Nintendo® Wii Therapy Improves Upper Extremity Motor Function in Children with Cerebral Palsy: A Systematic Review with Meta-Analysis

Background: Nintendo® Wii-based therapy (NWT) is a non-immersive virtual reality therapy used to recover upper extremity (UE) motor function in children with cerebral palsy (CP). We aimed primarily to elucidate the effectiveness of NWT in improving UE motor and functional impaired abilities in children with CP, compared to conventional therapy or no intervention. The secondary aim was to assess if NWT is more effective when used alone or combined with conventional therapy. Methods: A systematic review with meta-analysis was conducted from a bibliographic search in PubMed, Scopus, PEDro, Web of Science, and CINHAL, ending in October 2021, in accordance with PRISMA guidelines. We included randomized controlled trials that compared NWT vs. conventional therapy or no intervention in terms of their impact on different UE impaired abilities (grip strength, tip grip strength, UE dissociated movements, functional capacity in daily living activities, gross and fine motor dexterity, and grasping ability) in children with CP. Effect size was calculated with standardized mean difference (SMD) and its 95% confidence interval (95% CI). Results: Nine studies (276 participants) were included. NWT is more effective than conventional therapy at improving grip strength (SMD = 0.5, 95% CI 0.08, 0.91), tip grip strength (SMD = 0.95, 95% CI 0.3, 1.61), and grasping ability (SMD = 0.72, 95%CI 0.14, 1.3). NWT is more effective than conventional therapy at improving functional capacity in daily living activities (SMD = 0.83, 95% CI 0.07, 1.56). For fine manual dexterity, NWT was better than no intervention (SMD = 3.12, 95% CI 1.5, 4.7). Conclusions: Our results indicate that NWT is effective at improving various UE impaired motor skills in children with CP.


Introduction
Cerebral palsy (CP) is defined as a group encompassing permanent, changeable movement, posture, and motor function disorders [1,2] which arise as a consequence of damage to the developing fetal or infant brain [3]. Currently, CP is one of the major causes of childhood physical disability [4], with an estimated prevalence of 2-3 cases per 1000 [5,6]. CP produces general disorders in muscle tone, joint stiffness, muscle weakness, and muscle pain [7], and decreases motor control [8]. These changes influence physical function [9] and are more pronounced in upper-extremity-related functional capacity, and even more so in hand dexterity [10]. Approximately 83% of children with CP have upper extremity (UE) motor disabilities, and 36% have wrist and hand muscle contractures [11]. The changes in motor function of the UE lead to difficulties in performing daily living activities [12] and self-care, negatively impacting the quality of life of these patients, and increasing the burden on their caregivers. It also limits children's participation in family, school, and leisure activities [13]. Improving UE motor function may facilitate their daily living Scopus (TITLE-ABS-KEY ("cerebral palsy") AND TITLE-ABS-KEY ("nintendo wii" OR "wii" OR "wii fit" OR "wii balance board" OR "virtual reality") AND TITLE-ABS-KEY ("upper extremity" OR "upper limb" OR "arm" OR "hand" OR "elbow" OR "shoulder" OR "wrist")) Web of Science TOPIC:((cerebral palsy)) AND TOPIC:((nintendo wii OR wii OR wii fit OR wii balance board OR virtual reality)) AND TOPIC: ((upper

Study Selection: Inclusion and Exclusion Criteria
The study selection process was carried out independently by two authors (D.M.-C. and E.O.-G.), who reviewed all retrieved records by title and abstract. Each study was examined in detail by two reviewers. All disagreements that arose during this phase were resolved by a third author (M.C.O.-P.). A study was only included in our review if it met all the inclusion criteria: (1) RCTs; (2) including subjects diagnosed with CP; (3) with at least two groups of intervention, in which one received NWT in comparison with conventional therapy or no intervention; (4) studies that assessed the different outcomes related to UE motor function; and (6) studies that provided quantitative data of the outcomes of interest to carry out the meta-analysis. The exclusion criteria established were: (1) RCTs with groups of patients with different central nervous system diseases (for example, heterogeneous groups of neurological patients that included children with CP); (2) outcomes that could not be unified when performing the meta-analysis; (3) RCT studies that did not provide sufficient data to be included in the quantitative synthesis of this review.

