Upper Limb

Virtual Reality

Virtual reality interventions facilitate rehabilitation through computer based, interactive, and multisensory experiences that occur in real time. Users are able to engage with simulated objects or events in a motivating and fun environment to develop a range of skills, movements or task-based techniques. Most importantly, virtual reality interventions meet the four guiding principles of rehabilitation: intensity, task-specific training, biofeedback and motivation (Dias et al. 2019). In addition, virtual reality based neuro-rehabilitation has been shown to engage the mirror-neuron system including the frontal, parietal and temporal lobes to encourage cortical reorganization and functional recovery (Kirshblum et al. 2004). In light of this, a variety of virtual intervention systems have been designed specifically for therapeutic use (e.g. Cyber Touch glove or Toyra) or developed using existing gaming consoles (e.g. Nintendo Wii). As technology becomes increasingly accessible and affordable, virtual reality interventions have the potential to improve upper extremity function and transfer therapy gains into activities of daily living for innumerable people. Despite this, few studies have investigated the use of virtual reality interventions for upper extremity rehabilitation following spinal cord injury.

The methodological details and results of these studies (n=6) are presented in Table 6.

Author Year

Country

Research Design

PEDro Score

Sample Size

 Methods Outcomes
Prasad et al., 2018

India

RCT

PEDro=7

N=22

Population: Virtual reality and occupational therapy: Mean age=23.7±5.2 yr; Gender: males=11, females=1; Time since injury: 15.2 mo; Level of injury: C5=5, C6=6, C7=1; Severity of injury: AISA A=1, B=6, C=2, D=3.

Occupational therapy: Mean age=33.9±7.1 yr; Gender: males=10, females=0; Time since injury: 10.2 mo; Level of injury: C5=6, C6=3, C7=1; Severity of injury: AISA A=4, B=3, C=2, D=1.

Intervention: Participants were randomized to receive a virtual reality intervention (using Nintendo Wii) along with conventional occupational therapy (n=12), or conventional occupational therapy alone (n=10). Outcome measures were assessed at baseline, 2, 4, and 6 wk post-intervention.

Outcome Measures: CUE; BBT; SCIM.

 1.     No significant difference in hand function were observed between the groups for all outcome measures (p>0.05).
Dimbwadyo-Terrer et al., 2015

Spain

RCT

PEDro=6

N=31

Population: Conventional therapy and a virtual reality program: Mean age=34.5±13.7 yr; Gender: males=10, females=6; Time since injury: 4.3 mo; Level of injury: C5=7, C6=3, C7=5, C8=1; Severity of injury: AISA A=11, B=5.

Conventional therapy: Mean age=40.3±13.6 yr; Gender: males=12, females=3; Time since injury: 5.6 mo; Level of injury: C5=9, C6=2, C7=2, C8=2; Severity of injury: AISA A=10, B=5.

Intervention: Participants were randomized to receive a virtual reality program in combination with conventional therapy (n=16) or only conventional therapy (n=15). The intervention group received 15 sessions with Toyra virtual reality system for 5 wk, 30 min/d, 3d/wk in addition to conventional therapy. Outcome measures were assessed at baseline, after intervention, and at a 3 mo follow up.

Outcome Measures: MMT; FIM; SCIM III; BI; MB; MI.

1.     The control group showed significant improvements in the manual muscle test (p=0.043). in the follow-up evaluation.

2.     Both groups demonstrated clinical, but non-significant changes to their arm function.

3.     No significant differences were observed between groups for SCIM III, FIM, BI, MB, or MI.

4.     All patients showed a high level of satisfaction with the virtual reality system.

Dimbwaydo et al., 2013

Spain

RCT

PEDro=3

N=14

Population: Intervention: Time since injury: <6 mo; Level of injury: C5–C8.

Control: Time since injury: <6 mo; Level of injury: C5–C8.

Intervention: Participants were randomized to receive conventional therapy in addition to a virtual reality system (n=) for evaluation of ADLs or no virtual reality system and conventional therapy (n=). Outcome measures were assessed

Outcome Measures: Dexterity; Coordination and grip functions; Kinematic and functional parameters.

