Virtual Reality – Lower Extremity

One high quality RCT (Kim et al., 2009) investigated the effect of VR-based training on balance in patients with chronic stroke. This high quality RCT randomized patients to receive VR balance/gait training using the IREX system or no VR training; both groups received conventional physical therapy balance training. Balance was measured by the Berg Balance Scale (BBS) and the Balance Performance Monitor system (BPM, static balance: mean, sway area, sway path, maximal velocity; dynamic balance: anterior-posterior, medio-lateral) at post-treatment (4 weeks). Significant between-group differences were found in balance (BBS, BPM dynamic balance: anterior-posterior, medio-lateral), favoring VR balance/gait training using the IREX system vs. no VR training.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR balance/gait training using the IREX system is more effective than a comparison intervention (no VR training) in improving balance in patients with chronic stroke.
Note: This may in part be due to greater treatment time.

Two high quality RCTs (You et al., 2005; Kim et al., 2009) investigated the effect of VR-based training on functional mobility in patients with chronic stroke.

The first high quality RCT (You et al., 2005) randomized patients to receive VR gait training using the IREX system or no treatment. Functional mobility was measured by the Modified Motor Assessment Scale and the Functional Ambulation Category Test at post-treatment (4 weeks). Significant between-group differences were found in both measures of functional mobility, favoring VR gait training using the IREX system vs. no treatment.

The second high quality RCT (Kim et al., 2009) randomized patients to receive VR balance/gait training using the IREX system or no VR training; both groups received conventional physical therapy balance training. Functional mobility was measured by the Modified Motor Assessment Scale at post-treatment (4 weeks). Significant between-group differences were found, favoring VR balance/gait training using the IREX system vs. no VR training.

Conclusion: There is strong evidence (Level 1a) from two high quality RCTs that VR gait training using the IREX system is more effective than comparison interventions (no treatment, no additional VR training) in improving functional mobility in patients with chronic stroke.

One high quality RCT (Kim et al., 2009) investigated the effect of VR-based training on gait parameters in patients with chronic stroke. This high quality RCT randomized patients to receive VR balance/gait training using the IREX system or no VR training; both groups received conventional physical therapy balance training. Gait parameters (cadence, step time, swing time, stance time, single support time, double support time, step length, stride length) were measured by the GAITRite system at post-treatment (4 weeks). Significant between-group differences in three gait parameters (cadence, step time, step length) were found, favoring VR balance/gait training using the IREX system vs. no VR training.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR balance/gait training using the IREX system is more effective than a comparison intervention (no VR training) in improving gait parameters in patients with chronic stroke.

One high quality RCT (Kim et al., 2009) investigated the effect of VR-based training on walking speed in patients with chronic stroke. This high quality RCT randomized patients to receive VR balance/gait training using the IREX system or no VR training; both groups received conventional physical therapy balance training. Walking speed was measured by the 10-meter walking test at post-treatment (4 weeks). Significant between-group differences were found, favoring VR balance/gait training using the IREX system vs. no VR training.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR balance/gait training using the IREX system is more effective than a comparison intervention (no additional VR training) in improving walking speed in patients with chronic stroke.

One fair quality RCT (Yom et al., 2015) investigated the effect of VR-based training on gait parameters in patients with chronic stroke. This fair quality RCT randomized patients to receive VR ankle training or no training; both groups received conventional physical therapy. Gait parameters were measured by the GAITRite system (velocity, cadence, step length, stride length, stance time percentage, swing time percentage, double limb support percentage) at post-treatment (6 weeks). Significant between-group differences were found in all gait parameters, favoring VR ankle training vs. no training.

Conclusion: There is limited evidence (Level 2a) from one fair quality RCT that VR ankle training is more effective than no training in improving gait parameters in patients with chronic stroke.

One fair quality RCT (Yom et al., 2015) investigated the effect of VR-based training on mobility in patients with chronic stroke. This fair quality RCT randomized patients to receive VR ankle training or no training; both groups received conventional physical therapy. Mobility was measured by the Timed Up and Go test at post-treatment (6 weeks). Significant between-group differences were found, favoring VR ankle training vs. no training.

Conclusion: There is limited evidence (Level 2a) from one fair quality RCT that VR ankle training is more effective than no training in improving mobility in patients with chronic stroke.

One fair quality RCT (Yom et al., 2015) investigated the effect of VR-based training on spasticity in patients with chronic stroke. This fair quality RCT randomized patients to receive VR ankle training or no training; both groups received conventional physical therapy. Spasticity was measured by the Modified Ashworth Scale and the Tardieu Scale at post-treatment (6 weeks). Significant between-group differences were found for both measures of spasticity, favoring VR ankle training vs. no training.

Conclusion: There is limited evidence (Level 2a) from one fair quality RCT that VR ankle training is more effective than no training in reducing spasticity in patients with chronic stroke.