Data Extraction
Two authors (D.M.-C. and I.C.-P.) independently extracted data from the studies included using a Microsoft Excel standardized data-collection form designed specifically for this research, similar to that used in previous meta-analyses [25,43,44]. Disagreements were resolved by the third author (M.C.O.-P.).
Specifically, in this review, from each selected study, the characteristics collected were the following: (1) overall characteristics (authorship and publication date, study design, country); (2) participant characteristics (total sample size, number of groups, participants per group, age, sex ratio, type of CP, level of the Gross Motor Function Classification System [GMFCS], and time since diagnosis); (3) characteristics of the experimental and control intervention (type, number of sessions, number of weeks or months, number of sessions per week, and duration in minutes of each session); (4) data from outcomes of interest (mean and standard deviation); and (5) follow-up period (short-term, medium, or long-term follow-up). When a study did not provide standard deviations, it was estimated using standardized transformations through the standard error, range, interquartile range, and median [42].

Outcome Measures
The outcomes assessed in this systematic review with meta-analysis were the following: grip strength, tip grip strength, functional capacity in daily living activities and self-care, UE dissociated movements, gross manual dexterity, fine manual dexterity, and grasping ability. Table 2 summarizes information related to the outcome measures.

Risk of Bias and Quality of Evidence Assessment
The risk of bias in the studies selected was assessed through the Cochrane Collaboration Risk of Bias Tool. This scale comprises seven items (selection, performance, detection, attrition, reporting, and other bias), and gives the labels of low, uncertain, and high risk of bias [45].
The quality of evidence in each meta-analysis was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system [46]. We followed the GRADE checklist proposed by Meader [47] to estimate the quality of evidence, taking into account the risk of bias of each selected study, inconsistency, inaccuracy, lack of directiveness, and risk of publication bias. The quality of evidence could be high (if our findings were robust); moderate (if our results changed, including new studies); low (if the level of confidence in our pooled effect was very slight); and very low (when some items were not present). The quality of evidence of each meta-analysis was downgraded from high quality by one level for each factor that we found. In the case of the presence of several limitations, the overall quality level was downgraded by two levels.

Statistical Analysis and Additional Analyses
Comprehensive Meta-Analysis 3.0 (Biostat, Englewood, NJ, USA) [48] was used to perform the meta-analysis by two authors (E.O.-G. and I.C.-P.). The pooled effect was calculated using the Cohen's Standardized Mean Difference (SMD) [49] that may be interpreted at three effect strength levels: small (SMD = 0.2), moderate (SMD = 0.5), and large (SMD ≥ 0.8) [50].When an outcome was assessed with data from studies that use the same measurement test, we calculate the mean difference (MD) to compare with the minimally clinically important difference (MCID) for that test. Forest plots were used to display our findings [51]. Risk of publication bias was analyzed visualizing the funnel plot (asymmetric indicates risk of publication bias, and symmetric low risk) [52] and with the p-value for the Egger test (where if p < 0.1 it indicates that risk of publication bias exists) [53]. In addition, we calculated the adjusted effect considering any possible publication bias, using the Trim-and-fill method [54]. To estimate the risk of publication bias, our software requires almost three studies, so sometimes could not be estimated. Finally, the inconsistency was assessed with the level of heterogeneity through the degree of inconsistency (I 2 ) from Higgins (small <25%, moderate 25-50% and large ≥50%) and p-value for Q-test (p < 0.1 indicates presence of heterogeneity) [55,56].

Study Selection
Initially, 508 references were retrieved from databases after the bibliographic search, and a further 3 studies from other resources. After duplicate studies were removed, 277 studies were screened by title/abstract. Of these, 23 records were excluded for not being relevant and 245 references were removed for not meeting the inclusion criteria. Finally, 9 studies were included in the quantitative synthesis [57][58][59][60][61][62][63][64][65]. The PRISMA flow chart ( Figure 1) summarizes the screening and study selection phase.

Study Selection
Initially, 508 references were retrieved from databases after the bibliographic search, and a further 3 studies from other resources. After duplicate studies were removed, 277 studies were screened by title/abstract. Of these, 23 records were excluded for not being relevant and 245 references were removed for not meeting the inclusion criteria. Finally, 9 studies were included in the quantitative synthesis [57][58][59][60][61][62][63][64][65]. The PRISMA flow chart (Figure 1) summarizes the screening and study selection phase.

