Effects of exergaming on cognition, lower limb functional coordination, and stepping time in people with multiple sclerosis: a randomized controlled trial

Abstract Purpose To compare the effects of exergaming versus conventional exercises on cognition, lower-limb functional coordination, and stepping time in people with multiple sclerosis (PwMS). Methods Thirty-six PwMS were randomly assigned to either intervention (n = 18) or control (n = 18) group and received 18 training sessions during six weeks. The intervention group performed exergames that required multidirectional timed-stepping, weight-shifting, and walking while the control group performed conventional matched exercises. Trail making test (TMT part A, B; TMT-A, TMT-B, TMT B-A), six-spot step test (SSST), and choice stepping reaction time (CSRT-including reaction time (RT), movement time (MVT), and total response time (TRT)) were assessed pre- and post-intervention (short-term), and after three-month follow-up (mid-term). Results The intervention group showed faster TMT-B (p = 0.003) and TMT B-A (p = 0.002) at post-intervention and faster SSST at both post-intervention (p = 0.002) and follow-up (p = 0.04). The CSRT components showed no between-group differences at post-intervention; however, at follow-up, the intervention group had lower TRT (p = 0.046) and MVT (p = 0.015). TMT-A and RT had no significant between-group differences. Conclusions In short-term, exergames led to more improvements in complex attention, executive function, and lower-limb functional coordination comparing to the matched conventional exercises. In mid-term, exergaming was more effective for improving stepping time and lower-limb functional coordination. However, the two approaches did not show any superiority over each other for improving simple attention and RT. Implications for rehabilitation When designed properly, exergames have great potential to improve attention and executive function of people with multiple sclerosis (PwMS), at least in the short-term. Exergames seem like an appropriate option for improving lower limb coordination and decreasing choice stepping response time among PwMS in the mid-term. Exergames do not have superiority in improving the choice stepping reaction time compared to their matched conventional treatment.


Introduction
Balance and gait impairments, poor cognitive performance, deteriorated coordination in lower extremities, and reduced stepping ability are significantly associated with increased risk of falls in people with multiple sclerosis (PwMS) [1][2][3][4][5]. Hence, the multifactorial nature of falling in PwMS rises the need for designing specific interventions targeting as much risk factors as possible [5].
In recent years, the feasibility and effectiveness of the exercise gamification (also known as exergames) using virtual reality (VR) technology have been increasingly investigated [6,7]. Studies have supported the idea that VR-based neurorehabilitation positively affects neurological rehabilitation outcomes by augmenting motivation and participation [8]. By using VR technology, therapists can easily implement different types of exercises and adjust training attributes and levels of difficulty [7-10]. More specifically, previous studies in PwMS have revealed the promising effects of VR-based (compared to conventional) exercises on improving balance (as the most important risk factor of falls) and cognitivemotor functions [8, 11,12]. In cognitive demanding situations (like the most everyday activities), both cognitive capacity and cognitive-motor interaction are key factors to maintain the function [10,11]. As such, improving cognitive function would be an advantage for any balance and mobility rehabilitation intervention [10,11]. Cognitive impairment, as one of the important risk factors of falls and disability in PwMS, represents at all stages and subtypes of this disease and has devastating effects on the quality of life [13,14]. Impaired attention and executive function are among the most common cognitive deficits in PwMS [13]. Most involved subcomponents of attention are sustained, selective, and divided attention (complex attention in total) which prevent PwMS from performing many daily tasks like following a conversation, doing something carefully at work, resuming to previous task after an interruption, or keeping attention on a specific issue in the presence of other external stimuli [13]. Inhibitory control (ability to suppress irrelevant reactions to get the task done successfully), working memory (WM) (ability to hold information in the brain and work on it), and cognitive flexibility (ability to manage new environmental conditions and new tasks) are three cores of executive function that whenever work properly, the higher-order executive functions including anticipating, planning, goal setting, error correction, reasoning (self-regulation), and problem solving will be done successfully [15]. Eventually, a proper executive function leads to acceptable, effective, and creative social behavior [15]. Based on a recent multi-level meta-analysis, the exerciseinduced effects on cognition among PwMS are conflicting and the conventional exercises have relatively failed to improve cognitive domains in PwMS [16]. As such according to the recommendations of previous studies [10,16,17], we aimed to determine whether exergaming can be effective.
Proper cognition and motor control are always needed to have inter-limb coordination [18]. In PwMS, there is also evidence of strong correlations between both cognitive and walking impairments, with inter-limb coordination [18]. A previous study showed that, compared to no-intervention, balance and motor control training improved lower limb functional coordination in PwMS [19]. Since exergames have the potential to challenge coordination by triggering its main components including sequencing, timing, and accuracy [20][21][22], we investigated the effects of exergaming on improving lower limb functional coordination in PwMS. If the exergaming improves lower-limb functional coordination more than or even the same as the conventional exercises, it can be implemented as a more comprehensive intervention thanks to its extra features like placing individuals in cognitively demanding environments and using multisensory feedbacks (visual, auditory).
Appropriate inter-limb coordination plays a crucial role in walking and stepping [18]. According to previous studies, the inability to step in response to perturbations is one of the risk factors of falls in PwMS [2]. For fall prevention, a person should be able to take steps at proper time and with an acceptable speed [23,24]. In a previous study, video gaming enhanced stepping ability, but no comparison was made with conventional exercises to determine the potential superiority of the two approaches [3]. Given that VR has the potential to implement reactive exercises [10], another goal of the present study was to investigate the effect of stepping exercises through exergaming, that inherently trigger the response time, on stepping time and compare the results with conventional exercises that have not time constraints.
Totally, this study aimed to assess and compare the effects of exergaming versus conventional exercises on improving cognitive function, lower-limb functional coordination, and stepping time in PwMS. We hypothesized that exergaming used in the present study would be more effective than the conventional exercises. The results of this study can guide clinicians to apply more effective implementations of the exercises to reduce the identified impairments.

