Pediatric powered mobility training: powered wheelchair versus simulator-based practice

ABSTRACT Many children with physical disabilities lack independent mobility. Powered mobility can be a viable option, but to become proficient drivers, children need opportunities to practice. As is often the case, practice powered wheelchairs are scarce and direct therapy hours dedicated to powered mobility are often limited. Hence, alternative options are needed to enable safe, unsupervised practice. Simulator-based learning has been shown to be an effective training method for powered mobility and other skill-based tasks for adults. The goal of this study was to compare two training methods of powered mobility, powered wheelchair (control group) versus simulator-based (experimental group) practice to determine whether simulation is a feasible and effective method for youth. Method Participants included 30 children and adolescents (23 males, 13 females) with cerebral palsy and other neuromuscular diseases, aged 6–18. Data were collected and compared at baseline and after 12 weeks of home-based practice via a powered wheelchair or a simulator. Powered mobility ability was determined by the Powered Mobility Program (PMP), the Israel Ministry of Health’s Powered Mobility Proficiency Test (PM-PT) and the Assessment of Learning Powered Mobility (ALP). Results All participants practiced for the required amount of time and both groups reported a similar user experience. Both groups achieved significant improvement following the practice period as assessed by the PMP and PM-PT assessments, with no significant differences between them. A significant improvement was found in the ALP assessment outcomes for the powered wheelchair group only. Conclusions This is the first study, to our knowledge, that compares two different wheelchair training methods. Simulator-based practice is an effective training option for powered mobility for children with physical disabilities aged 6–18 years old, demonstrating that it is possible to provide driving skill practice opportunities safe, controlled environments.


Introduction
Self-locomotion is a major milestone in a child's development (Anderson et al., 2013) since it facilitates exploration of the environment, peer interaction and communication while improving children's self-esteem and enhancing societal perception of their abilities (Adolph & Franchak, 2017;Anderson et al., 2013;Campos et al., 2000).When children with physical disabilities lack independent mobility, their development is delayed and disparities with typically developing peers are widened (Anderson et al., 2013;Campos et al., 2000).
Powered mobility (PM) is a viable option for young children (Butler et al., 1983;Galloway et al., 2008).Being able to drive safely entails key skills including self-navigation, spatial perception, environmental awareness, planning and judgment (Smith et al., 2020); thus most children become proficient drivers only after considerable practice (Gefen et al., 2019;Jones et al., 2012).Practice options may be limited due to a lack of access to powered wheelchairs (PW), insufficient therapist time and the need for supervision when practicing.Simulator-based practice allows the user a safe environment to practice PM without constant supervision (Arlati et al., 2020) and has been shown to be an effective learning tool for adults with motor or cognitive impairments (Archambault et al., 2012, Tao & Archambault, 2016).Simulation-based practice may include variations of a given route as well as providing gradual increases in difficulty as skill level improves.These and other assets known in the virtual reality (e.g., enhanced virtual presence, avatars, documentation of results (Parsons et al., 2009) justify the use of simulation for PM practice.
A comprehensive survey of PM training methods was provided in Kenyon et al. (2018) systematic review.
Although training on an actual PW is the most commonly used technique, other methods such as simulators (Arlati et al., 2020), smart robot wheelchairs (McGarry et al., 2012) and modified "ride on cars" have also been investigated (Huang et al., 2018).The importance of training in a natural environment is recognized (Butler et al., 1984;Jones et al., 2012).Jones et al. (2012) provided PW to 14 children to practice in their home environment for 12 months, supervised by their parents.Butler et al. (1984) provided a PW to 12 children to practice under parental guidance at home.In both studies, children were able to make progress in their driving skills in their natural environments.
Most studies investigating pediatric PW training have been carried out on very young children, with very few focusing on older children who have not yet mastered independent driving.Rosenberg et al. (2021) evaluated an intensive three-week PM summer camp for children and adolescents with cerebral palsy (CP); all 24 participants significantly improved their PM skills as assessed by the Powered Mobility Program (PMP) and Assessment of Learning Powered Mobility (ALP).
A scoping review examined different aspects of PM training with a simulator focusing on the sense of presence, user interaction and perception as well as on the effectiveness of this type of training (Arlati et al., 2020).All studies reported improvement in driving skills as measured in both simulator and realworld tasks, thus supporting the ability to transfer skills obtained via simulator practice.Increased training duration with the simulator improved driving skills (Arlati et al., 2020).
To date, simulator-based training has not been compared directly to PW training of children.The main objective of this study was to assess simulator-based PM training and to determine its effectiveness in comparison to PW training.The hypothesis was that children practicing on a chair or via a simulator at home would improve their PM skills significantly, with no significant differences between training methods.This was accomplished by comparing the pre-and posttraining PMP, Israel Ministry of Health's Powered Mobility Proficiency Test (PM-PT) and ALP outcomes in two equivalent groups of children (PW versus simulator).