Characteristics of the Studies Included in the Review
The studies included were carried out between 2012 and 2021 in Turkey [57,59,60], Saudi Arabia [58,62], Australia [61], India [63], South Korea [64], and China [65]. The studies included reported data from 274 participants (mean age 9.56 ± 2.35 years old and 48% girls). Eight studies [57][58][59][61][62][63][64][65] reported data from participants diagnosed with spastic, diplegic, or tetraplegic CP, and one study [60] reported non-specific CP. Related to GMFCS levels, 2 studies [57,59] included patients in levels I and II; one study [58] included patients in level III; 2 studies [60,63] included patients in levels I-IV; another study [61] included patients in all levels I-V; and 3 studies [62,64,65] did not report the GMFCS level. The experimental group consisted of 138 individuals receiving NWT or NWT + conventional therapy, and the control group included 136 individuals (conventional therapy or no intervention). Seven studies [57,[59][60][61][62][63]65] compared the effect of NWT plus conventional therapy versus conventional therapy; one study [64] compared NWT versus conventional therapy, and another study [58] compared NWT vs. no intervention. The duration of the therapy in weeks was 6 weeks or less in two studies [57,63] and more than 6 weeks in seven studies [58][59][60][61][62]64,65]. The duration of each session was less than 30 minutes in three studies [57][58][59] and more than 30 minutes in six studies [60][61][62][63][64][65]. Four studies [57,59,64,65] applied the therapy twice a week; 3 studies [60][61][62] applied it three times a week; and 2 studies [58,63] applied it more than three times a week. Table 3 shows the main characteristics of the studies included in this review.  Table 4 shows the risk of bias in the studies included. A high risk of selection bias was found in five studies [57,58,60,61,64]. No studies were able to blind the type of therapy to the CP participants, increasing the risk of selection bias. Five studies [57][58][59][60]65] did not blind the assessors after the treatment, increasing the risk of detection bias. In general, the overall quality of the included studies was moderate due to the possible presence of selection, performance, and detection bias. Abbreviations: "+" = high risk of bias, "-" = low risk of bias, "?" = inadequate data for the evaluation.

Tip Grip Strength
Two RCTs [60,62] provided data for 70 children with CP to analyze the effect of NWT on tip grip strength (Table 5 and Figure 2). On the one hand, very low-quality evidence with a large effect (SMD = 0.95; 95% CI 0.3, 1.61; p = 0.004) favored NWT + conventional therapy vs. conventional therapy, without heterogeneity. On the other hand, very lowquality evidence with a large effect (SMD = 0.8; 95% CI 0.01, 1.49; p = 0.047) favored NWT vs. conventional therapy without heterogeneity.

Grasping Ability
Two RCTs [57,63] reported data for 48 children with CP to assess the effect of NWT on grasping ability (Table 5 and Figure 2). The results showed very-low quality evidence of a medium-large effect (SMD = 0.72; 95% CI 0.14, 1.30; p = 0.014) favoring NWT+ conventional therapy vs. conventional therapy, with an increase of 4.35% (95% CI 0.96, 7.75; p = 0.003) in the "grasp" domain of the QUEST questionnaire. Heterogeneity was not found, and the precision level was low (24 participants per study).

Functional Capacity in Daily Living Activities and Self-Care
Four RCTs [57,59,60,65] reported data for 102 children with CP to assess the effect of NWT on functional capacity in daily living activities and self-care (Table 5 and Figure 3). No statistically significant differences (SMD = 0.43; 95% CI −0.04, 0.91; p = 0.074)) were

Tip Grip Strength
Two RCTs [60,62] provided data for 70 children with CP to analyze the effect of NWT on tip grip strength (Table 5 and Figure 2). On the one hand, very low-quality evidence with a large effect (SMD = 0.95; 95% CI 0.3, 1.61; p = 0.004) favored NWT + conventional therapy vs. conventional therapy, without heterogeneity. On the other hand, very low-quality evidence with a large effect (SMD = 0.8; 95% CI 0.01, 1.49; p = 0.047) favored NWT vs. conventional therapy without heterogeneity.