Study design
This study was a two-arm, parallel, single-blinded randomized controlled trial, designed according to the CONSORT guideline [25]. The trial was conducted at the Musculoskeletal Rehabilitation Research Center located in Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran between May 2018 and May 2019. The study was approved by the University Ethics Committee (code: IR.AJUMS.REC.1396.1114) and registered in the Iranian Registry of Clinical Trials (registration ID: IRCT20171106037286N2).

Participants
People with a definitive neurologist diagnosis of relapsing-remitting or secondary-progressive MS [26,27], were recruited from the Khuzestan MS Patients' Society by phone or during monthly visits. Inclusion criteria were 18-64 years of age and Expanded Disability Status Scale (EDSS) higher than 2 and lower than 6. Exclusion criteria were MS relapse in the past three months; apparent cognitive impairment (mini-mental state examination below 24); uncorrected visual or auditory impairments; and pregnancy. All participants signed a written informed consent form before the study.

Randomization
Following the accomplishment of the baseline assessment, participants were randomly assigned to the intervention (exergaming) or control (conventional exercise) group using a computerized random allocation sequence with different block sizes. To conceal the allocation sequence, sealed envelopes were used. Randomization was performed by an investigator not involved in any other parts of the study.

Intervention
The participants in both groups received 18 training sessions, three times per week for six consecutive weeks, by a physical therapist who was blinded to the assessment results. The intervention group performed exercises using the Xbox360 with Microsoft's Kinect V R three exergames which required multidirectional timed stepping, weight-shifting, and walking in different directions, while the control group performed matched (in terms of type, form, and duration) conventional exercises. Progression of exercises was individualized for each participant. To avoid fatigue, appropriate rest periods according to the participant's needs were provided between the exercises. The net treatment duration in each session was 35 min for the both groups. The duration of a 6week period was selected as a recent meta-analysis reported that 6 weeks of exergaming (three 30-min sessions per week) is the median dosage based on previous studies [28]. Detailed information about the exergames and conventional exercises, progression of the exercises, and handling of fatigue are provided in the Supplementary file.

