The Effect of Virtual Reality Intervention Programs on the Functionality of Children and Adolescents with Cerebral Palsy. A Systematic Review

Background: Cerebral Palsy (CP) is a neurological disorder of movement and posture. Recent studies have shown that Virtual Reality (VR) is a useful and low-cost tool used in treating children and adolescents with cerebral palsy. Nevertheless, there is no substantial evidence supporting that VR therapy can help CP patients, not only as the primary treatment, but as a supplement. Objectives: The present systematic review aimed to investigate the effectiveness of VR intervention programs on the functional capacity of children and adolescents with CP, according to the International Classification of Functioning, Disability and Health (ICF). Methods: A systematic online search was conducted in PubMed, Scopus, and PEDro databases, as well as in the Google Scholar search engine, from inception till September 2022. The methodological quality of included studies was rated with the PEDro scale. Results: Twenty-two randomized-controlled trials were eligible for inclusion. The results indicated that there was a significant improvement after the implementation, of interventional VR programs, in balance and visual perception, while the results were controversial for muscle strength, coordination, gross motor function, gait, upper limb function, independence in activities of Daily Life Activities and participation. Conclusion: Significant balance and visual perception improvements may result from VR programs applied in children and adolescents with CP. Important factors that may influence the results are the functional level of the participants, the sample size, the context in which the therapeutic intervention is carried out (rehabilitation center, home), and the conventional treatments that the VR intervention programs are compared against.


INTRODUCTION
Cerebral Palsy (CP) is a group of neurological permanent disorders of movement and posture, caused by non-progressive interference in the developing brain (Yu et al., 2018). Also, CP is a socio-economic problem, since its associated conditions can impose a significant economic burden on the affected families, health care system, and general economy, as it requires long term supportive care services. Generally, the cost of cerebral palsy is estimated expenditure of $1.47 billion per year (Australian Cerebral Palsy Register Report, 2018). Cerebral Palsy has "varying severity and complexity" across the lifespan, so its management requires a multidisciplinary approach (Trabacca et al., 2016), focusing on "maximizing individual function, choice and independence", in line with the International Classification of Functioning, Disability and Health (National Guideline Alliance (UK), 2017). Modern CP rehabilitation is based on motor 12 IJKSS 11(2):11-24 CP may not be able to participate in interactive activities, in a safe environment and this is done in real time, through VR (Chiu et al., 2014;Harris & Reid., 2005;Park et al.,2021, Rostami et al., 2012. In recent years, several systematic reviews have been published, that evaluate the effectiveness of VR rehabilitation programs in CP. Ravi et al. (2016) included 31 studies with various research designs and reached conflicting conclusions. In particular, they reported that the intervention of VR programs could improve balance and mobility while it did not significantly affect the function of the upper limbs (Ravi et al., 2017). The systematic review and meta-analysis by Chen et al. (2018) included 19 randomized studies, reporting positive effects of VR on arm function, gait and balance in children and adolescents with CP . In addition, Rathinam et al. (2019) assessed the effectiveness of VR in children with CP but the results reported from 6 RCTs only for hand function were conflicting (Rathinam et al., 2019). Similarly, Fandim et al. (2020) included 23 randomized trials and observed short-term benefits in upper and lower limb function and balance, after using VR, as an adjunctive therapy (Fandmin et al., 2020). Similar conclusions were reached by Alrashidi et al. (2021), who supported that VR was ineffective in the arm function of children with CP.
All these studies point out that the use of different assessment tools, the heterogeneity of the results obtained, and the difference in the methodological quality of the studies precluded any generalization of the results and the production of valid conclusions. The present systematic review aimed to assess the effectiveness of VR intervention programs on the functional capacity of children and adolescents with CP based on the classification of their functions in the framework of the ICF. More specifically, the effects of VR training on physical structures and functions, activities, and participation of children and adolescents with CP, were evaluated according to the variables examined in the selected literature.

Research Design
This systematic review has been conducted following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Page et al, 2021).

Eligibility Criteria
Initially, the eligibility criteria of the studies were according to the research question, which was formulated according to the principles of the PICO method. PICO is an acronym for the words: Problem/Population, Intervention, Comparison, Outcome, and is used to formulate the clinical question, in literature reviews and to specify keywords. These four words are the basic elements for searching the research question according to EBP (Evidence-Based Practice) (Mamédioda Costa Santos et al 2007). Also, studies published in full text in English, randomized controlled trials (RCTs), were included, in which, the participants were diagnosed with cerebral palsy aged between 5-18 years old. Additionally, VR training had to be the main treatment of the experimental group, without surgery performed before or during the intervention. Alongside, VR training had to be used in both groups and the intervention group not to be compared with typically developed children.