Grasping Ability
Two RCTs [57,63] reported data for 48 children with CP to assess the effect of NWT on grasping ability (Table 5 and Figure 2). The results showed very-low quality evidence of a medium-large effect (SMD = 0.72; 95% CI 0.14, 1.30; p = 0.014) favoring NWT+ conventional therapy vs. conventional therapy, with an increase of 4.35% (95% CI 0.96, 7.75; p = 0.003) in the "grasp" domain of the QUEST questionnaire. Heterogeneity was not found, and the precision level was low (24 participants per study).

Functional Capacity in Daily Living Activities and Self-Care
Four RCTs [57,59,60,65] reported data for 102 children with CP to assess the effect of NWT on functional capacity in daily living activities and self-care (Table 5 and Figure 3). No statistically significant differences (SMD = 0.43; 95% CI −0.04, 0.91; p = 0.074)) were found between NWT+ conventional therapy vs. conventional therapy without risk of publication bias (Egger p = 0.66 and 0% variation in the trim-and-fill method). However, very low-quality evidence of a large effect favored NWT (SMD = 0.82; 95% CI 0.07, 1.56; p = 0.032) in comparison with conventional therapy. found between NWT+ conventional therapy vs. conventional therapy without risk of publication bias (Egger p = 0.66 and 0% variation in the trim-and-fill method). However, very low-quality evidence of a large effect favored NWT (SMD = 0.82; 95% CI 0.07, 1.56; p = 0.032) in comparison with conventional therapy.

Upper Extremity Dissociated Movements
Two RCTs [57,63] reported data from 48 children with CP to analyze the effect of NWT on UE dissociated movements (Table 5 and Figure 3). Our results showed very lowquality evidence of a medium-large effect favoring NWT + conventional therapy (SMD = 0.73; 95% CI 0.15, 1.3; p = 0.013) compared to conventional therapy, increasing the score of the domain "dissociated movements" of the QUEST questionnaire by 4.094% (95% CI 0.75, 7.43; p = 0.016). Heterogeneity was not found (I 2 0%; p = 0.33) and the precision level was low (24 participants per study).

Gross Manual Dexterity
Four RCTs [57,60,61,63] provided data for 135 children with CP to assess the effect of NWT on gross manual dexterity (Table 5 and Figure 4). No statistically significant differences between NWT+ conventional therapy vs. conventional therapy (SMD = −0.28; 95% CI −0.67,0.1; p = 0.149) and between NWT vs. conventional therapy (SMD = −0.12; 95% CI −0.84 to 0.6; p = 0.74) were found without risk of publication bias. Heterogeneity was not present in any subgroup, and the precision level in both comparisons was low (30 and 35 participants per study, respectively).

Upper Extremity Dissociated Movements
Two RCTs [57,63] reported data from 48 children with CP to analyze the effect of NWT on UE dissociated movements (Table 5 and Figure 3). Our results showed very low-quality evidence of a medium-large effect favoring NWT + conventional therapy (SMD = 0.73; 95% CI 0.15, 1.3; p = 0.013) compared to conventional therapy, increasing the score of the domain "dissociated movements" of the QUEST questionnaire by 4.094% (95% CI 0.75, 7.43; p = 0.016). Heterogeneity was not found (I 2 0%; p = 0.33) and the precision level was low (24 participants per study).

Gross Manual Dexterity
Four RCTs [57,60,61,63] provided data for 135 children with CP to assess the effect of NWT on gross manual dexterity (Table 5 and Figure 4). No statistically significant differences between NWT+ conventional therapy vs. conventional therapy (SMD = −0.28; 95% CI −0.67,0.1; p = 0.149) and between NWT vs. conventional therapy (SMD = −0.12; 95% CI −0.84 to 0.6; p = 0.74) were found without risk of publication bias. Heterogeneity was not present in any subgroup, and the precision level in both comparisons was low (30 and 35 participants per study, respectively).

Fine Manual Dexterity
Three studies [58,61,64] provided data for 114 children with CP to assess the effect of NWT on fine manual dexterity (Table 5 and Figure 4). No statistically significant differences (SMD = −0.05; 95% CI −0.51, 0.41; p = 0.82) were found between NWT+ conventional therapy vs. conventional therapy without heterogeneity. However, very low-quality evidence of a large effect (SMD = 3.12; 95% CI 1.53, 4.7; p < 0.001) favored NWT vs. no intervention without heterogeneity. The precision level in both comparisons was low (40 and 34 participants per study, respectively).

