Systematic search results
The flow of different search stages and reasons for exclusion were outlined using a PRISMA diagram flow as shown in (Fig. 1). A total of potentially relevant 2347 citations were yielded, 968 from Pubmed, 571 from Web of Science, 105 from the Cochrane Library, 524 from Google Scholar and 179 from Science Direct. The identified citations were exported to Mendeley software which initially removed 383 duplicates. Then, the full-text of 271 citations was retrieved after the screening of the titles and abstracts of the remaining identified citation. Finally, a total of 19 studies were included in our review based upon full-text examination.
Characteristics the included studies
A detailed information about the study populations, study interventions strategy and outcomes measured is shown in Tables 1, 2 and 3.
Studies participants
A total of 645 participants were recruited with a 570 were who continued to the post-intervention assessments. Sample size ranged from 16 to 102 participants. The majority of the recruited participants were males 299 (54.56%) from a total of 548 participants in 15 studies which specify the participants sex with 6 of them removed the withdrawn participant (dropped-out) from the personal characteristics [32, 35, 38, 43, 47, 56], while 3 studies did not specify the gender of the both groups [37, 40, 41]. The mean age was ranging from 7.05 to 11.67 years in the experimental group and from 7.25 to 12.4 in the control group with two study did not report the mean for each group [40, 41]. Participants in seven of the selected trials were hemiplegic [34, 36, 38, 42,43,44,45] with mixed topographical distribution in five [32, 37, 48, 49, 56], diplegic in four [35, 39, 42, 46], whereas three did not clarify the distribution of CP [40, 41, 47]. At least 55.56% (355) of the patients were affected by spastic hemiplegic, 128 (20.03%) with spastic diplegic CP, 10 (1.56%) with spastic quadriplegic, 5 (0.78%) with triplegic forms of CP.
Types of intervention
One of the included studies had a four comparison arms, comparing VR and rehabilitation to constraint-induced movement therapy (CIMT) and rehabilitation to VR, CIMT and rehabilitation to rehabilitation alone [34]. Two groups (VR versus control group (CG)) were selected for inclusion in our review. However, all the other studies had two arms comparing either VR alone or when combined with usual care or VR training with transcranial direct current stimulation to usual care or no intervention or sham transcranial current direct stimulation. In the term of sophostication, fifteen studies utilized the commercially low-cost sets, whereas four used the engineer-built. Participants in fourteen studies received VR as an adjunctive therapy to conventional treatment, whereas others in another three studies received VR alone and in two studies, the participants received VR followed by a period of conventional treatment or no treatment or vice versa utilizing a crossover design. Location of VR therapy varied from laboratory, clinic or home-based. An overview of the characteristics of the eligible studies is presented in Table 2.
Types of outcome measures
A variety of assessment tools were used to evaluate different aspects of neuromotor status (e.g., coordination, strength, muscle tone) and functional performance.
The International Classification of Function (ICF) outlines two main domains of function for assessment: body function and structure domain and activity and participation domain (subdivided into activity subdomain and participation subdomain).
Through the use of ICF classification, we found that the majority of the outcome measures used in the included studies fit within the activity subdomain of the ICF model with lesser extent measures falling under the body function and structure domain while the participation domain having the least number of outcome measures. Table 3 represents different assessment scales used with their ICF classification.
Fine motor coordination
Under the body function and structures lies joint kinematics and Visual Motor integration (VMI) test, whereas Jebsen Taylor Test of Hand Function (JTTHF), Nine-hole Peg test and Peabody Developmental Motor Scale-2 PDMS-2 assess activity. BurininksOsteretsky Test of Motor Proficiency (BOTMP) “subset 8” lies under both categories.
JTTHF was used in three studies, joint kinematics, VMI test, PDMS-2 and Nine-hole Peg test are used once.
Gross motor coordination
Under the body function and structure lies Modified sensory organization test, reactive balance, Static Posturography, joint kinematics, 10 s climbing test (10s CT), standing durations, 3D temporo-spatial and full-body kinematic gait and motor evoked potential analysis, while Pediatric balance scale (PBS), Box and Blocks test (BBT), Rhythmic weight shift, Walking Speed and Distance, 10 min walking test (10m WT),sit-to-stand test(STST), Timed up and Go test, functional reaching tests, Wii Nintendo Fit Balance and Game Scores lie under the activity section of coordination. BOTMP-2 lies under both categories.
PBS was used in five studies, stabilometric evaluation center of pressure (COP) in 4, 3D temporo-spatial and full-body kinematic gait analysis, timed up and go (TUG) in 3, joint kinematics and BBT twice and Modified sensory organization test, reactive balance, Rhythmic weight shift, 10 sCT, STST, standing durations, Walking Speed & Distance, 10 m WT, functional reaching tests, Wii Nintendo Fit Balance & Game Scores and motor evoked potential were used once.
Intervention protocols
Studies used different treatment strategies with different durations from 3 up to 20 weeks, session duration ranging from 15 to 90 min and frequency 2 to 6 sessions/week.
Effect of interventions
Fine motor coordination
Six studies provided post-intervention assessment of fine motor coordination on 255 participants [34, 36, 38, 40, 42, 44]. There was a low certainty level according to Cochrane RoB that VR a large beneficial effect than the controls immediately post-intervention (SMD 0.75, 95% CI 0.02–1.51) (Fig. 2). Between-study heterogeneity was considerable (I2 = 86%). On the other hand, when assessed by modified Sackett scale, the overall evidence for fine motor coordination was moderate that VR intervention is better than the control groups (83.3% n = 5 studies were scored as Ib).
Gross motor coordination
Fourteen studies on 363 children with CP were found to carry a low certainty level with respect to Cochrane RoB about the non-significant small beneficial effect on gross motor coordination brought out immediately following VR (SMD 0.15; 95% CI, 0.09 to 0.40) [32, 33, 35,36,37, 39, 41, 43, 45,46,47,48,49, 56] (Fig. 3). On modified Sackett’s scale, 55.04% of studies have a moderate evidence. Between-study heterogeneity was negligible (I2 = 24%). Since only 55.04% of the studies carry a moderate evidence, pooling of individual evidence scoring for each study to reach the overall evidence outweighs a limited evidence.
Correlation of the total scores obtained with both quality scales
There was a non-significant difference between Cochrane RoB and PEDro scores (p value = 0.248) carrying a fair positive correlation (r = 0.28). The degree of overall agreement between the total scores of the two quality scales was slight (κ = 0.02; (95% CI − 0.02 to 0.50) with non-significant difference (p value = 0.433).
High and low PEDro scores revealed moderate agreement with Cochrane RoB (κ = 0.43 (95% CI 0.36–0.49) and 0.46; (95% CI 0.40–0.51)), respectively, with a significant difference (p = 0.0001) for both. However, when compared with Cochrane RoB, moderate quality studies on PEDro exhibited slight agreement with no significant difference (κ = 0.10; (95% CI − 0.09 to 0.29), p = 0.404).
Quality assessment
Figure 4 and Table 4 display a summary of the quality appraisal scores for each study by Cochrane RoB and PEDro, respectively. Scores were heterogeneous depending on the trial and the quality scale used. When assessed by Cochrane risk of bias assessment tools, all the included studies are considered to have a high risk of bias.
Whereas, when assessed by PEDro scale, more than half of the included studies (57.9%, n = 11) were of moderate quality [32, 34, 35, 37, 39, 41, 42, 44,45,46, 48], about the quarter of studies (26.13%, n = 5) of a high quality [36, 38, 43, 47, 49], whereas three studies (15.8%) were of a low quality [33, 40, 56].