Search Strategy and Screening
A systematic search of the PubMed, Scopus, and PEDro databases, as well as the Google Scholar search engine was conducted from inception to September 2022. The keywords used were: cerebral palsy, children, adolescents, virtual reality, video games, exergaming, walking, gait, balance, fine motor skills, gross motor skills, participation, activities, which arose from the analysis of the exploratory question, Also, the review was conducted using the controlled vocabulary of pre-defined terms [Medical Subject Headings (MeSH) terms] wherever possible. The entire articles were studied without a time limit. Two independent researchers studied and evaluated all studies initially selected from the electronic searches. Duplicates were removed manually. Then, several articles were rejected after reading their title or studying their abstracts. Finally, in case of doubt, the entire article was examined to decide whether to include it in the review, according to the inclusion and exclusion criteria already set.

Data Extraction
Data relating to the included population characteristics (age, CP type and classification), the VR intervention characteristics (type, frequency, duration), as well as the ICF-based outcomes each study utilized were extracted from included studies.

Data Synthesis and Analysis
Results were presented and analyzed according to an ICFbased categorization of the outcomes. Since clinical heterogeneity was present between-studies, a narrative synthesis was performed. Specifically, the combination of quantitative results between studies was performed according to the vote-counting method based on the direction of effect reported from each study, as described by McKenzie & Brennan (2022).

Assessment of Methodological Quality
The methodological quality of the studies was carried out with the PEDro scale, which has high enough validity and reliability (Maher et al., 2003). It consists of 11 criteria, 10 related to the internal and external validity of the results and are summarized for the overall score (Verhagen et al, 1998). Each criterion is scored 1 point, with a maximum score of 10 and a minimum of 0. From zero to three points, studies are rated as "low quality", from four to six as "moderate", and from seven to ten as "high" (Cahin & McAuley, 2020).
The International Classification of Functioning, Disability and Health (ICF) can be used during the evaluation, the forma- tion of therapeutic goals and the selection of the intervention (Martinuzzi et al., 2010). The ICF consists of 2 parts: the 1 st part includes the parameters-domains of functioning and disability, and the 2 nd part includes the parameters-domains of the context factors (environmental and individual factors) (Rosenbaum & Stewart, 2004;Palisano et al., 2006;WHO, 2001).

Search Results
The electronic search of the databases amassed a total of 523 articles. From these, 91 duplicates were excluded. According to the procedure, from the remaining 432 articles, 383 were rejected, after the assessment based on the title or the summary. Specifically, 163 articles had no relevant content, 34 were not in English, while 186 were not RCT. The evaluation of the full text concerned 49 articles, of which, 27 were considered inappropriate for our review. Particularly, 5 articles included population outside the specified age range, 7 combined VR with specialized treatment, 3 included surgery population and 3 compared the sample to a typically growing population. One survey had not been published in all of its extent, while 6, included a VR program in both groups. Finally, the articles included in our systematic review were 22 (Figure 1).

Assessment of Methodological Quality
The evaluation of the methodological quality of the reviews was carried out with the PEDro scale, and all studies were rated separately by two different assessors. Specifically, 3 studies were of low quality (2-3 points), 10 moderate (4-6 points) and 9 high (7-9 points). The average PEDro score for the randomized studies, included in our systematic review was 4.5. The final score of the studies was formed after discussion by the assessors and are presented in Table 1.

Characteristics of the Participants
The present review incorporated 22 studies that included 728 patients in total. Of the patients initially recruited, 689 completed their participation in the study. All patients had been diagnosed with CP, aged 5-18 years. In each study the CP type varied among the patients. The functional capacity of those patients was classified according to the five levels of the GMFCS (Gross Motor Function Classification System) and MACS (Manual Ability Classification System). Furthermore, all studies included in their research protocol participants of both sexes ( Table 2).

Duration of the VR Programs
The duration of the VR interventional programs ranged from 4 weeks to 5 months. In most papers, the frequency of the intervention was 2 or 3 times per week. As for the exercise duration, it varied from 20 to 145 minutes ( Table 3).