Discussion
The main aim of our meta-analysis was to evaluate the effect of NWT on UE motor function recovery in children with CP. We included nine RCTs that reported quantitative data from 274 subjects with CP to analyze seven abilities dependent on UE motor function. Our meta-analysis found the following: (1) NWT can be considered effective for the treatment of UE function in children with CP; (2) NWT plus conventional therapy is better than conventional therapy alone for recovering grip strength, tip grip strength, grasping ability, and UE-dissociated movements in children with CP; (3) NWT is better than conventional therapy to improve tip grip strength and functional capacity in daily living activities and self-care in children with CP; and (4) NWT is better than no intervention to improve fine manual dexterity in children with CP. Our study differs from previous reviews [34,66] as it is the first meta-analysis to assess the effect of NWT in this variety of UE motor dimensions. It is important to point out that, although this is a novel study, these results should be considered with caution due to the low number of studies included in each meta-analysis, which reduces the robustness of the results. However, taking into account the low level of evidence, this meta-analysis is the first that provide information about the use of NWT to restore UE motor-function in children with CP. Previous reviews assessed the effect of some non-immersive VR devices, such as NWT or custom video

Fine Manual Dexterity
Three studies [58,61,64] provided data for 114 children with CP to assess the effect of NWT on fine manual dexterity (Table 5 and Figure 4). No statistically significant differences (SMD = −0.05; 95% CI −0.51, 0.41; p = 0.82) were found between NWT+ conventional therapy vs. conventional therapy without heterogeneity. However, very low-quality evidence of a large effect (SMD = 3.12; 95% CI 1.53, 4.7; p < 0.001) favored NWT vs. no intervention without heterogeneity. The precision level in both comparisons was low (40 and 34 participants per study, respectively).

