding to the critical skin Table 5 Time to Don and Doff the ABLE Exoskeleton areas. Two device-related skin lesions occurring in the Mean ± SD Median; IQR Range same participant were rated as moderate (grade II sacral pressure injury). These led to a training pause of 8 weeks. tDON (min:sec) 04:43 ± 1:52 04:15; 2:18 2:00–12:30 Due to an insufficient effect of adding extra padding, tDOFF (min:sec) 02:07 ± 1:10 01:43; 01:32 0:30–7:24 the training was resumed by using a hydrocolloid and a tDON + ­tDOFF (min:sec) 06:50 ± 2:50 06:05; 3:39 2:30–15:21 foam plaster over the critical skin areas during training. Mean, median and range of time to Don and Doff the device are shown Although the skin of this study participant was intact at screening, it is important to note that this participant had a known history of pressure injuries at the right ischial Feasibility and usability outcome measures tuberosity before involvement in the clinical study and Donning and doffing of the device required an average therefore had a generally high risk for developing pres- total time of 6 min and 50 s (Table 5). sure injuries. Of the 21 skin issues that were assigned to There was a significantly reduced LoA over the training other causes, 11 were rated as probably device-related. sessions for donning and doffing (p < 0.001 for 16 partici- Most of these cases (10/11) were non-blanchable red- pants who completed all 12 training sessions). At the end ness (EPUAP grade I) that appeared directly after the of the training, the majority of participants (68.8%) were exoskeleton training and recovered the same day without able to complete both donning and doffing either inde- the need for any medical treatment. Mild bruising on the pendently (25%), with supervision only (18.8%) or with shin was noted before the 11th session for one partici- minimal assistance (25%) (see Fig. 4). pant, however, this was resolved by the end of training. Independence to carry out the activity tasks (see Fig. 5) It was not clear if this was device-related given that he in the device (Sit-to-stand, Walk 10m, Turn 180º, Stand- was involved in many other therapy activities, and that he to-sit) increased significantly for all participants as the did not present with this bruising in any of the 10 pre- training progressed (p < 0.001 for 16 participants, who vious sessions. Two skin lesions on the lower leg were completed all 12 training sessions). In session 1, 54.2% rated as “possibly and unlikely” device-related, two skin of the 24 participants were able to walk 10 m with the lesions on the foot were classified as “probably or pos- device, and required total assistance (4.2%), maximum sibly related” to the medical procedure and one redness assistance (20.8%) or moderate assistance (29.2%). By on the back could have a “possible relation” to the disease session 12, the 16 participants who completed all train- condition. Five skin lesions happened between the train- ing sessions were able to walk 10 metre with the device, ing sessions and were not related to any of the three cat- and 56.3% of them required low levels of assistance egories (device-, procedure- or disease-related). (37.5% with minimum assistance and 18.8% with total Most of the device-related pain issues were mild (2/3) independence). and were resolved in the same or subsequent session. In session 1 all participants were able to complete One event of a device deficiency caused a pain score of sit-to-stand and stand-to-sit tasks, with one partici- VAS 8/10. The exoskeleton lost connection to the mobile pant able to do so with a minimum level of assistance. application and as a consequence, the participant needed By the end of the training programme, 50.0% of the to be assisted to sit down on the chair with extended 16 participants who completed session 12 were able legs which led to pain in the posterior part of the knee. to Sit-to-Stand with minimum assistance, and 68.8% The pain resolved immediately after the exoskeleton was performed stand-to-sit with low levels of assistance doffed and the participant sustained no further issues. (31.3% minimum assistance 25.0% supervision, 12.5% Including the above-mentioned device deficiency that total independence). The independence to carry out the led to a device-related AE, a total of 16 device deficien- 180º turn task also improved throughout the training cies have been reported during the clinical investigation, programme, with 62.5% of the participants completing of which, 15 were classified as malfunctions and one as a this task with minimum assistance (43.8%), supervision user error of the participant. Two of these malfunctions (6.3%) or total independence (12.5%) in session 12. corresponded to mechanical issues of the exoskeleton, while the rest were related to the exoskeleton firmware, the mobile app software or a combination of both. All Therapy metrics device deficiencies were promptly addressed and resolved A mean therapy time per session of 34.8 min ± 6.5 by the manufacturer and the changes made to the devices (median 34.9, IQR 6.0 min, range 13.1–57.5 min) was were reported to the regulatory agencies. Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 10 of 18 (A) (B) LoA Donning LoA Doffing # participants 24 22 21 20 20 20 19 19 18 18 17 16 24 22 21 20 20 20 19 19 18 18 17 16 100% Percentage of participants 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Training session Training session Total Assistance Maximum Assistance Moderate Assistance Minimum Assistance Supervision Independence Fig. 4 Progression of the Level of Assistance in donning and doffing. Level of Assistance (LoA) for A Donning and B Doffing is shown over the course of the training sessions achieved. Figure 6 shows a clear trend towards higher session 12 compared to session 1 (Table 7). In addition standing and walking times throughout the train- to the 16 participants who completed all 12 sessions, 2 ing programme. As stated previously, sessions 1 and participants missed 2 and 6 sessions respectively, but still 12 were used to perform outcome measures with the completed walking tests with the exoskeleton at sessions exoskeleton, including standardised rest breaks, there- 1 and 12 and were therefore included in final calcula- fore therapy metrics for these sessions were not con- tions. Participants who were able to complete the TUG, sidered. Participants’ mean walking time in session 11 10MWT and 6MWT at sessions 1 and 12 showed a sig- was 2.6 times higher compared to session 2 for those nificant improvement (p < 0.01), with average gait speed who completed these training sessions (n = 17). Besides during the 10MWT being 2.1 times faster in session 12 the 16 participants who completed all 12 sessions, one (0.3 [m/s] ± 0.1, 0.1–0.5) compared to session 1 (mean additional participant completed sessions 2–11 (only 0.1 [m/s] ± 0.1, 0–0.3). These participants also walked a missing session 12 due to back pain) and was there- greater distance (an average of 1.9 times further) during fore considered for this calculation of therapy met- the 6MWT at the end of the training (Table 7). rics. The mean number of steps, speed and distance all The BORG-scale was measured after the 6MWT even increased by 3.0, 1.8 and 2.9 times respectively, from when the test was aborted (Table 7). A significant reduc- session 2 to 11. tion of the RPE was seen for individuals who performed All participants who completed sessions 2–11 showed this test in session 1 and session 12 with the exoskeleton a significant improvement in walking time, number of (p < 0.05). One participant had to abort all 3 walking tests steps, speed and distance from beginning to the end of at session 12, but still completed the BORG-scale. the training programme (p < 0.001) (Table 6, Fig. 7). No This trend was also observed in the walking tests com- significant differences were found between successive pleted without the device. There were 3 more participants sessions. who were able to complete the walking tests without the device at post-training compared to baseline, with signifi- Secondary outcome measures cant improvements in TUG and 6MWT (p < 0.05). The Activity outcomes BORG-scale was measured without the exoskeleton for An increase was seen in the number of participants who 12 (50%) participants post-training compared to 8 (33%) were able to complete all walking tests with the device in at baseline, with no change seen in scoring. The median Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 11 of 18 (A) (B) LoA Sit-to-Stand LoA Walk 10m # participants 24 22 21 20 20 20 19 19 18 18 17 16 24 22 21 20 20 20 19 19 18 18 17 16 100% Percentage of participants 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Training session Training session (C) (D) LoA Turn 180° LoA Stand-to-Sit # participants 24 22 21 20 20 20 19 19 18 18 17 16 24 22 21 20 20 20 19 19 18 18 17 16 100% Percentage of participants 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Training session Training session Unable Total Assistance Maximum Assistance Moderate Assistance Minimum Assistance Supervision Independence Fig. 5 Progression of the Level of Assistance in therapy tasks. The Level of Assistance (LoA) for A Sit-to-Stand, B Walk 10m, C Turn 180°, and D Stand-to-Sit tasks is shown over the course of the training sessions WISCI II score changed significantly from 3.0; IQR 6.0 changes in scores in the mobility section (p < 0.05). Sig- (mean 4.3 ± 4.5; range 0–13) pre-training to 5.5; IQR 6.0 nificant improvements in WISCI II and SCIM III scores (mean 6.1 ± 4.6; range 0–16) post-training (p < 0.01). Par- were seen in participants with both motor complete and ticipants showed significant positive changes in the total motor incomplete injuries, however the difference was SCIM III score (p < 0.05) from baseline to follow-up from highest in participants with incomplete SCI within the median 68.0; IQR 10 (mean 68.7 ± 6.1; range 58–76) to first 6 months (Table 3). 72.5; IQR 8 (mean 73.0 ± 6.1; range 59–88), with notable Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 12 of 18 mean 36.1 ± 18.2), as well as in the three subscales, 60 pertaining to “competence” (median 15, IQR 14; mean 50 15.1 ± 8.2), “adaptability” (median 10, IQR 7; mean 10.7 ± 4.8) and “self-esteem” (median 10, IQR 11, mean 40 10.4 ± 6.7). There was no significant difference between post-training and follow-up or between participants Minutes 30 in the subacute vs. chronic stage. It can be noted that the overall PIADS score and all subscores were around 20 the middle of the positive range (+ 1, + 2). Only a minor number of the subscores were reported nega- 10 tively, ranging from 2.2% of the scores for “adaptabil- ity” to 4.0% of the scores for “competence” and 4.3% for 0 2 3 4 5 6 7 8 9 10 11 “self-esteem”. Training session Outcomes of personal and environmental factors Therapy Time Upright Time Walking Time The median satisfaction with the exoskeleton was Fig. 6 Average therapy time progression. The average total therapy rated (total QUEST 2.0 score) as 33.0; IQR 7.5 (mean time (time spent standing, walking or sitting in the