Dependent Variables and Assessment Tools
Dependent variables, as well as the assessment tools that were used were categorized according to the domains of ICF. Specifically, categorization referred to: a) body structures and functions, b) activity and c) participation. The results are presented in Table 4.

Effectiveness of Virtual Reality Intervention Programs on Structure and Function
Initially, the effect of VR programs on CP patients' strength, was evaluated in 4 papers (Avcil et al., 2020;    Chen et al., (2012) found positive changes in the measurements, after the home-based Virtual Cycling Training (hVCT) intervention. Particularly, there was a statistically significant difference in the measurements of both angular velocities in the knee flexors (60 o /s: p=0.028; 120 o /s: p=0.003) and in the knee extensors (60 o /s: p=0.045; 120 o /s: p=0.003). Also, El-Shamy & El-Banna, (2018), found increase in the strength of all participants, in both hands, while the interventional group had better performance. In contrast, Chiu et al., (2014) noted no statistically significant difference, between groups of the upper limb muscles. However, participants in the intervention group had more strength compared to participants in the control group (p=0.10 in 6 weeks and p=0.19 at 12 weeks). Similarly, Avcil et al., (2020), reported no statistically significant changes between the groups (p>0, 05), although there was a change in both examined handles, with a statistically significant difference in each group (p<0.05).

Voluntary control and coordination
Voluntary control was evaluated only by Jung et al., (2020). Researchers, after the intervention, found signif-icant improvements in all variables (ankle dorsiflexion (p=0.008), knee extension (p=0.008), left hip abduction (p=0.032)] except right hip abduction relative to the control group (p=0.151). Additionally, Al Saif & Alsenany, (2015) observed significant changes in the intervention group compared to the control group in eye-hand coordination. Specifically, on the BOTMP scale, the intervention group improved from 2.23 ± 0.47 to 3.78 ± 0.39, while the control group improved from 2.82 ± 0.51 to 3.12 ± 0.66. In contrast, Chiu et al., (2014) observed no change between groups. Specifically, no statistically significant difference emerged, either at 6 or 12 weeks, when the measurements were made in the elbow joint (6 weeks: p=0.30, 12 weeks: p=0.15) and in the finger joints (6 weeks: p=0.54, 12 weeks: p=0.92). James et al., (2015), assessed visual perception and there was a statistically significant improvement in the intervention group, compared to the control group. The final TVPS-3 score had a difference of 6.79 (p=0.001), while its subscales, also showed a statistically significant difference [visual discrimination (p=0.017), spatial relations (p=0.01) and visual loss (p=0.03)].

Balance
The effect of VR programs on balance was studied in 11 papers ( Specifically, using Balancia program and mFRT, there was a statistically significant difference (p<0.05), while with K-TCMS there wasn't (p=0.102). Similarly, Wade & Porter (2012) evaluating sitting balance observed a statistically significant improvement in 2 of the 8, Chaily level evaluation items (p<0.05), as well as, in the final SACND score (p<0.05).
In contrast, Ramstrand & Lygnegård (2012) observed no significant changes in the intervention group (p>0.05). Also, Pin & Butler (2019) after their intervention, did not find any changes in the experimental group, as did Arnoni et al (2019), (p=0.085). Finally, Jha et al (2021) found improvement in both groups after their assessment with the PBS and Kids-Mini-BESTest scales. Statistical significance found only in the second scale (p=0.001), while in the first p =0.06.  Saif et al., (2015) found positive effects on walking. Also, Jung et al., (2020) found an improvement in the intervention group compared to the control group, but it was not statistically significant (p>0.05).

Gross mobility
In the present review 5 papers assessed gross mobility ( observed significant improvements in sections D (standing) (p=0.021) and E (Walking, Running and Jumping) (p=0.008) of the gross mobility rating scale (GMFM). Also, Sahin et al., (2019), reported significantly better gross motor performance in the group that followed the interventional VR program compared to the control group (p=0.001).

Arm function
Upper Limbs function concerns movements of the shoulder, elbow and fingers (fine motility) and were evaluated in 11 studies ( On the other hand, Avcil et al., (2020) reported similar results in arm function between the intervention group compared to the control group. Similarly, Chiu et al., (2014), found no improvement comparing the results of the two groups using the Nine-hole Peg Test (6 weeks: p=0.91; 12 weeks: p=0.34) or using the Jebsen -Taylor Test of Hand Function (6 weeks: p=0.89; 12 weeks: p=0.46). Wang et al., (2021) and Reid & Campbell, (2006) did not observe statistically significant changes in the fine motor skills of the experimental group compared to the control group. Additionally, James et al., (2015) reported that a web based therapy program was not superior compared to usual care (consultative sessions with medical and allied health professionals) regarding arm function, of children with unilateral cerebral palsy. Finally, Jannink et al., (2008) found little difference in only two, of the ten children (9% and 13%) who participated in the experimental group.