Discussion
The main aim of our meta-analysis was to evaluate the effect of NWT on UE motor function recovery in children with CP. We included nine RCTs that reported quantitative data from 274 subjects with CP to analyze seven abilities dependent on UE motor function. Our meta-analysis found the following: (1) NWT can be considered effective for the treatment of UE function in children with CP; (2) NWT plus conventional therapy is better than conventional therapy alone for recovering grip strength, tip grip strength, grasping ability, and UE-dissociated movements in children with CP; (3) NWT is better than conventional therapy to improve tip grip strength and functional capacity in daily living activities and self-care in children with CP; and (4) NWT is better than no intervention to improve fine manual dexterity in children with CP. Our study differs from previous reviews [34,66] as it is the first meta-analysis to assess the effect of NWT in this variety of UE motor dimensions. It is important to point out that, although this is a novel study, these results should be considered with caution due to the low number of studies included in each meta-analysis, which reduces the robustness of the results. However, taking into account the low level of evidence, this meta-analysis is the first that provide information about the use of NWT to restore UE motor-function in children with CP. Previous reviews assessed the effect of some non-immersive VR devices, such as NWT or custom video games, but did not specifically assess the effect of NWT protocols for these seven UE abilities. Previous studies such as that of Alrashidi M et al. analyzed the effect of VR on UE function in children with CP but, in contrast with our meta-analysis, they did not analyze the effect of a particular VR device, they included fewer trials and they only assessed two domains of UE function. Other reviews only reported the effect of all VR devices on global UE function, but we showed that it is necessary to assess the effect of this therapy for each specific domain of UE in children with CP. We consider that grip strength and fine manual dexterity are different skills, and the effect of NWT was likely to be different for each. NWT may be of interest to physical clinicians as it is cheaper, more accessible, and easier to use than other non-immersive VR devices, and it is important to analyze the effect that NWT can produce for different UE motor skills.
For grip strength, our results showed that NWT is effective when used as complementary therapy with conventional therapy, leading to an increase of 1.8 points in dynamometer measurement. For this variable, both therapies are effective separately, without significant differences between them. For tip grip strength, our results showed that NWT is effective when it is used as a single therapy and when it is used combined with conventional therapy, both in comparison with conventional therapy. The improvement in grip strength and tip grip strength may be due to grasping the Nintendo joystick and pressing its buttons to play and interact with the virtual environment [67]. The effect of NWT on grip strength was not assessed in previous reviews. As such, our results cannot be directly compared with other reviews as they did not specifically assess the effect of NWT on these outcomes. Only Alrashidi, M and colleagues included three RCTs (two of NWT) to assess the effect of non-immersive VR in improving grip strength in children with CP. However, although it appears that non-immersive VR is effective in improving grip strength, their review did not present meta-analyses and some important studies were not included, so their findings may be questionable, meaning that our findings are difficult to compare with them. In addition, in other previous meta-analyses that assessed the effect of the Leap Motion Controller, as an example of non-immersive VR, on grip strength [68], non-immersive VR was found to be useful to improve this variable in different neurological diseases, such as CP. Therefore, our results reinforce the use of non-immersive VR devices to increase grip strength in children with CP.
For grasping ability, our results showed that the use of NWT more conventional therapy is better than conventional therapy, with a medium-high effect. In this case, it was not possible to assess if NWT alone is better than conventional therapy or no intervention due to a lack of studies. The improvement in grasping ability can be also justified by the use of the remote control, as in grip strength and tip grip strength, since in order to perform the different functions requested by the games, it is necessary to hold the Nintendo ® Wii joystick constantly in a deliberate position between the hands. This continuous grip of the joystick and increase in grip strength may explain the improvement in grasping ability. This is of significant clinical interest as NWT may help patients with CP to better grip objects used in everyday life and may in this way increase personal autonomy and self-care.
The results obtained for functional capacity in daily living activities and self-care showed an improvement with a large effect for NWT compared to conventional therapy. When NWT plus conventional therapy was compared to conventional therapy, no significant differences were found between therapies. Both comparisons are at the limit of significance. Future studies will indicate if both therapies could be complementary. For now, the results indicate that NWT can be considered a good therapeutic option to improve functional capacity in activities of daily living and self-care. According to Avcil E et al. (2020), the improvements in this variable may be because the exercises performed in the chosen video games were similar to the gestures performed in day-to-day activities. The use of these video games, despite the fact that they were not developed for rehabilitation purposes, could indirectly produce an improvement in the gestures required for daily living activities. This could justify the use of NWT alone as a treatment option.
For UE-dissociated movements, our results showed an improvement with a mediumlarge effect of the NWT more conventional therapy vs. conventional therapy. However, we could not analyze the effect of NWT vs. conventional therapy due to a lack of studies.
Alrashidi, M and colleagues assessed the effect of non-immersive VR (NWT and Xbox) in UE abilities using QUEST, showing improvements in this outcome. However, it is difficult to compare our review with theirs, as our meta-analysis clarifies its findings, indicating that NWT combined with more conventional therapy is able to improve children's UE abilities. Nintendo ® Wii games require widely differing movements related to the affected UE to correctly complete the virtual activities, and to complete the virtual tasks it is necessary to be able to dissociate both UEs. These movements are carried out with a joystick and displayed on a screen. The continuous and repetitive exposition and training of the UE movements in these virtual environments may explain the improvement in the ability to dissociate UE movements in children with CP.