exoskeleton) 31.7 ± 5.5) out of 40 at follow-up. No significant differ- together with the standing and walking time for each session of the ences were found between visits when the QUEST 2.0 training programme of the 17 participants that completed sessions was performed (session 1, session 6, session 12 and fol- 2–11 is shown. Error bars represent standard deviation low-up). Results obtained for each of the rated items are shown below in Fig. 8. “Safety” was the best rated item (median 4.5, IQR 1.0; mean 4.4 ± 0.8), followed by “Dura- Outcomes of body function & structure bility” (median 4.0, IQR 1.0; mean 4.1 ± 1.0) and “Device The International Standards for Neurological Classifica- dimensions” (median 4.0, IQR 1.0, mean 4.1 ± 0.9). tion of Spinal Cord Injury (ISNCSCI) lower extremity “Safety” was also rated as the most important category motor score (LEMS) changed significantly over the train- for the majority of the participants (83.3%), followed by ing programme (p < 0.05) with an average of 12% ± 3.8% “Effectiveness” (58.3%) and “Ease of use” (41.7%) (see improvement in LEMS reported for 7 of the 8 motor Fig. 9). incomplete participants (Table 3). All other participants’ A total of 11 physiotherapists completed the QUEST scores were unchanged. 2.0, scoring an average total satisfaction with the exoskel- eton of 31.6 ± 3.2 out of 40. The lightweight characteristic Participation outcomes of the device (“Weight” category) appears to be the best Overall psychosocial impact measured by the PIADS rated category by therapists, followed by “Dimensions” score at follow-up was positive (median 34, IQR 34; and “Durability” (Fig. 8). “Ease of use” was rated as the Table 6 Overview of therapy metrics Session Median (IQR) Mean (SD) Range p-value Walking time (min) 2 3.0 (5.0) 4.8 (6.5) 0–27 < 0.001 6 7.0 (13.0) 9.1 (8.9) 0–31 11 8.0 (17.0) 12.4 (10.7) 0–32 Steps (n) 2 68.0 (128.0) 138.8 (239.7) 0–1025 < 0.001 6 144.0 (335.0) 282.1 (322.8) 0–1288 11 222.0 (610.0) 422.8 (415.5) 4–1301 Speed (m/s) 2 0.1 (0.1) 0.1 (0.1) 0–0.3 < 0.001 6 0.1 (0.1) 0.1 (0.1) 0–0.4 11 0.2 (0.1) 0.2 (0.1) 0.1–0.3 Distance (m) 2 23.0 (59.7) 55.2 (111.9) 0–476.5 < 0.001 6 41.0 (145.1) 108.8 (150.4) 0–602.5 11 87.1 (265.3) 157.8 (164.1) 1.4–491.0 Median, interquartile range (IQR), mean, standard deviation (SD), range and Friedman-test p-value of walking time, steps, speed and distance of session 2, 6 and 11 for 17 participants, who completed sessions 2–11 Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 13 of 18 ** *** 1.250 *** *** 1.000 ** ** Number of steps ** 750 ** ** 500 250 0 2 3 4 5 6 7 8 9 10 11 Training session outliers p<0.01 p<0.001 ** *** Fig. 7 Progression of number of steps with the ABLE Exoskeleton. A boxplot of the number of steps for the 17 participants, who completed sessions 2–11, is shown Table 7 Walking test results with the ABLE Exoskeleton at session 1 and 12 Session Number of Number of participants Median (IQR)* Mean (SD)* Range* p-value* participants completed session 1 and 12 completed (%) (%) TUG (s) S. 1 12 (50) 9 (37.5) 147.0 (133.0) 167.0 (89.2) 70–352 0.004 S. 12 17 (70.8) 86.0 (45.0) 78.2 (27.7) 44–128 10MWT (s) S. 1 13 (54.2) 10 (41.7) 74.5 (115.0) 120.2 (102.7) 32–343 0.002 S. 12 17 (70.8) 35.0 (30.0) 41.3 (20.4) 22–81 6 MWT (m) S. 1 10 (41.7) 9 (37.5) 33.0 (55.0) 48.4 (35.8) 10–118 0.004 S. 12 15 (62.5) 100.0 (109.0) 94.1 (37.0) 22–131 BORG (6–20) S. 1 14 (58.3) 11 (45.8) 15.0 (7.0) 14.3 (4.2) 6–19 0.016 S. 12 18 (75) 11.0 (5.0) 11.9 (3.3) 6–17 Results of the Timed-up and Go test (TUG), the 10 Metre Walk Test (10MWT), the 6 Min. Walk Test (6MWT) and the BORG-scale at sessions 1 and 12 are shown *Only for participants who completed both session 1 and 12 most important category for the majority of therapists we based our protocol on the in- and exclusion criteria, (83.3%), followed by “Safety” and “Effectiveness” (Fig. 9). training methodology and outcome measures most com- monly used in previous exoskeleton trials (see Additional Discussion file 2 for an overview). We placed a special emphasis on With the aim of achieving consistent comparability of the assessment of feedback from exoskeleton end-users our study results with other safety and feasibility studies as well as therapists. of robotic lower-limb exoskeletons in persons with SCI, The assessments used in our study were implemented with the aim of covering all domains of the ICF to ensure Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 14 of 18 Dimensions minor bruising that may have been overlooked in pre- 5 vious studies. In this study, 8 device-related AEs were Effectiveness 4 Weight reported, with most of them (5/8) being of mild severity. 3 Two device-related AEs were reported as a consequence 2 of a device deficiency. In both of these cases, the partici- 1 pant sustained no lasting damage. Other device-related 0 AEs in this study corresponded to skin lesions and pain. Comfort Adjustments Skin lesions and pain in the areas of the body that are in direct contact with the device are well-known risks of the use of exoskeletons, with previous studies reporting between 2 and 13 device-related skin alterations when Easy to use Safety using similar devices with a comparable number of par- ticipants [1, 2, 16]. Episodes of orthostatic hypotension Durability due to the standing up procedure also occurred in this Participants Therapists study, however, these are to be expected given the pro- Fig. 8 QUEST 2.0 mean scores for device satisfaction items. The figure pensity to this issue in the SCI population and have also