DISCUSSION
In the present systematic review several variables were evaluated, in order to investigate the effect of VR programs to children and adolescents with CP. The results showed that some variables had a positive impact, and others had no effect.
In particular, El-Shamy & El-Banna, (2018) and Chen et al., (2012) report a positive effect of VR programs on participants' muscle strength, upper limb and lower limb, respectively. In contrast, Avcil et al., (2020) and Chiu et al., (2014) found no statistically significant improvement between the two groups, but differences were observed in the intervention group. This may be due to the duration of the program, which was twice, as long as, in El-Shamy & El-Banna, (2018) study. The long-term effects of the treatment program were not examined in the Avcil et al., (2020) and El-Shamy & El-Banna, (2018) studies. Finally, the heterogeneity of the sample in terms of the type of CP and the functional level of the children are limitations acknowledged in those studies.
Interventional VR programs may contribute to the improvement of the voluntary motor control and coordination of children and adolescents with CP (Saif & Alsenany, 2015;Jung et al., 2020). Specifically, Al Saif & Alsenany, (2015), observed an improvement in upper limb and eye coordination. Additionally, Jung et al., (2020), report positive results in the voluntary motor control, of lower limb movements of children who participated in the group. In contrast, Chiu et al., (2014) found no significant differences in coordination, after completing the VR intervention program. The difference between the results may be due to the different context (home or clinic), in which the research was conducted. James et al (2015) report a significant improvement in visual perception, after completion of VR interventional programs. However, the weekly frequency and duration of the intervention program, was not adhered by all participants.    in the above studies may be important factors for differentiating the results (Arnoni et al., 2019;Jung et al., 2020;Pin & Butler, 2019;Park et al., 2021;Ramstrand & Lygnegård, 2012;Wade & Porter, 2012). The results for the effect of VR on walking were conflicting, since positive indications were found in two of four studies. Al Saif & Alsenany, (2015) and Tarakci et al (2016) observed significant changes in children's gait. In contrast, Jung et al (2020) and Pin & Butler (2019) did not discern a statistically significant improvement in children's gait, after completing the VR intervention programs. The different results may be due to the small number of participants, the different level of mobility, which varied from I-IV on the GMFCS scale, and the location of the study. In particular, the study by Al Saif & Alsenany (2015), Jung et al (2020) and Tarakci et al (2016) was carried out in a supervised therapeutic setting while Pin & Butler (2019) was carried out in an unsupervised school.
Three out of five studies included in this systematic review reported a positive effect of VR programs on gross motor function of children and adolescents with CP. Specifically, Arnoni et al., (2019) and Sahin et al., (2019) argued that the application of VR programs in addition to classical therapy has positive effects on gross motor function. However, Sahin et al., (2019) used toys, which did not include movements used in daily activities and were not suitable for direct gross motor training. Chen et al., (2012) found similar results applying a VR program to ambulatory children and adolescents, who did not receive any other treatment at the same time. In contrast, Pin & Butler, (2019) and Jha et al., (2021) did not find a significant improvement in gross mobility in the experimental group, after a combined VR program. The different results may be due to the different mobility level of the participants (GMFCS: I-IV, MACS: I-III), differences in intervention duration (20 -60 minutes), as well as the small number of participants (Arnoni et al., 2019;Chen et al., 2012;Pin & Butler, 2019).
The results for the effect of VR upper limb function were conflicting, since six out of eleven studies did not demonstrate improvement through VR interventions. Specifically, Al Saif & Alsenany, (2015), Okmen et al., (2015), El-Shamy & El-Banna, (2018) and Rostami et al., (2012) found significant improvements in the VR intervention groups compared with the corresponding control groups. Similarly, Sahin et al., (2019) found statistically significant changes in upper limb function in the group that received the VR intervention, even though the intervention group's games were not considered suitable for direct training. In contrast, in the intervention groups of Avcil et al., (2020), Chiu et al., (2014), Wang et al., (2021), James et al (2015), Jannink et al., (2008) and Reid & Campbell, (2006) an improvement of function capacity in the upper limbs was not observed after the VR intervention. The varied results may be due to the different types of VR intervention, used by the researchers in each study. In particular, with the use of the Nintendo Wii additionally to the usual treatment, of the children and adolescents participated, it appeared to improve the performance of the upper limb function capacity, but also, in its use in daily activities.