One high quality RCT (Lee et al., 2014) investigated the effect of VR-based training on balance in patients with chronic stroke. This high quality RCT randomized patients to receive VR postural control training or no VR training; both groups received physical therapy. Balance was measured by the Berg Balance Scale at post-treatment (4 weeks). No significant between-group differences were found.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR postural control training is not more effective than no VR training in improving balance in patients with chronic stroke.

One high quality RCT (Park et al., 2013) investigated the effect of VR-based training on functional ambulation in patients with chronic stroke. This high quality RCT randomized patients to receive VR postural control training or no VR training; both groups received conventional physical therapy. Functional ambulation was measured by the Functional Ambulation Profile at post-treatment (4 weeks) and at follow-up (1 month). No significant between-group differences were found at either time point.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR postural control training is not more effective than no VR training in improving functional ambulation in patients with chronic stroke.

Two high quality RCTs (Park et al., 2013; Lee et al., 2014) investigated the effect of VR-based training on gait parameters in patients with chronic stroke.

The first high quality RCT (Park et al., 2013) randomized patients to receive VR postural control training or no VR training; both groups received conventional physical therapy. Gait parameters (velocity, cadence, step/stride length of paretic/non-paretic limbs) were measured by the GAITRite system at post-treatment (4 weeks) and at follow-up (1 month). No significant between-group differences were found at post-treatment. At follow-up, a significant between-group difference was found for stride length (paretic and non-paretic limbs) only, favoring VR postural control training vs. no VR training.

The second high quality RCT (Lee et al., 2014) randomized patients to receive VR postural control training or no VR training; both groups received physical therapy. Gait parameters (velocity, cadence, step length/stride length of paretic/non-paretic limbs) were measured by the GAITRite system at post-treatment (4 weeks). Significant between-group differences were found for most gait parameters (velocity, step length/stride length of paretic/non-paretic limbs), favoring VR postural control training vs. no VR training.

Conclusion: There is conflicting evidence (Level 4) between two high quality RCTs regarding the effect of VR postural control training on gait parameters in patients with chronic stroke. While one high quality RCT found that VR postural control training was not more effective than a comparison intervention (physical therapy provided for 60-minute sessions, 5 times/week for 4 weeks); a second high quality RCT found that VR postural control training was more effective than a comparison intervention (physical therapy provided 30-minute session, 3 times/week for 4 weeks).
Note: Both studies provided the intervention at a comparable intensity and duration; accordingly, differences in the intensity of the comparison intervention between the two studies could account for differences in findings.

One high quality RCT (Lee et al., 2014) investigated the effect of VR-based training on mobility in patients with chronic stroke. This high quality RCT randomized patients to receive VR postural control training or no VR training; both groups received physical therapy. Mobility was measured by the Timed Up and Go test at post-treatment (4 weeks). No significant between-group differences were found.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR postural control training is not more effective than no VR training in improving mobility in patients with chronic stroke.

One high quality RCT (Park et al., 2013) investigated the effect of VR-based training on walking speed in patients with chronic stroke. This high quality RCT randomized patients to receive VR postural control training or no VR training; both groups received conventional physical therapy. Walking speed was measured by the 10-meter walking test at post-treatment (4 weeks) and at follow-up (1 month). No significant between-group differences were found at either time point.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR postural control training is not more effective than no VR training in improving waking speed in patients with chronic stroke.

One high quality RCT (Mirelman et al., 2008) investigated the effect of VR-based training on gait parameters in patients with chronic stroke. This high quality RCT randomized patients to receive VR + the robotic Rutgers Ankle Rehabilitation System (RARS) or RARS only. Gait parameters (number of steps/day, average daily distance walked, velocity, cadence, stride length, longest consecutive locomotion period, longest distance traveled) were measured by the Patient Activity Monitor at post-treatment (4 weeks). A significant between-group difference was found in three gait parameters (number of steps/day, average daily distance, velocity) at post-treatment, favoring VR + RARS vs. RARS alone.

A further analysis (Mirelman et al., 2010) reported on gait parameters (change in ankle moment at push-off barefoot/shoes on; ankle power at push-off barefoot/shoes on; ankle range of motion barefoot/shoes on; knee flexion range of motion during stance/swing barefoot/shoes on; hip flexion range of motion during swing barefoot/shoes on; onset of push-off; self-selected velocity) measured by the Vicon Motion Capture System + Plug-In Gait model at post-treatment (4 weeks) and at follow-up (3 months). Significant between-group differences were found for some gait parameters at post-treatment (change in ankle moment at push-off – barefoot; ankle power at push-off – barefoot; knee flexion range of motion during stance/swing – barefoot; onset of push-off; self-selected velocity); and at follow up (ankle power at push-off – barefoot; ankle range of motion – barefoot; knee flexion ROM of the affected side during stance/swing – barefoot; onset of push-off; self-selected velocity), favoring VR + RARS vs. RARS alone.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR + robotic lower extremity training is more effective than a comparison intervention (robotic lower extremity training alone) in improving gait parameters in patients with chronic stroke.