Outcome measurements
All participants underwent three assessments at baseline, postintervention (short-term), and after three-month follow-up (mid-term). Demographic data were recorded on the day of the baseline assessment. Assessments were conducted by a trained physical therapist blinded to the group allocation. The assessor remained blinded until the end of the follow-up assessment. Preintervention tests had random order of administration to prevent the effects of fatigue on the results. The same orders were utilized at post-intervention and follow-up. In addition, participants were allowed to have adequate rest (2-5 min) between the measurements. After the post-intervention assessment, participants were not received any specific exercise advice and were asked to continue their usual care until the follow-up assessment.
A set of outcome measures were utilized to evaluate attention and executive function (trail making test (TMT)), lower limb functional coordination (six-spot step test (SSST)), and components of choice stepping reaction time (CSRT test including total response time (TRT), reaction time (RT), and movement time (MVT)). TMT is a measure of cognitive processing which is one of the most widely used measures in neuropsychological research and clinical settings [29]. TMT includes two parts: part A (TMT-A) that assesses simple attention, visual scanning and information processing speed (PS), and part B (TMT-B) that additionally assesses complex attention, WM, inhibitory control, and set-switching abilities [2, [29][30][31]. TMT-A requires the participants to draw lines to link numbered circles in a numerical order (i.e., 1-2-3, etc.). In TMT-B, the participants have to connect the circles contained numbers and letters in an alternating numeric and alphabetic sequence (e.g., 1-A-2-B). Both parts asked to be completed as fast as possible. Total time to accomplish each test was measured. The difference between TMT-B and TMT-A execution time (TMT B-A) was calculated to estimate the executive function [2, 29,31,32]. TMT is highly sensitive to the variety of neurological conditions, also to detect pre-clinical manifestations of cognitive decline such as in Alzheimer's [29]. Also, it is significantly correlated with the instrumental activities of daily living in the elderly and neurological conditions [29]. This test is able to differentiate between PwMS and healthy controls, and its validity to evaluate executive function and attention deficits in PwMS (if there is an impairment in part B while part A is normal) has been confirmed [33,34].
Six-spot step test was performed in a 1 � 5 m (width � length) field, with plastic cones placed at the distance of 1 and 3 m from the starting line at one side, and 2 and 4 m at the other side; also, at the center of starting and finishing lines. Participants were asked to walk as quickly as possible, while kicking the cones out of the field with specified foot, alternating between the medial and lateral sides of the foot. Each participant performed two trials for each foot, and the average time of the four trials was calculated as the final point [35,36]. The SSST was selected to assess lower limb functional coordination. This test challenges the coordination and dynamic balance, which can better evaluate the complex functions of the lower limbs in walking compared to measures assessing straight walking [36]. The SSST is a lower limb equivalent of the Nine-Hole Peg Test (NHPT) in upper extremity to evaluate fine movements [35]. The SSST has acceptable reliability and validity in PwMS [36][37][38].
The choice RT measures the individual's ability to maintain attention and make/inhibit motor responses to the target/non-target stimulus [39]. In this study, CSRT was used, which measures an individual's ability to step quickly and properly in response to visual stimulus [40]. The participants stood on a long force plate (EquiTest CRS V R NeuroCom V R International, Clackamas, OR), with feet hip-width apart, while looking at the screen in front of them. Highlighted boxes were randomly appeared in right or left side of the screen and participants were instructed to step with the corresponding foot as fast as possible and then return to the starting position. Each participant performed 10 trials, each contained five random steps. Times from stimulus appearance to movement initiation (lift off), and from movement initiation to foot strike were measured as RT and MVT, respectively. The TRT was obtained as the sum of RT and MVT [3].

Statistical analysis
Data were analyzed using IBM SPSS statistics software version 22 for Windows (SPSS Inc., Chicago, IL) and the p value was set at 0.05. All variables were assessed for normality using Kolmogorov-Smirnov's test and were normalized via log transformation when appropriate (applied for TMT-A, TMT-B, and TMT B-A; in which the normality was confirmed after the log transformation for all of them). An intention-to-treat analysis based on the last observation carried forward approach was performed for the three participants who discontinued the intervention and two participants who lost the follow-up. The clinical and demographic characteristics of the participants in each group were compared at baseline using independent sample t-test, Mann-Whitney's U test, or Chi-Squared test. Repeated measures analysis of variance (ANOVA) was used to determine within-group differences during repeated measurements. To assess between-group differences at post-intervention and follow-up, the analysis of covariance (ANCOVA) adjusting for baseline values of the outcome measures was conducted. In addition, Cohen's d coefficient was calculated to determine the effect sizes (ESs) for changes within and between the groups [41].

Sample size
Since SSST has a confirmed cut-off to score real change for therapeutic purposes (greater than 19% of change), we used this measure to determine sample size. Average SSST value for the first 10 participants in the study (pre-intervention SSST was 15.35 s). Since our goal was to improve this time by at least 3.07 s (based on a change of 20% from 15.35 s), we achieved a 0.9 ES based on the previous study's standard deviation (3.45 s) [36]. By setting the ES to 0.9, power to 0.8, and the alpha to 0.05 using the GPower software (version 3.1), 16 participants were needed in each group. Figure 1 demonstrates the CONSORT flow diagram. A total of 36 PwMS participated in this study who were assigned into intervention (n ¼ 18) and control (n ¼ 18) groups. One participant of the intervention group (transport problems) and two of the control group (lack of interest and work schedule) discontinued the intervention. As indicated in Table 1, at baseline there were no significant statistical differences in demographic and clinical characteristics of participants between the groups. No significant adverse events were reported in either group after the interventions. Adherence to exergames and conventional exercises were 97% (mean, 17.4 of 18 sessions; min, 8; max, 18) and 93% (mean, 16.7 of 18 sessions; min, 5; max, 18), respectively.