Exceptions are 2 researches. The study by Chiu et al., (2014) evaluated fine motor skills, with the NHPT and JTTHF scales. Their intervention produced no change and respectively Wang et al., (2021) who used BOT-2, PMAL-R and ABILHAND-Kids scales. Some less famous toys, such as Mitii, Eye Toy, Mandala Gesture Xtreme, did not show improvements in the examined variable, perhaps, because they are not designed for therapeutic purposes (James et al 2008).
Researches that had evaluated the effect of VR programs on upper limb, did not select a specific type of CP to include. Instead, they included children with any type of CP and different classification of upper limb functionality (MACS: I-V). In addition, the duration of the virtual reality intervention (20 to 90 minutes), as well as, the number of participants (10 to 102) are important factors, that may account for the differentiation of the results.
Of the 5 researches that evaluated DLA independence, a positive effect was reported in 2 of them. Specifically, in Sahin et al., (2019) and Tarakci et al., (2016) the independence of participants in the intervention group improved. However, in the study by Sahin et al., (2019) the games of the therapeutic program did not include movements that are used in daily activities, so they were not considered suitable for direct training. Additionally, Tarakci et al., (2016) reported that due to the different types of children's the results cannot be generalized. In contrast, Jha et al (2020), Okmen et al (2022) and Wang et al (2021) observed no significant changes between the 2 groups. The researchers report that the duration of the study was probably not long enough to support the effectiveness of the intervention, suggesting interventions lasting more than 6 weeks to achieve this goal. It is noted that children's functional capacity (GMFCS: I-III, MACS: I-III) and treatment time (from 45 minutes to 145 minutes) differed among the above studies.
Based on the analyzed studies, VR programs did not have a significant effect on the participation of children and adolescents with CP. Specifically, Atasavun Uysal & Baltaci, (2016), Reid & Campbell, (2006), Wang et al., (2021) did not observe any improvement in participation in daily activities, after the intervention programs. According to Atasavun Uysal & Baltaci, (2016) and Reid & Campbell, (2006), the results may be due to the small number of participants, the individual sessions and the lack of motivation, which did not contribute to the socialization of the participants. In contrast, in the study by James et al., (2015) a positive effect was observed, even though the maximum weekly dosage was not followed, by all in the intervention group. It should be noted that the duration of the intervention and the number of participants differed in the above research. In particular, in the studies of Atasavun Uysal & Baltaci, (2016) and James et al., (2015) each session lasted 30 minutes, while in Reid & Campbell, (2006) the session duration was 90 minutes and in Wang et al., (2021) 145 minutes.
The information from studies included and analyzed herein indicated the necessity to conduct more studies in VR-based rehabilitation to derive more concrete results on the effectiveness of such programs on the participation/ body structures and the functionality/activity of children and IJKSS 11(2):11-24 adolescents with CP. Furthermore, research on the type of VR training (used as a single intervention or in combination with parallel interventions), the frequency and duration of sessions, so that it can be included in the therapeutic routine of children and adolescents with cerebral palsy. Amongst the limitations of this systematic review is the exclusion of studies that were not randomized, that only studies written in the English language were included, no information on the magnitude of effects was provided, nor did it account for differences in the relative sizes of the included studies, based on the type of data analysis followed.

CONCLUSIONS
The purpose of the present systematic review was to investigate the effectiveness of VR therapeutic programs on the functionality of children and adolescents with cerebral palsy, in the light of the ICF framework. Analysis of the relevant studies revealed a variety of results, regarding its effectiveness. Specifically, the effects of VR on visual perception and balance were positive, while the effects on muscle strength, coordination, gross mobility, gait, upper limb function, DLA independence and participation were conflicting. Additionally, the methodological quality of most studies included in this review, was low to moderate. Important factors that may differentiate the results are the functional level of the participants, the sample size, the context in which the therapeutic intervention is carried out (rehabilitation center, home), as well as the conventional treatment that is carried out alongside the VR intervention programs. Therefore, it is considered necessary to conduct more randomized studies with a larger number of participants, greater homogeneity of the sample and VR as a unique means of treatment, in order to, derive reliable results for its effect on the functionality of children and adolescents with cerebral palsy.