shows the satisfaction with the device for participants and therapists been noted in other exoskeleton trials [4, 21]. As in previ- at follow-up. Blue and red dots represent the standard deviation both ous studies, no falls occurred [1, 16]. These findings indi- for study participants’ and therapists’ scores, respectively cate that the ABLE Exoskeleton is comparable to other devices in regard to safety. Dimensions Feasibility and usability outcomes 100 The mean average time to don and doff the ABLE Exo- Weight skeleton device was 6 min 50 s. This finding constitutes Effectiveness 75 an improvement with respect to other exoskeletons that 50 reported donning and doffing times averaging between 9 25 and 30 min, suggesting that the use of the ABLE Exoskel- eton is feasible within the timeframes available in real- Comfort 0 Adjustments world rehabilitation settings [2, 6, 13, 24]. Moreover, as participants progressed through the training programme with the exoskeleton, the amount of assistance required to perform both the donning and doffing of the device Easy to use Safety decreased. The majority of participants were able to com- plete this process either independently, with minimum Durability assistance, or with supervision by the end of the training learn to use the Ekso™ exoskeleton found that 3/7 partic- % Participants % Therapists programme. Similarly, a study investigating the effort to Fig. 9 Importance of QUEST 2.0 categories for participants and therapists. The percentage of participants and therapists that selected ipants could independently don the device by the end of each QUEST 2.0 category as the most important is shown study [13]. Although a direct comparison of outcomes of exoskeleton studies is generally challenging due to the dif- ferences in the design of the devices (e.g. the ABLE Exo- the possibility of analysing the meaningfulness of the skeleton investigated in this study is only knee powered results for individuals with SCI. while the ­ EksoGTTM is hip-knee powered), we believe that it is valid to compare devices in respect to usability Safety outcomes aspects. A strict and exhaustive safety assessment was performed By the end of training, 15 of 16 participants were to overcome the reporting limitations of previous studies able to complete all four activity tasks (sit-to-stand, [9]. There are differences in the severity grading of AEs walk 10 m, turn 180° and stand-to-sit) with the ABLE among studies, with events such as fractures not consid- Exoskeleton, with 8 of them able to do so with mini- ered to be serious adverse events in every trial [10]. Thus, mum assistance, supervision or complete independ- we took the most conservative approach and strictly ence. Moreover, all participants showed a reduction in reported all the adverse events that occurred during the assistance needed to complete these tasks by the end of study, including minor issues such as skin marking and the training programme. In 77.7% of cases, training was Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 15 of 18 delivered with the supervision of only one physiothera- individuals with multiple sclerosis, and considerably bet- pist, showing the potential of the ABLE Exoskeleton to ter than the Marsi Active Knee device (22.4 ± 3.2) when reduce the human resources required to deliver func- used by individuals with stroke and multiple sclerosis tional gait training in persons with complete SCI. How- [27]. In respect to weight (Median: 4.0/Mean: 4.0 ± 1.0), ever, the sit-to-stand transfer required most assistance easiness to use (Median: 4.0/Mean: 3.8 ± 0.9) and safety compared to the other tasks. The reason for this is that (Median: 4.5, Mean 4.4 ± 0.8) the ABLE Exoskeleton the knee motors are not strong enough to complete the was also better in comparison to the Marsi Active Knee of the participant. Other studies with the Ekso™ device stand-up process without active arm and trunk support Safety: Median 3.6) and comparable to the EksoGT™ device (Weight: Median 2.8; Easiness to use: Median 2.6; also reported a reduction in assistance needed as train- device (Weight: Mean 3.8 ± 1.0; Easiness to use: Mean to walk in the Ekso™ device for 30 min with minimal ing progressed. Six out of 7 participants managed 3.7 ± 0.9; Safety: Mean 4.3 ± 0.9) [26, 27]. These results show that participants are satisfied with the ABLE Exo- assistance in a median of 8 sessions, with 5 of them able skeleton and its usability. The specific characteristics to do so with a contact guard or close supervision assis- of the ABLE Exoskeleton of being lightweight and of tance in a median of 15 sessions [13]. Another study smaller size were rated best by both participants and the Ekso™ device at the end of an 18-session training reported minimum assistance or less for walking with therapists. Participants rated their satisfaction highest in the QUEST 2.0 with respect to safety when using the programme, whilst a need for moderate or maximum device, while therapists scored it slightly lower. This find- assistance was required for all sit-stand transitions [21]. ing indicates that participants are more satisfied with the As presented in our study, all participants increased safety of the device than the therapists. A reason for this their level of independence to complete all four therapy could be that the therapists answered the QUEST after activities throughout the training programme, demon- completing sessions with several participants where they strating how the ABLE Exoskeleton can be used as a had to provide different