Between-group differences
Results of ANCOVAs are presented in Table 2

Within-group differences
Results of repeated measure ANOVA (Table 2) showed significant  improvements of TMT-A, TMT-B, TMT B-A, SSST, TRT, and MVT in intervention group at both post-intervention and follow-up. In control group, TMT-A at follow-up revealed significant withingroup change. In both groups, RT did not show significant improvement neither at post-intervention nor at follow-up.

Discussion
The results of this study showed that in the short term, six weeks of exergaming leads to more improvements in complex attention, executive function, and lower limb functional coordination compared to conventional exercises. In the mid-term (three months), lower limb functional coordination, stepping response time, and stepping MVT were significantly different favoring the exergaming group. The two groups were not different in simple attention, and it improved in both groups in mid-term. None of the interventions reduced the RT, so future studies are needed to focus on the design of protocols specifically targeting this outcome. Although both therapeutic approaches were able to improve simple attention (constant attention to perform a simple task) as a prerequisite for all higher-level mental activities [42], the complex attention (i.e., ability to control, shift, and divide attentional resources) as one of the crucial elements in the quality of life of PwMS [42] showed greater improvement in the exergaming group. The VR-based training in this study has been designed to challenge the elements of complex attention and executive function. These challenges include continuous attention and focus on the game environment (sustain and selective attention), simultaneous attention to several elements for the successful completion of the tasks (divided attention), adjusting various reactions and trying to avoid the irrelevant ones (inhibitory control), trying to perform the required tasks in various environments in different games with different degrees of difficulty and task demands, and learning the processes to succeed in the more advance stages of the games (WM and cognitive flexibility). These features are not generally considered in the control group due to the non-disturbing nature of conventional exercises (fixed environment, without a specific time limit and without any real time feedback/rewards, etc.). In the other words, the virtual environment designed for this study enabled us to expose the participants to more attentional demands than the conventional exercises. Contrary to our results, Hoang et al. showed no significant differences in attention and executive function following step training using video-gaming in PwMS [3]. This finding may be related to the type of implemented training program that only included one kind of stepping exercise in a fixed environment. In the present study, on the other hand, use of three different exergames with different levels of difficulty, might be attentionally more challenging for the exergaming group. Similarly, the results of a recent RCT in older adults with mild cognitive impairment indicated that VR training led to more improvement in executive function (TMT-B) than conventional training [43]. In addition, previous studies in PwMS have shown that the exergaming in virtual environment challenges cognitivemotor skills and leads to significant improvement in these functions [11,12,44]. It is worth to note that significant post-intervention between-group differences were not observed in the followup assessment. Comparing with pre-intervention, performance of the exergaming group in TMT-B and TMT-B-A was better in both post-intervention and follow-up, while there were no changes for the control group. Taking a closer look to the results (Table 2), compared to post-intervention, the standard deviation of the scores in the control group was much higher in the follow-up. The unchanged mean alongside with the high standard deviation supports the idea that some individuals in the control group performed better and some performed worse in the follow-up. This might be the reason of disappearance of the post-intervention between-group differences at the follow-up.
The SSST considers lower limb coordination in addition to walking speed which are two important risk factors for falling in PwMS [35]. According to a recent RCT in PwMS, conventional progressive balance exercises (compared to no intervention) significantly improved SSST [19]. Findings of our study support the idea of superiority of exergames over the conventional exercises on improving lower-limb functional coordination. It has been showed that for therapeutic purposes, changes more than 19% in the SSST values would be real and satisfactory in PwMS [38]. In the present study, the SSST changes at both post-intervention and follow-up were above this value only in the exergaming group ( Table 2). Exergames of this study required the participants to put their feet with the maximum speed in the specified places in order to earn points/rewards. While, the speed of the exercises and the exact location of the feet did not matter in the conventional exercises. As such, the probable cause of observed difference in the SSST, where fast and correct hitting of obstacles is necessary to perform the test acceptably, is the demand for more controlled movements in the exergaming group. Since exergames compared to their conventional counterparts, presented significant improvement in SSST in both short-term and mid-term, also due to having the positive features of VR medium that were mentioned throughout the text, clinicians can use exergames for offering a more comprehensive rehabilitation to PwMS who have complex walking impairments such as inter-limb coordination deficits.