Study design
This study used a pre-test/posttest pragmatic design where participants were recruited from a list of children waiting to be provided with a PW from the ALYN Hospital PM lending program.The waiting list at the start of the study (May 2018) was continually updated with additional children until the study end date (June 2020).Since the effectiveness of simulator practice was unproven, i.e., there was no assurance that it would lead to proficiency in wheelchair driving, the hospital's Institutional Review Board required that participants in the simulator group be assigned from the bottom half of the waiting list.In contrast, participants in the wheelchair group (training via an actual PW that was considered "standard service procedures") were assigned from the top half of the waiting list according to their place on the waiting list.Since there were no differences in age, gender, and other demographics of the participants in the two halves, this allocation scheme merely ensured that PW training was given on a "first in-first out" basis.Moreover, a commitment was made to the parents of all children who practiced on the simulator that practice with a PW would be offered upon completion of the study if proficiency was not obtained.All children and parents were provided with a verbal overview of the study prior to giving their informed consent to participate.

Participants
Thirty-six children (23 males, 13 females), aged 6-18 years (mean age: 10.58 yrs, SD: 3.58), with CP and other neuromuscular diseases were recruited through the ALYN Hospital PW lending program.All participants were able to control the chair with a joystick.Participants were excluded if they had any unstable medical conditions that prevented them from participating in PM training (e.g., uncontrolled epilepsy); or had visual acuity or field deficits that interfered with their ability to view and respond to stimuli presented on a screen or while driving.

Outcome measures and screening classifications
Demographic Questionnaire was completed by a parent and therapist in the community regarding the participant's diagnosis, medical and developmental history, functional level, hand dominance, mobility, transportation, home/school accessibility and prior experience with powered mobility.

Screening classifications
Level of Sitting Scale (Fife et al., 1991) uses an eight-point scale to rate how much support a child needs to sit independently.
Gross Motor Function Classification Scale Expanded and Revised (Rosenbaum et al., 2014) uses a five-point scale to rate the gross motor level of children with CP aged 2-18 years.
Manual Ability Classification Scale (Rosenbaum et al., 2014) uses a five-point scale to rate the child's ability to handle objects during daily activities.It can be used for children with CP aged 4-18 years.
Communication Function Classification Scale (Rosenbaum et al., 2014) uses a five-point scale to rate the everyday communication abilities with familiar and unfamiliar partners of children with CP, aged 2-18 years.
Brooke Upper Limb Function Scale (Brooke et al., 1981) is a six-point scale that describes the functional level of the upper extremities of children with Duchenne and other neuromuscular diseases.
Vignos Lower Limb Function Scale (Florence et al., 1984) is a 10-point scale that categorizes the walking abilities of children with Duchenne muscular dystrophy and other neuromuscular diseases.

Primary outcome measures of PM proficiency
Powered Mobility Program (Furumasu et al., 1996) is a threepart evaluation of up to 34 driving skills, scored on a 6-point scale, that measures a child's ability to drive a PW in and outdoors.Skills include the ability to drive through doors, enter a room, drive down a corridor, drive up and down an outside slope.
Assessment of Learning Powered Mobility Use (Gefen et al., 2020;Svensson & Nilsson, 2021) -The ALP defines eight phases within three stages of learning how to use a powered mobility device: Explore Function, Explore Sequence, Explore Performance.Categories of observation include: Interaction and Communication, Understanding of Tool Use, Activity and Movement, Expressions and Emotions, and Attention.Israel Ministry of Health PM Proficiency Test (Gefen et al., 2020) is a three-point scale (pass = 5, needs practice = 3, fail = 1) Hebrew -language test with seven items that assesses the user's PW skills.Skills include the ability to stop voluntarily when asked to, to drive around obstacles, drive through a doorway, and drive up to 25 meters.A person is considered a competent PM driver if they are able to pass each item in a consistent manner four out of five times.It is used by therapists to determine who is entitled to government funding of a PW.The evaluation was developed based on Butler et al. (1984) proposed skills.Median scores were calculated as the final score.
All three PM assessments are reliable and valid (Gefen et al., 2020).