1.     Significant improvements were observed in parameters related to dexterity, coordination and grip functions (p<0.05) after treatment in the intervention group.

2.     No significant differences in kinematic variables and functional status were observed between groups (p>0.05).
Dimbwadyo et al., 2015

Spain

PCT

N=9

Population: Intervention: Mean age=54.3±9.9 yr; Gender: males=5, females=1; Time since injury: 5.8 mo; Level of injury: C4=1, T4=4; Severity of injury: AISA A=5, D=1.

Control: Mean age=44.2±22.9 yr; Gender: males=2, females=1; Time since injury: 5 mo; Level of injury: T4=2, T6=1; Severity of injury: AISA A=3.

Intervention: Participants in the intervention group (n=6) underwent a virtual reality training program with the use of a data glove for two weeks, while participants in the control group (n=3) only underwent traditional rehabilitation. Outcome measures were assessed at baseline and at 2wk.

Outcome Measures: MB; BI; SCIM; NHPT; JHFT.

1.     No statistical significance was found in any of the outcome measures.

2.     The data glove group seemed to obtain clinical changes in MB, functional parameters, dexterity, coordination and fine grip tests.
Seanez-Gonzalez et al., 2016

USA

Pre-Post

N=5

Population: Mean age=44.6 yr; Gender: males=5, females=0; Time since injury: 11.6 yr; Level of injury: C5 – C6; Severity of injury: Not reported.

Intervention: Participants performed five visu-spatial motor training tasks over 12 sessions (two to three sessions per wk). Subjects controlled a cursor with movements of the shoulders using a body-machine interface. Outcome measures were assessed at baseline and within two days of training completion.

Outcome Measures: MMT; Isometric force; Beck Depression Inventory (BDI); FIM; Fractional anisotropy (FA).

1.     The total MMT score improved significantly for all subjects after training (p=0.037).

2.     The total isometric force exerted by the subjects’ shoulders improved significantly after 12 training sessions (p=0.012).

3.     No significant differences were observed over time for the BDI or FIM (p>0.05).

4.     Motor training significantly increased FA, indicating localized white matter microstructure changes (p=0.03).
Dimbwadyo et al., 2015

Spain

Pre-Post

N=15

Population: Mean age=34.5±13.7 yr; Gender: males=9, females=6; Time since injury: 4.3 mo; Level of injury: C5=7, C6=3, C7=4, C8=1; Severity of injury: AISA A=10, B=5.

Intervention: Participants received daily conventional therapy complemented with virtual reality ADL training. Outcome measures were assessed at baseline and 4 wk.

Outcome Measures SCIM; ROM.

1.     A statistically significant improvement was observed in the total score of SCIM III self-care category and 2 of the 6 self-care category variables (Bathing upper body and Grooming) (p<0.05).

2.     ROM improved significantly when comparing pre- and post-assessments for 4 out of 5 ADL tasks (eating, drinking, spoon and sponge exercises) (p<0.05).

3.     No significant difference was observed pre and post assessment for the comb exercise (p>0.05).
Foldes et al., 2015

USA

Post Test

N=3

Population: Mean Age: 28 yr; Gender: males=3, females=0; Level of Injury: C2=1, C5=2; Severity of Injury: AIS A=2, AIS B=1, Unspecified=2.

Intervention: Patients with complete hand paralysis participated in a virtual hand grasping task. The virtual stop-motion hand was projected onto a screen and was controlled by the patient’s sensorimotor rhythms (SMRs). The SMRs were utilised via magnetoencephalography. Patients were asked to grasp or rest the virtual hand and were required to hold the position for a set time depending on difficulty level of the trial. Patients were also asked to attempt grasping and resting their own paralysed hand during each virtual hand trial. The intervention consisted of 200 trials (75% grasp, 25% rest) in a pseudorandom order with a 1min break after every 20 trials. Trials were also broken down into segments of 50 trials for analysis purposes. Assessments were performed at baseline and during each trial through to post-treatment.