We did not observe significant differences between the two exercise approaches in CSRT results post-intervention. After three months, the stepping TRT and stepping MVT in the exergaming group were significantly lower than the control group. Based on the within-group results (Table 2), in the follow-up period, the control group returned toward the pre-intervention mean values (for TRT) and even worsened it (for MVT), but the exergaming group was able to preserve the improvement, which led to a statistically significant between-group differences at follow-up. The changes of 2-10% of the MVT and 3-12% of the response time are usually considered as clinical significance [3,44]. In the present study, the mid-term changes (changes of mean values from preintervention to follow-up) for exergaming group were 10% for the MVT and 8% for the response time that could be considered as clinically significant.
Our results show that the faster response in the exergaming group, compared to the control group, was due to the greater improvement in stepping MVT, not stepping RT. RT includes three components of sensory input transmission, central information processing (CIP), and motor response, which 80% of that assigns to the CIP component [45]. Among the CIP subcomponents, the two that are most focused on the assessments of PwMS are WM and PS [46]. The PS is more impaired than WM in PwMS, and this disrupts whole CIP [46]. TMT-A is also used to evaluate PS [29,34,47], and we did not observe any statistically significant between-group difference in this measure neither at post-intervention nor at follow-up. It can be concluded that both exergames and conventional exercises were able to improve PS (Table  2; pre-intervention to follow-up for TMT-A in both groups) and the two approaches have no superiority over each other on improving PS, so CIP and ultimately RT. The reason for the shorter MVT in the exergaming group may be due to the nature of exercises in the virtual environment. As mentioned above, the participants in the exergaming group needed to do the movements at maximum speed within a certain time frame to earn points/ rewards while the conventional exercises were performed at preferred speed with no time constrains. Hoang et al. showed that performing a 12-week specific timed stepping exercises with VR in PwMS, compared to no intervention, resulted in a significant improvement in all components of the CSRT test including stepping RT [3]. Since the implemented video-game in the study by Hoang et al. was quite similar to the CSRT test procedure, the specificity of training principal might be the reason for observing significant improvement in the stepping RT. Also, the duration of intervention was different from our study. In the Hoang et al. study, participants were asked to practice at least two sessions per week for 12 weeks [3]. Although the exact number of training sessions was not reported, their participants probably had greater amount of exercise. Further studies on the exergame protocols targeting the improvement of the RT are needed to address the issue.
We acknowledge some limitations. First, the present study had relatively narrowed samples. Future studies with larger sample size are needed to investigate generalizability of our findings to a broader sample of PwMS. Second, for executing the TMT test, we did not measure dexterity directly. However, the participants were asked to write their names and test date on the TMT sheets to make sure that they had no dexterity limitation. Third, we examined clinical and functional lower limb coordination. Future studies should be conducted to use laboratory equipment (e.g., motion capture systems) to assess kinematic coordination to examine the effects of exercises on intra-limb and inter-limb coordination. Fourth, the exercises in both groups were completely matched in terms of form and duration, but not using a scale such as Borg to ensure the intensity matching of the exercises can be another limitation of this study. Fifth, the cognitive status of the participants in the present study was measured globally by MMSE, and none of the referred patients scored lower than the designated cut-off and did not exclude due to the MMSE score. This may have led to a ceiling effect in our cognitive outcomes. We recommend that future studies use domain specific measures to assess cognitive status and include cognitively impaired individuals to increase generalizability of the results. Sixth, the wide range of participants' age (18-64) may have led to some age-related cognitive changes whereby older people may need more training dosage than younger individuals to achieve cognitive improvements. Seventh, in the present study, considering the cost and logistical issues, follow-up duration was 3-month. We suggest that future studies evaluate the retention of the improvements over longer follow-up periods. Also, longitudinal studies investigating the relationship between the improvement of the studied risk factors and future falls rate could be very beneficial.

Conclusions
Compared to matched conventional exercises, exergaming in virtual environments forms a mid-term improvement in lower limb functional coordination, stepping TRT, and stepping MVT. Also, exergames can improve complex attention and executive function at least in the short-term. However, exergames did not have any superiority in improving simple attention and stepping RT over their conventional counterparts. Further studies on broader community of PwMS that include specific exercises targeting stepping RT, and assess the rate of falls are suggested.