Instruments
Powered Wheelchairs for the control group were provided through the ALYN Lending Program.Chairs were selected from a pool of units with key characteristics (size, tilt, recline) representing commonly used models (e.g.Invacare, Pride Mobility, Permobil).Each chair was configured optimally for the participant's use.
The McGill Immersive Wheelchair Simulator (MiWe) (Archambault et al., 2012;Bigras et al., 2019;Tao & Archambault, 2016) was developed with Unity to compare real power mobility movement to that of the simulator, as a clinical assessment and practice tool for adults and to examine upper extremity reaching during powered mobility.The MiWe runs on a Windows computer, with an operating system of at least Windows 7, with a graphics card of at least 1 GB of video memory and able to run DirectX 11.It displays on a standard screen.The user operates a joystick to interact with a series of six simulated environments (navigating an elevator or car ramp, driving in a supermarket or mall, entering a narrow room or crossing a street) via a first person, nonstereoscopic viewpoint.Pausing at a defined spot for several seconds (dwelling) activates an elevator button or used cashier services.Joystick actions, the simulated wheelchair trajectory, collisions and tasks duration are recorded.Each environment has three levels differing in the number of tasks, obstacles and the time allotted per task.The tasks incorporated elements from the Wheelchair Skills Test (Kirby et al., 2004).
The MiWe was modified for use at ALYN with children (Gefen et al., 2021); two routes were added for testing purposes, and a Hebrew language interface was provided.Photos and screenshots in Figure 1 show several example views of the routes.

Procedures
Participants in both groups were fitted with a PW and seating supports (if required) by a seating expert and access site for wheelchair activation.The positioning of the participants in the simulator group was documented to ensure that they would sit as similarly as possible during the pre-and post-intervention evaluations.Participants from both groups drove the chair throughout a spacious room for about 15 minutes before the initial PM evaluation.They then drove the chair for 30-60 minutes along one of the two pre-determined test routes, which entailed maneuvering it through hallways and doorways, driving to specified destinations, stopping upon request and driving up an incline.An experienced occupational therapist walked next to the child to ensure safety.The extent of provided help was recorded as part of the PMP evaluation.Driving along the route was videotaped by a parent or second therapist for later analysis.The wheelchair group then left with a loaner chair which they kept until the end of the study.The simulator group participants were provided with a laptop, joystick and the simulator program and their families were instructed how to use the simulator for practice at home or school.Both the wheelchair and simulator groups were instructed to practice at least four times a week for a minimum of 20 minutes.The wheelchair group was encouraged to practice using the chair inside and outside.The simulator group was encouraged to practice all tasks (indoor and outdoor environments).
Participants returned to ALYN for a post-training evaluation after 12 weeks.Both groups drove along a pre-determined route on a PW.Videotapes of each evaluation were later analyzed by raters using the PMP, ALP and PM-PT.Participants in either group who demonstrated proficiency were referred to the Ministry of Health for procurement of their own chair.Children who did not demonstrate proficiency were provided additional practice opportunities; these additional data are not included in the current analysis.

Data analysis
Following calculation of descriptive statistics, group background differences were tested by the Pearson Chi-square test and the Fisher exact test for the categorical variables and the Mann-Whitney test or independent samples t-tests for ordinal or interval/ratio variables (respectively).With regard to the outcome measures, practice effects overall and within group differences were determined using the Wilcoxon signed rank test; pre-and post-practice, between group differences were determined using the Mann-Whitney test.Effect size was calculated by converting the z statistic, produced by these tests, to Pearson's r (Field, 2009) and interpreted as follows: small (r = 0.10); medium (r = 0.30); large (r > 0.50) (Field, 2009).The minimal clinically important differences (MCID) were estimated for all three assessments (PMP, ALP and PM-PT) by comparing their SEM (SD/√n) to the mean of the pre-post difference.If the SEM was less than the mean pre-post change, then this change was considered to be clinically significant (Copay et al., 2007;Musselman, 2007;Rai et al., 2015).
All analyses were carried out using SPSS software, version 25; significance was set to α ≤ 0.05.When appropriate, a false discovery rate (FDR) correction was made for multiple comparisons which resulted in p values needing to be ≤ .02 in order to reach significance (Benjamini & Hochberg, 1995).