levels of assistance, in contrast tool during rehabilitation in clinical settings for loco- to the participants who rated only their own experience. motion training, with increasing independent use. However, despite the difference in satisfaction scores Participants’ walking time at the end of training was 2.6 between therapists and participants, the overall level of times longer than at the start, indicating an increasing satisfaction scores amongst therapists still suggests a high tolerance for gait training with the ABLE Exoskeleton, acceptance of the ABLE Exoskeleton, which represents which is crucial to deliver intensive therapy as soon as an essential prerequisite for its integration into clinical possible after the injury in our cohort which contained gait training programmes. To the best of our knowledge, mainly participants in the acute or subacute stage. this is the first study where the QUEST 2.0 was used to Results from the device metrics in this study were in evaluate the satisfaction with the device among clinical keeping with other exoskeleton studies, with increases professionals. Therapists represent an important stake- in the average number of steps (300%), speed (180%) holder group responsible for the successful implementa- Similarly, studies with the Ekso™ device also reported and distance (290%) respectively from session 2 to 11. tion of new technology into clinical practice [33]. The findings of the psychosocial impact assessment an increase in the number of steps of 249% within 18 (PIADS) in this study suggest that the ABLE Exoskeleton, training sessions and 352% within a 25 session train- used in a clinical setting, may have a positive impact on ing programme, respectively [1, 21]. The wide range and quality of life and well-being, to be added to the more vis- standard deviation of the results in our study could be ible gait and motor function improvements. It should be caused by our broad in- and exclusion criteria, which noted that the PIADS is more relevant to persons in the allowed us to recruit acute, subacute and chronic individ- chronic stage of SCI. In our study, no differences in the uals with a wide range of motor impairments. However, PIADS results between participants in the subacute and as our results show, participants improved significantly chronic phase after SCI were seen, but as there were only over the course of the gait training programme in all gait three participants in the chronic group, no conclusions parameters, similar to previous studies. can be made on this. Future studies should investigate the psychosocial impact of exoskeleton use on individuals Outcomes of participation and personal/environmental with chronic SCI. factors The average satisfaction score obtained for the ABLE Activity outcomes parable to the EksoGT™ exoskeleton (31.3 ± 5.70) [26] Exoskeleton (31.6 ± 5.7) in study participants was com- Although demonstrating the efficacy of the ABLE Exo- skeleton was not the primary endpoint of this study, and the ReWalk device (29.4 ± 2.5) [32] when tested in significant improvements for outcomes on exertion Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 16 of 18 and gait were measured by the BORG-scale, 10MWT, selection bias towards physically fit and cooperative par- 6MWT and TUG performed with the exoskeleton were ticipants was present. This limits the generalisation of the achieved and are consistent with the results from other results to the SCI population as a whole. The heterogeneity devices [6]. Similarly, improved LEMS scores seen after of the study population together with the relatively small training in participants in this study with motor incom- sample size lowered the statistical power in calculations. plete lesions were also reported in previous exoskeleton Furthermore, the primary endpoint of this study was on trials [4, 22]. Besides the functional aspects, the average safety and feasibility and not on efficacy. Since many study value of the BORG-scale indicates that the training with participants were in the acute and subacute phase after the ABLE Exoskeleton represents a moderate-intensity injury, during which spontaneous recovery occurs, no physical exercise with an expected positive effect on the causal relationship between the observed improvements cardiovascular system. However, another clinical study and the exoskeleton use can be made. Finally, due to the comparing the knee-powered ABLE Exoskeleton with COVID-19 pandemic, there was reduced access to poten- Knee-Ankle-Foot orthoses (KAFOs) found no signifi- tial participants who could have participated, with both cant difference of performance in these walking tests. centres initially unable to recruit outpatients and also a This study proposes that providing powered assistance reduction of in-patients being admitted for neurorehabili- only on the knee joints is not enough to provide sufficient tation. Despite the challenging circumstances we were still trunk stability and reduced metabolic costs [11]. We also able to recruit 24 participants with a wide range of motor identified the lack of trunk stability in cervical and high impairments and time after SCI. thoracic lesions, but also the presence of spasticity in leg muscles and missing proprioception as main factors lim- Conclusions iting the exoskeleton therapy outcome. The knowledge In conclusion, the results of our multicentre study show gained in this clinical study in this regard led to some that the ABLE Exoskeleton is safe and feasible for gait changes in the design of the ABLE