Outcome Measures: Grasp success rate, SMR modulation, time to successful grasp.

1.     Overall grasp success rates varied between 62 and 64% with success rate significantly better than chance for each patient (p<0.001).

2.     Although grasp success rates improved after breaks between trials, the success rate was not significantly different when compared to trials before breaks (p=0.22).

3.     Success rates were also significantly greater than chance during grasp-only and rest-only trials (both p<0.001).

4.     Two patients demonstrated a significant increase in their ability to modulate their SMRs by 14.9pp and 15.0pp (both p<0.05) from baseline to post-treatment. The remaining patient did not exhibit any significant improvement in modulating SMRs.

5.     ANOVA analyses revealed a significant interaction between patient and session-segment (p<0.001) and a significant main effect of session-segment (p<0.001).

6.     Patients took an average of 1.96sec to complete a successful grasp, indicating that grasping using SMRs had been learnt quickly.
Robinson et al., 2014

UK

Post-Test

N=10

Population: Tetraplegic Group (n=5): Mean age: 39±6 yr; Gender: males=5, females=0; Level of injury: C5=4, C5/6=1; Mean time since injury: 17.6 yr; Severity of Injury: AIS A=3, AIS B=1, AIS C=1.

Control group (n=5): Mean age: 38±7 yr; Gender: males=5, females=0.

Intervention: Aiming movements were performed in two directions (20 cm away or toward), with or without vision with a ball transfer unit by both SCI patients and age-matched neurotypical controls. Trials that contained a sub-movement phase (i.e., discontinuity in velocity, acceleration or jerk) were identified.

Outcome Measures: Kinematic variables, Frequency and distribution (velocity, acceleration or jerk discontinuity), Amplitude and duration of sub-movements.

1.     The percentage of trials containing a sub-movement did not differ significantly between the tetraplegic and control groups (p>0.1).

2.     For % of type 3 sub-movements, there was a significant for direction (p<0.05), indicating that both groups made more type 3 sub-movement corrections when aiming away than toward the body.

3.     A significant effect was shown in direction for movement time (p<0.05) and a condition × direction × group interaction for both movement time (p<0.01) and peak velocity (p<0.05).

4.     Peak acceleration indicated significance for group and direction (p<0.02).

5.     Primary movement amplitude was greater when aiming away from than toward the body (p<0.05); this difference was somewhat larger in the vision than no vision condition (p<0.05).

6.     Amplitude revealed significance of group, with tetraplegics making larger corrections than controls (p<0.05).

7.     No significant for duration of corrective sub-movements between groups (p=0.08).

8.     Magnitude of spatial variability at peak velocity in sub-movement trials showed significance in group (p<0.05) as well as condition x direction x group interaction (p<0.05).

9.     Both groups made a greater percentage of functional than non-functional corrections when aiming toward, irrespective of vision (p<0.01).
Scott & Vare 2015

Australia

Post Test

N=16

Population: SCI population (n=11): Mean age: 37.5 yr; Level of Injury: C1-C4=5, C5-C6=6; Severity of Injury: AIS A=11.

Non-injured group (n=5): Mean age: 29.8 yr.

Intervention: Both the SCI and non-injured groups completed target matching tasks using a user command controller triggered by head position to manipulate a virtual hand representation. Participants using head movements matched the virtual hand to different targets. There were 10 targets split between the trials, some of which had different locations or sizes compared to each other. Additionally, the speed of the virtual hand was altered in four speed increments progressively throughout the experiment with a low of speed 1 (18 on-screen units/s) to a high of speed 4 (196 on-screen units/s).

Outcome Measures: Absolute performance on task matching (time to complete (TTC)), Efficacy of completion on task matching (integral of the error (IOE)), Ability to issue appropriate commands using the virtual hand (percentage of errors (POE)).

1.     The non-injured participants had significantly faster TTC scores than the SCI participants on completing Targets 3 and 4 (p>0.05).

2.     Additionally, high cervical participants were found to have significantly slower TTC scores than the mid cervical group (p<0.05).