Results
The Consort diagram, shown in Figure 2, shows participant allocation to each group including number and reason for dropouts.Data from 30 participants were analyzed.

Participant characteristics
Table 1 summarizes the background characteristics of the participants and their baseline outcome scores.There were no statistically significant differences between the groups apart from practice location where a greater percentage of children in the wheelchair group practiced both at home and school.The simulator group was characterized by a greater proportion of children who can propel a manual wheelchair.Parents reported that their children in both groups practiced for at least the recommended amount of time.
Analysis of the PMP (Table 2, Figure 3a) revealed a significant effect for practice overall (p < .001)with a medium effect size (r = 0.47) and a significant effect for practice within each group separately (Simulator: p = .04;PM: p = .003)with medium and large effect sizes (r = 0.39, r = 0.52, respectively).The overall between-group effect was not significant (p = .95)and neither were the effects between the groups at each time-point (prepractice: p = .22;post-practice: p = .47).All significant effects for this outcome measure survived FDR correction for multiple comparisons, except the practice effect for the simulator group that fell above the critical value following the correction (p = .02).
Analysis of the ALP (Table 2, Figure 3b), revealed a nominal significant effect for practice overall (p = .02)with a medium effect size (r = 0.30) and a significant effect for practice within the powered mobility group only (Simulator: p = .33;PM: p = .03)with medium effect size (r = 0.39).However, the significant effects found for this outcome measure did not survive FDR correction for multiple comparisons since they fell above the critical values following the correction (practice effect overall: p = .008;practice effect PM: p = .02).The overall between-group effect was not significant (p = .38)and neither were the effects between the groups at each time-point (pre-practice: p = .22;post-practice: p = .67).
Analysis of the PM-PT (Table 2, Figure 3c) revealed a significant effect for practice overall (p < .001),with a medium effect size (r = 0.46) and a significant effect for practice within each group (Simulator: p = .01;PM: p = .007)with medium effect sizes (r = 0.45, r = 0.47, respectively).The overall between-group effect was not significant (Mann-Whitney test: p = .45)and neither were the effects between the groups at each time-point (pre-practice: p = .58;postpractice: p = .67).All significant effects for this outcome measure survived FDR correction for multiple comparisons.
The MCID was estimated for all three assessments (PMP, ALP and PM-PT) as indicated in the data analysis section.The PMP (mean = 0.63; SEM = 0.15) and ALP (mean = 0.52; SEM = 0.24) demonstrated significant MCIDs.The PM-PT did not reach significance (mean = 0.55; SEM = 0.68).