Exoskeleton, which training in persons with SCI with neurological injury now includes motor-powered hip joints, optional shoul- levels ranging from C5 to L3 in a rehabilitation hospi- der straps and an articulated ankle joint, to increase the tal setting, regardless of the time since injury (acute or support provided to the trunk and therefore extend the chronic) and its severity (motor complete or incomplete group of potential end users. SCI). Similarly, it was feasible for all the participants to Furthermore, the significant improvement in the WISCI II don and doff the device within an efficient timeframe, scores seen in this study shows that participants also devel- which maximises the effective gait training time of each oped less dependency on walking aids without the exoskel- rehabilitation session. The degree of independence to eton [29]. Another study also reported significant changes in complete therapy activities improved throughout the the Ekso™ device in recently injured participants [4]. Posi- walking ability within an 8-week training programme with training, whilst also making progress with gait perfor- mance. Furthermore, the protocol formulated for this tive changes in the SCIM III were seen after training in this study has intended to bring together the key aspects study and have also been identified post-training in a 2-week of previous safety and feasibility trials with lower-limb training programme by a study with the AIDER exoskeleton, exoskeletons. Thus, we were able to maximise the com- indicating improvements in the degree of independence to parability of our results with these previous studies. It is complete activities of daily living [22]. Whilst it is not pos- hoped that our study protocol can serve as a framework sible to conclude that all of these changes have occurred only for future exoskeleton trials to allow for more consist- as a direct result of exoskeleton training due to a number of ency and comparability of results. Such a framework is confounding variables and lack of a control group, they do not only beneficial for the early evaluation of a specific suggest that the participants became more tolerant to gait exoskeleton against other types of exoskeletons and/or training in the device, were able to master the basic tech- standard rehabilitation interventions for persons with nique of walking more efficiently by the end of training, and SCI, but also for the assessment of its impact on func- had an improved function outside of training sessions. These tion and health in a community-based setting outside results suggest that gait training with the ABLE Exoskeleton the hospital. has the potential to become an effective tool for improving standing and walking function of individuals with SCI in Abbreviations combination with conventional rehabilitation techniques. 10MWT 10 Metre Walk Test 6MWT 6 Minute Walk Test AD Autonomic dysreflexia Study limitations AE Adverse events There were several limitations to this study. Firstly, this was AIS American Spinal Injury Association Impairment Scale CVS Cardiovascular system an open and non-blinded/non-controlled study, therefore Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 17 of 18 DBP Diastolic blood pressure project “20121 ABLE Exoskeleton”, an activity that has received funding from DVT Deep venous thrombosis EIT Health. EIT Health is supported by the European Institute of Innovation DXA Dual Energy X-Ray Absorptiometry and Technology (EIT), a body of the European Union. EIT Health is co-funded EPUAP European Pressure Ulcer Advisory Panel by the European Union. The main objective of EIT Health is to promote health‑ ICF International Classification of Functioning, Disability and Health care innovation and its impact on the improvement of citizens’ quality of life. ISNCSCI International Standards for Neurological Classification of Spinal Open Access funding enabled and organized by Projekt DEAL. Cord Injury IQR Interquartile range Availability of data and materials LEMS Lower Extremity Motor Score The datasets used and analysed during the current study are available from LoA Level of Assistance the corresponding author on reasonable request. PIADS Psychosocial Impact of Assistive Devices Scale PROMs Patient-Reported Outcome Measures QoL Quality of life Declarations QUEST Quebec User Evaluation of Satisfaction with assistive Technology ROM Range of Motion Ethics approval and consent to participate RPE Rate of perceived exertion All sites received ethics approval from their respective Ethics Committee and SAE Serious adverse events National Competent Authority. All subjects provided signed informed consent SBP Systolic blood pressure prior to participation. Site approvals were as follows: (1) Heidelberg University SCI Spinal cord injury Hospital: Ethics Committee of the Medical Faculty of Heidelberg University, SCIM Spinal cord independence measure MZmo-557/2020, (2) Institut Guttmann: Fundació Unió Catalana d’Hospitals, STROBE Strengthening the Reporting of Observational Studies in CEI 20/40. Epidemiology SD Standard deviation Consent for publication TUG​ Timed up and go N/A. VAS Visual Analogue Scale WISCI Walking Index for Spinal Cord Injury Competing interests AC, AMV and JLP were employees of ABLE Human Motion S.L., the sponsor of the clinical investigation when the study was conducted and the publication Supplementary Information prepared. The online version contains supplementary material available at https://​doi.​ org/​10.