When fine manual dexterity was assessed after application of NWT, our results showed that NWT is better than no intervention. However, we did not find a greater effect for NWT more conventional therapy or NWT over conventional therapy for fine or gross manual dexterity. Both types of therapies improve gross manual dexterity and fine manual dexterity, but neither therapy is superior to the other. Our results, although we only assessed NWT, may be in consonance with previous reviews that determined that non-immersive VR devices could be useful to improve fine manual dexterity in various non-traumatic neurological diseases, including CP [36,68].
The improvements that NWT can produce in the UE motor skills of children with CP may be based on neuroplasticity mechanisms and motor learning induced by multisensory stimulation during the use of NWT. Nintendo ® Wii games help to immerse patients in virtual environments and allow them to dynamically interact with situations and activities that are impossible in the real world. In neuroimaging studies carried out in children with CP, it has been observed that the use of VR induces neuroplasticity and causes changes at the level of the cerebral cortex [69]. Using NWT, activity-dependent plasticity is mainly achieved based on the execution of specific activities repeatedly and within certain intensity and time parameters, as well as multisensory stimulation [70,71]. It also produces improvements at the motor level with respect to conventional therapy, thus favoring the use of systems that relate to perception, memory, and learning in children with CP [72,73]. Various Nintendo ® Wii games could comply with the principles of neuroplasticity and motor learning indicated above. Several studies have analyzed the effect of VR in patients with neurological disorders, concluding that repetitive exercises and multisensory stimulation focused on DLAs stimulate motor evoked potentials, influencing corticospinal interactions in the premotor and motor areas of the brain [74]. This allows the reorganization of the neural networks that are responsible for controlling movement [75,76], due to increased activation of the primary motor cortex and the somatosensory cortex on the affected side [76,77]. Likewise, several studies in subjects who have suffered a stroke suggest that the use of non-immersive VR produces changes in the volume of gray matter in the motor and premotor regions of the affected extremity, which are correlated with the improvement of motor skills in brain areas that are not damaged. Specifically, correlations between changes in gray matter volume, increased range of motion of the extremity, and increased grip strength were found [78]. In addition, the use of VR and conventional therapy seems to generate a reduction in disabling motor symptoms such as spasticity [79], which can be explained by greater inhibitory control and a decrease in alpha neuron reflexes, which produces better motor control [80].
Several authors have affirmed that, when a VR device is adapted to the patient, benefits are obtained in terms of UE mobility [80] and strength [81]. In addition, it seems that the use of VR in children increases their motivation during therapy and favors the practice of physical activity within safe environments. This translates into greater adherence to treatment compared to other types of therapies [33]. Finally, this type of therapy can be applied at home, favoring the continuity of treatment over time [82]. This study presents some strengths. To begin, it is the first meta-analysis to assess the specific use of NWT to improve different motor and functional UE abilities in children with CP. Next, authors carried out a detailed synthesis of the tests employed in the studies included, allowing us to identify in each in each outcome whether NWT is effective. Finally, our meta-analysis points to NWT as a useful non-immersive VR device to be used in the UE recovery of children with CP. However, we cannot argue that NWT is better than other non-immersive VR devices. These findings may allow clinicians to make decisions about the employment of NWT to improve UE motor function in these patients.
This systematic review has several limitations. First, generalization of our results may be affected by the small number of RCTs included in each meta-analysis. Additionally, the small number of comparisons regarding the type of intervention in each meta-analysis markedly reduced the quality of the evidence in our findings. Likewise, the effect of therapy could only be assessed on the affected side due to the absence of data for the non-affected side. Another limitation is the small number of participants per study, which reduces the precision of the findings, despite the fact that the sample size tends to be low in this type of study. The inability to blind participants and therapists, and in some cases assessors, may have led to the presence of selection and classification bias in the studies included in the meta-analysis. In addition, variations in the trim-and-fill method of more than 10% indicate the presence of publication bias, which also makes it difficult to generalize our results. The low level of evidence in each meta-analysis is an important limitation that may affect the generalization of our findings. However, a strength of this study is that there are not enough studies published that clarify or replicate other researchers' results, so the meticulous analysis of each measurement instrument used by each of the RCTs included in this review, as well as the identification of the domains and subscales of each of the instruments, allowed us to delimit the different outcomes to be analyzed within the greater construct that is UE function. This effort to analyze and identify outcomes makes this the first systematic review with meta-analysis to focus on UE functioning in this type of patient. It would be interesting in future research to carry out RCTs of high methodological quality, with a larger sample size, in order to analyze in depth this type of intervention in children with CP.

Conclusions
Our results showed that NWT can be considered effective for the treatment of UE motor function in children with CP. We also found that NWT plus conventional therapy may be better than conventional therapy to recover grip strength, tip grip strength, grasping ability, and UE-dissociated movements in children with CP. NWT is better than conventional therapy to improve tip grip strength and functional capacity in daily living activities and self-care in children with CP. Finally, NWT is better than no intervention to improve fine manual dexterity in children with CP, but no effectiveness of NWT alone or combined with conventional therapy was found for gross manual dexterity. Although more studies are needed to evaluate the effect of NWT on all outcomes related to UE function in children with CP, the findings presented in this meta-analysis may help clinicians to include NWT in the UE recovery of children with CP.

Conflicts of Interest:
The authors declare no conflict of interest.