Discussion
Our goal was to determine whether simulator-based PM practice is a viable training option for children with mobility impairments.Participants as young as six years improved their actual PW driving skills over a 12-week program regardless of practice method.In previous studies of PM training, only younger children were assessed, the training period was 6-12 months, and only a single training method was tested (Jones et al., 2012;Kenyon et al., 2018).We note that a relatively short training period (12 weeks) was used in the current study since the goal was to compare the ability to progress in PM skills rather than ensuring driving proficiency (although this was an important secondary aim that was achieved by 27% of the wheelchair group and 30% of the simulator group).When studying proficiency, longer training periods have shown that home-based practice with parents being responsible for ensuring sufficient practice opportunities can promote proficiency (Gefen et al., 2019;Jones et al., 2012).
In two systematic reviews (Arlati et al., 2020;Kenyon et al., 2018), different PW training methods were evaluated for children with and without disabilities.The studies described the use of a PW, smart navigation, virtual reality programs and physical PW simulators with each investigating a single training method.The direct comparison of two raining methods, as carried out in the current study, provides, for the first time, evidence that confirms the feasibility and effectiveness of using a simulator for PW practice.
Each type of practice method has advantages and limitations that are important to recognize when considering their suitability as training options.For example, using a simulator can be safer, in contrast to the constant supervision a child needs when beginning to practice on a PW.The chair and simulator can be used simultaneously to increase and enhance practice time and opportunities, e.g., at school on a powered chair and via a simulator at home.If a child lives in an inaccessible home, a simulator can be used to expedite the practice process until the family moves to a more adapted home.Moreover, a simulator would help acclimatize users to PM in cases of potential apprehension when, for example, a young child is confronted with a large, powered chair, especially when it is moving, Our results showed that participants who used the MiWe-C improved their real wheelchair skills, which is consistent with previous research.For example, Morère et al. (2020) demonstrated improvement in outdoor driving of eight children with disabilities following practice with a simulator.Adelola et al. (2009) demonstrated skill transfer by two children with CP who practiced with a VR system.These results provide some evidence for the transfer of skills following practice with a simple PW simulation to driving a real PW.
Furthermore, simulator training may be effective for the training of complex PW driving skills.Indeed, crossing the street, avoiding potholes and other barriers, or driving in non-idyllic weather conditions are all challenging when a child practices on a powered wheelchair.Being able to master these skills in a more controlled environment via a simulator is important.Morère et al. (2020) assessed 12 children and young adults with CP who practiced outdoor driving skills on a simulator.Eleven of the 12 participants improved their indoor and outdoor driving skills and five of the proficient indoor drivers were able to pass their outdoor driving test, showing that a simulator can be used to master challenges they were not explicitly trained for.
Most simulators used in previous studies, as noted in Arlati et al. (2020) and Lam et al. (2018) systematic reviews, were laboratory-based, and appropriate primarily for research in comparison to the MiWe-C, used in the current study, which is a clinical or home-based tool that can be incorporated into a user's daily routine.Having a simulator that can be used for practice in the community rather than requiring specific laboratory conditions is a considerable advantage since it facilitates intensive practice leading to improve performance (Arlati et al., 2020).The MiWe-C also supports the use of home-based, caregiver directed rehabilitation which has been shown to be effective in different populations including stroke survivors (Guillén-Climent et al., 2021;Rozevink et al., 2021), spinal cord injury (Osuagwu et al., 2020), cystic fibrosis (Del Corral et al., 2018), and CP (Ferre et al., 2017).Similar to the protocol used in the current study, these programs were based on caregiver directed rehabilitation and promoted a family-centered approach.Home-based rehabilitation appears to increase and prolong therapy dosage after discharge and to take advantage of technology provided it is adapted to the home environment.
The results of the current study were reported to the Israel Ministry of Health who have now adopted a policy of fully funding powered wheelchairs for children who show proficiency in a real powered wheelchair following training solely on a simulator.In countries where proficiency is a criterion for funding a powered wheelchair, having additional, affordable practice options may help more children become proficient powered wheelchair drivers.Inter-rater reliability (r = 0.99, p < .01);intra-rater reliability (r = 0.99, p < .01)(Furumasu et al., 1996;Gefen et al., 2020).b Inter-rater reliability (K = 0.83-0.87)(Gefen et al., 2020;Svensson & Nilsson, 2021).c Inter-rater reliability ICC 2,1 mean = 0.89; Intra-rater ICC 2,1 mean = 0.93 (Gefen et al., 2020).severely restricted; hence only 36 suitable participants were tested and only 30 completed the full protocol; this small sample size is comparable to other recent PM studies (Butler et al., 1983;Jones et al., 2012;Mockler et al., 2017;Rosenberg et al., 2021).The current study included multiple comparisons and most of the significant results survived the FDR correction procedure.However, some comparisons did not survive this procedure, mainly for the ALP outcome measure.The ALP assesses "softer" attributes such as learning and understanding, in contrast to the PMP and PM-PT outcome measures that assess more "concrete" behavioral attributes that are easier to define and assess.These latter measures are less exposed to unexplained variance, as opposed to the ALP.The ALP results should thus be considered to be preliminary in the current study and should be replicated in future studies with larger sample sizes in order to overcome the possible unexplained variance.
We contend that these results contribute to the evidence supporting the use of home-based simulator practice to enhance PM proficiency.Nevertheless, generalization of the results should remain cautious.Finally, although parents reported that their children practiced more than the mandated time, direct documentation of practice time was not recorded.Future studies should address whether children with cognitive impairment could benefit from practicing on a PW simulator and if it could be used by children who only have scanner or switch access.Other future studies should be randomized controlled trials to demonstrate the efficacy of the virtual training more conclusively.

Conclusions
Powered mobility simulators provide children and adults practice opportunities to achieve proficiency that may not be available otherwise due to the cost of a powered chair for training, therapy time and safety concerns.

Figure 3 .
Figure 3. Differences between the median pre-and post-intervention scores of the (a) PMP, (b) ALP and (c) PM-PT for the simulator practice group (gray) and the powered mobility group (white).The plots represent the inter-quartile ranges.

Table 2 .
Pre-and post-practice differences. a