​1186/​s12984-​023-​01165-0. Author details 1 Institut Guttmann, University Institute attached to the Universitat Autònoma de Barcelona, Badalona, Barcelona, Spain. 2 Universitat Autònoma de Barcelona, Additional file 1. Literature review results: This document provides the Bellaterra, Cerdanyola del Vallès, Spain. 3 Fundació Institut d’Investigació en detailed results of a literature review of studies that evaluated the safety, Ciències de la Salut Germans Trias i Pujol, Badalona, Barcelona, Spain. 4 Spinal feasibility, and usability of exoskeletons for the SCI population, which was Cord Injury Center, Heidelberg University Hospital, Schlierbacher Landstraße conducted on PubMed, Science Direct, Cochrane library, DOAJ and BMJ 200a, 69118 Heidelberg, Germany. 5 ABLE Human Motion S.L., Barcelona, Spain. between the 7th and 30th January 2020 for studies completed since 1 January 2010. Received: 7 June 2022 Accepted: 30 March 2023 Additional file 2. Recommendations for clinical study protocols of early studies with robotic lower limbs exoskeletons. This document lists the recommended inclusion/exclusion criteria, training methodology and outcome measures as a framework for future clinical trials to assess the safety, feasibility and usability of robotic lower-limb exoskeletons in References individuals with SCI. 1. McIntosh K, Charbonneau R, Bensaada Y, Bhatiya U, Ho C. The safety and feasibility of exoskeletal-assisted walking in acute rehabilitation after spinal cord injury. Arch Phys Med Rehabil. 2020;101(1):113–20. Acknowledgements 2. Tefertiller C, Hays K, Jones J, Jayaraman A, Hartigan C, Bushnik T, The authors would like to thank all study participants for their willingness to et al. Initial outcomes from a multicenter study utilizing the indego contribute to the scientific progress during the challenging conditions of the powered exoskeleton in spinal cord injury. Top Spinal Cord Inj Rehabil. COVID-19 pandemic. Furthermore, the authors would like to thank the physi‑ 2018;24(1):78–85. otherapists Irene Hartmann, Jonathan Alt, Tomas Lemlein, Eva-Maria Schmidt, 3. Singh A, Tetreault L, Kalsi-Ryan S, Nouri A, Fehlings MG. Global preva‑ Josep Medina, Lucas Conesa, Montse Corrons and Narda Murillo for conduct‑ lence and incidence of traumatic spinal cord injury. Clin Epidemiol. ing the training sessions with the ABLE Exoskeleton and Christan Schuld for 2014;6:309–31. his help with the statistical analysis. 4. Bach Baunsgaard C, Vig Nissen U, Katrin Brust A, Frotzler A, Ribeill C, Kalke YB, et al. Gait training after spinal cord injury: safety, feasibility and gait Author contributions function following 8 weeks of training with the exoskeletons from Ekso MAW and FH contributed equally to study design, data collection, data analy‑ Bionics. Spinal Cord. 2018;56(2):106–16. sis and manuscript writing. CH, SF, NW and JV helped with the recruitment 5. Rodríguez-Fernández A, Lobo-Prat J, Font-Llagunes JM. Systematic review and screening of participants. AC-C, AM-V and JL-P contributed to the design on wearable lower-limb exoskeletons for gait training in neuromuscular of the study protocol, the management of the approval process, provided 6. Kolakowsky-Hayner S. Safety and feasibility of using the Ekso™ Bionic impairments. J Neuroeng Rehabil. 2021;18(1):22. clinical training and technical support throughout the study and reviewed the manuscript. It is important to note that these three authors were not involved exoskeleton to aid ambulation after spinal cord injury. J Spine. 2013. in the collection, analysis or production of data presented in this paper. EO https://​doi.​org/​10.​4172/​2165-​7939.​S4-​003. and RR designed the study protocol and reviewed the manuscript. All authors 7. Zeilig G, Weingarden H, Zwecker M, Dudkiewicz I, Bloch A, Esquenazi A. read and approved the final manuscript. Safety and tolerance of the ReWalk exoskeleton suit for ambulation by people with complete spinal cord injury: a pilot study. J Spinal Cord Med. Funding 2012;35(2):96–101. For the publication fee, we acknowledge financial support by Deutsche 8. Contreras-Vidal JL, Bhagat NA, Brantley J, Cruz-Garza JG, He Y, Manley Forschungsgemeinschaft within the funding programme “Open Access Q, et al. Powered exoskeletons for bipedal locomotion after spinal cord Publikationskosten” as well as by Heidelberg University. This study is part of the injury. J Neural Eng. 2016;13(3):031001. Wright et al. Journal of NeuroEngineering and Rehabilitation (2023) 20:45 Page 18 of 18 9. He Y, Eguren D, Luu TP, Contreras-Vidal JL. Risk management and regula‑ 27. Puyuelo-Quintana G, Cano-de-la-Cuerda R, Plaza-Flores A, Garces- tions for lower limb medical exoskeletons: a review. Med Devices (Auckl). Castellote E, Sanz-Merodio D, Goñi-Arana A, et al. A new lower limb 2017;10:89–107. portable exoskeleton for gait assistance in neurological patients: a proof 10. Miller LE, Zimmermann AK, Herbert WG. Clinical effectiveness and of concept study. J Neuroeng Rehabil. 2020;17(1):60. safety of powered exoskeleton-assisted walking in patients with spinal 28. Fundarò C, Giardini A, Maestri R, Traversoni S, Bartolo M, Casale R. Motor cord injury: systematic review with meta-analysis. Med Devices (Auckl). and psychosocial impact of robot-assisted gait training in a real-world 2016;9:455–66. rehabilitation setting: a pilot study. PLoS ONE. 2018;13(2): e0191894. 11. Rodríguez-Fernández A, Lobo-Prat J, Tarragó R, Chaverri D, Iglesias X, 29. Day H, Jutai J, Campbell KA. Development of a scale to measure the psy‑ Guirao-Cano L, Font-Llagunes JM. Comparing walking with knee-ankle- chosocial impact of assistive devices: lessons learned and the road ahead. foot orthoses and a knee-powered exoskeleton after spinal cord injury: a Disabil Rehabil. 