3.     The high cervical group had significantly higher IOE than the middle cervical group and the non-injured participants for Targets 3 and 4 (p<0.05).

4.     Non-injured participants had a significantly lower POE than those with SCI in completing Targets 3 and 4 (p<0.05).

5.     On examination of TTC, IOE and POE for Targets 5 and 6, no significant differences were found between SCI and non-injured participants (p>0.05).

6.     There was a significant increase in the TTC for Target 8 for SCI participants over non-injured participants (p<0.05).

7.     There was a significant increase in IOE for Target 7 by SCI participants when compared to non-injured participants (p<0.05).

8.     There was a significant increase in the POE commands for Target 7 and Target 8 for SCI participants compared to controls(p<0.05).

9.     Non-injured participants were significantly faster than SCI participants in completing Target 10 (p<0.05), but there was no significant difference between the two groups for Target 9 (p>0.05).

10.   For speeds 1, 2, and 3, TTC scores were significantly lower for SCI participants (p<0.05).

11.   For IOE scores, non-injured participants had higher scores at speeds 1 and 3 compared to SCI participants (p<0.05).

12.   For POE scores, non-injured participants were scored significantly lower than the SCI participants at all four speeds (p<0.05).

Discussion

In subacute and chronic stroke patients, improvements in upper limb function with virtual reality have been demonstrated; however, the evidence of its application in spinal cord injury is still very limited. The small number of studies presented here demonstrate that virtual reality interventions produce similar results to conventional therapy for upper limb function. A minority of studies, demonstrated significant improvement in aspects of hand function such as dexterity, coordination, and grip, as well as, specific activities of daily living. While the results of these studies are promising, they are rather preliminary. In this sense, virtual reality should not replace conventional therapy, however, it may be well suited as a supplement. The incorporation of virtual reality as a rehabilitation supplement has been shown to improve several motivating and social factors including perceived control, curiosity, exploration, imagination, cooperation, competition and social interaction (Lohse et al. 2013). Moreover, virtual reality may provide a more engaging treatment by allowing patients to interact with virtual objects in a variable environment selected by themselves (e.g. games, characters or levels). In turn, this may increase motivation and subsequently increase the dose of therapy received. However, as outlined by Prasad and colleagues, future research should focus on: (1) comparing virtual reality systems to conventional therapy with randomized controlled trials in a larger population, (2) development of telerehabilitation programs to compliment virtual reality intervention, and (3) efficacy of virtual reality systems and types of exercises included.

Conclusion

There is level 1b evidence (from two randomized controlled trials: Prasad et al. 2018, Dimbwadyo-Terrer et al. 2015) that virtual reality interventions (Nintendo Wii & Toyra) produce similar results to conventional therapy in upper limb function.

There is level 2 evidence (from one randomized controlled trial: Dimbwadyo-Terrer et al. 2013) that a virtual reality intervention (Toyra) significantly improves dexterity, coordination and grip functions in comparison to conventional therapy.

There is level 2 evidence (from one prospective controlled trial: Dimbwadyo et al. 2015) that a virtual reality intervention (Cyber Touch) produces similar results to conventional therapy and clinically improves dexterity, coordination and grip, although, not significantly.

There is level 4 evidence (from one pre-post test: Seanez-Gonzalez et al. 2016) that use of an interactive body machine interface in patients with high level SCI improves upper-body movement ability and stimulates structural brain changes.

There is level 4 evidence (from one pre-post test: Dimbwadyo et al. 2015) that conventional therapy complimented with virtual reality training (Toyra) for activities of daily living significantly improves self-care scores and range of motion in tasks related to eating, upper body bathing and grooming.

There is level 4 evidence (from one post-test: Foldes et al. 2015) that patients with complete hand paralysis can learn to significantly modulate their sensorimotor rhythms using a virtual hand task over time.

Rehabilitation using virtual reality interventions produces similar results to conventional therapy and may help to improve hand function, as well as activities of daily living, through an engaging platform as a supplement to conventional therapy.

Chapter Downloads
Patient Handouts
Outcome Measures
Toolkits and CPGs
Videos
Active Clinical Trials