2002;24(1–3):31–7. randomized, crossover clinical trial. Sci Rep. 2022;12(1):19150. 30. Devitt R, Chau B, Jutai JW. The effect of wheelchair use on the qual‑ 12. Rupp R, Biering-Sørensen F, Burns SP, Graves DE, Guest J, Jones L, Read ity of life of persons with multiple sclerosis. Occup Ther Health Care. MS, Rodriguez GM, Schuld C, Tansey KE, Walden K, Kirshblum S. Interna‑ 2004;17(3–4):63–79. tional standards for neurological classification of spinal cord injury. Top 31. Rigby P, Ryan S, Joos S, Cooper B, Jutai JW, Steggles I. Impact of electronic Spinal Cord Inj Rehabil. 2012;27(2):1–22. https://​doi.​org/​10.​46292/​sci27​ aids to daily living on the lives of persons with cervical spinal cord inju‑ 02-1. ries. Assist Technol. 2005;17(2):89–97. 13. Kozlowski AJ, Bryce TN, Dijkers MP. Time and effort required by persons 32. Kozlowski AJ, Fabian M, Lad D, Delgado AD. Feasibility and safety of a with spinal cord injury to learn to use a powered exoskeleton for assisted powered exoskeleton for assisted walking for persons with multiple walking. Top Spinal Cord Inj Rehabil. 2015;21(2):110–21. sclerosis: a single-group preliminary study. Arch Phys Med Rehabil. 14. Wu CH, Mao HF, Hu JS, Wang TY, Tsai YJ, Hsu WL. The effects of gait train‑ 2017;98(7):1300–7. ing using powered lower limb exoskeleton robot on individuals with 33. Turchetti G, Vitiello N, Trieste L, Romiti S, Geisler E, Micera S. Why effective‑ complete spinal cord injury. J Neuroeng Rehabil. 2018;15(1):14. ness of robot-mediated neurorehabilitation does not necessarily influ‑ 15. Spungen AM, Asselin PK, Fineberg DB, Kornfeld SD, Harel NY, editors. ence its adoption. IEEE Rev Biomed Eng. 2014;7:143–53. Exoskeletal-assisted walking for persons with motor-complete paraple‑ gia. North Atlantic Treaty Organization (NATO), research and technology organization, human factors and medicine panel, HFM-228 symposium; Publisher’s Note 2013 April 15–17; Milan. Springer Nature remains neutral with regard to jurisdictional claims in pub‑ 16. Yang A, Asselin P, Knezevic S, Kornfeld S, Spungen AM. Assessment of lished maps and institutional affiliations. in-hospital walking velocity and level of assistance in a powered exo‑ skeleton in persons with spinal cord injury. Top Spinal Cord Inj Rehabil. 2015;21(2):100–9. 17. Sale P, Russo EF, Russo M, Masiero S, Piccione F, Calabro RS, et al. Effects on mobility training and de-adaptations in subjects with Spinal Cord Injury due to a Wearable Robot: a preliminary report. BMC Neurol. 2016;16:12. 18. Craven B, Blencowe L, McGillivray C, Adachi J. Detection and treatment of sublesional osteoporosis among patients with chronic spinal cord injury: proposed paradigms. Top Spinal Cord Inj Rehabil. 2009;14:1–22. 19. Ditunno JF Jr, Ditunno PL, Scivoletto G, Patrick M, Dijkers M, Barbeau H, Burns AS, Marino RJ, Schmidt-Read M. The Walking Index for Spinal Cord Injury (WISCI/WISCI II): nature, metric properties, use and misuse. Spinal Cord. 2013;51(5):346–55. https://​doi.​org/​10.​1038/​sc.​2013.9. 20. Meseguer-Henarejos AB, Sánchez-Meca J, López-Pina JA, Carles- Hernández R. Inter- and intra-rater reliability of the Modified Ashworth Scale: a systematic review and meta-analysis. Eur J Phys Rehabil Med. 2018;54(4):576–90. https://​doi.​org/​10.​23736/​S1973-​9087.​17.​04796-7. 21. Gagnon DH, Escalona MJ, Vermette M, Carvalho LP, Karelis AD, Duclos C, et al. Locomotor training using an overground robotic exoskeleton in long-term manual wheelchair users with a chronic spinal cord injury living in the community: lessons learned from a feasibility study in terms of recruitment, attendance, learnability, performance and safety. J Neuro‑ eng Rehabil. 2018;15(1):12. 22. Xiang XN, Ding MF, Zong HY, Liu Y, Cheng H, He CQ, et al. The safety and feasibility of a new rehabilitation robotic exoskeleton for assisting individuals with lower extremity motor complete lesions following spinal cord injury (SCI): an observational study. Spinal Cord. 2020;58(7):787–94. 23. National Pressure Ulcer Advisory Panel, European Pressure Ulcer Advisory Panel and Pan Pacific Pressure Injury Alliance. Prevention and treatment Ready to submit your research ? Choose BMC and benefit from: of pressure ulcers: quick reference guide. Haesler E, editor. Osborne Park: Cambridge Media; 2014. • fast, convenient online submission 24. Hartigan C, Kandilakis C, Dalley S, Clausen M, Wilson E, Morrison S, et al. Mobility outcomes following five training sessions with a powered • thorough peer review by experienced researchers in your field exoskeleton. Top Spinal Cord Inj Rehabil. 2015;21(2):93–9. • rapid publication on acceptance 25. Birch N, Graham J, Priestley T, Heywood C, Sakel M, Gall A, et al. Results of • support for research data, including large and complex data types the first interim analysis of the RAPPER II trial in patients with spinal cord injury: ambulation and functional exercise programs in the REX powered • gold Open Access which fosters wider collaboration and increased citations walking aid. J Neuroeng Rehabil. 2017;14(1):60. • maximum visibility for your research: over 100M website views per year 26. Fernández-Vázquez D, Cano-de-la-Cuerda R, Gor-García-Fogeda MD, Molina-Rueda F. Wearable robotic gait training in persons with At BMC, research is always in progress. multiple sclerosis: a satisfaction study. Sensors (Basel, Switzerland). 2021;21(14):4940. Learn more biomedcentral.com/submissions