One high quality RCT (Mirelman et al., 2008) investigated the effect of VR-based training on walking endurance in patients with chronic stroke. This high quality RCT randomized patients to receive VR + the robotic Rutgers Ankle Rehabilitation System (RARS) or robotics RARS alone. Walking endurance was measured by the 6-Minute Walk Test at post-treatment (4 weeks) and at follow-up (3 months). No significant between-group differences were found at either time point.
Note: However, subgroup analyses of patients with moderate walking speed at baseline revealed significant between-group differences at post-treatment and at follow-up, favoring VR + RARS vs. RARS alone.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR + robotic lower extremity training is not more effective than a comparison intervention (robotic lower extremity training alone) in improving walking endurance in patients with chronic stroke.

One high quality RCT (Mirelman et al., 2008) investigated the effect of VR-based training on walking speed in patients with chronic stroke. This high quality RCT randomized patients to receive VR + the robotic Rutgers Ankle Rehabilitation System (RARS) or RARS alone. Self-selected walking speed over 7 meters was measured at post-treatment (4 weeks) and at follow-up (3 months). No significant between-group differences were found at either time point.
Note: However, subgroup analyses of patients with moderate walking speed at baseline revealed significant between-group differences at post-treatment and at follow-up, favoring VR + RARS vs. RARS alone.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR + robotics lower extremity training is not more effective than a comparison intervention (robotics lower extremity training alone) in improving walking speed in patients with chronic stroke.

One high quality RCT (Llorens et al., 2015) and one quasi-experimental design study (Llorens et al., 2012) investigated the effect of VR-based training on balance in patients with chronic stroke.

The high quality RCT (Llorens et al., 2015) randomized patients to receive VR stepping exercises or no VR training; both groups received conventional physical therapy. Balance was measured by the Berg Balance Scale, Brunel Balance Assessment, and Tinetti Performance Oriented Mobility Assessment (balance, gait subtests) at post-treatment (4 weeks). Significant between-group differences were found on 2 of 3 balance measures (Berg Balance Scale, Brunel Balance Assessment) at post-treatment, favoring VR stepping exercises vs. no VR training.

The quasi-experimental design study (Llorens et al., 2012) assigned patients to receive VR stepping exercises. Balance was measured by the Berg Balance Scale, Brunel Balance Assessment, and Tinetti Performance Oriented Mobility Assessment (balance subtest) at baseline, at post-treatment (4-5 weeks) and at follow-up (8 weeks). Significant improvements were found on 2 of 3 balance measures (Berg Balance Scale, Brunel Balance Assessment) at post-treatment; differences did not remain significant at follow-up.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR stepping exercises are more effective than no VR training in improving balance in patients with chronic stroke. A quasi-experimental design study also found improvements in balance following VR stepping exercises.

One high quality RCT (Llorens et al., 2015) investigated the effect of VR-based training on walking speed in patients with chronic stroke. This high quality RCT randomized patients to receive VR stepping exercises or no VR training; both groups received conventional physical therapy. Walking speed was measured by the 10-Meter Walking Test at post-treatment (4 weeks). Significant between-group differences were found, favoring VR stepping exercises vs. no VR training.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR stepping exercises are more effective than no VR training in improving walking speed in patients with chronic stroke.

Three high quality RCTs (Kang et al., 2012; Cho & Lee, 2013; Cho & Lee, 2014) and three fair quality RCTs (Jaffe et al., 2004; Yang et al., 2011; Kim et al., 2015) investigated the effect of VR-based training on balance in patients with chronic stroke.

The first high quality RCT (Kang et al., 2012) randomized patients to receive VR treadmill gait training with optic flow, conventional treadmill gait training or physical exercises; all groups received conventional physical therapy. Balance was measured by the Functional Reach Test at post-treatment (4 weeks). Significant between-group differences were found, favoring VR treadmill gait training with optic flow vs. physical exercises; there were no significant differences between VR treadmill gait training and conventional treadmill gait training.
Note: Significant between-group differences were also found favoring conventional treadmill gait training vs. physical exercises.

The second high quality RCT (Cho & Lee, 2013) randomized patients to receive VR treadmill gait training or conventional treadmill gait training; both groups received conventional rehabilitation. Balance was measured by the Berg Balance Scale (BBS) at post-treatment (6 weeks). Significant between-group differences were found, favoring VR treadmill gait training vs. conventional treadmill gait training.

The third high quality RCT (Cho & Lee, 2014) randomized patients to receive VR treadmill gait training or conventional treadmill gait training; both groups received conventional rehabilitation. Dynamic balance was measured by the BBS, and static balance was measured by the Good Balance SystemTM(postural sway: anteroposterior, mediolateral, velocity moment) at post-treatment (6 weeks). Significant between-group differences were found for dynamic balance only, favoring VR treadmill gait training vs. conventional treadmill gait training.

The first fair quality RCT (Jaffe et al., 2004) randomized patients to receive VR treadmill gait training or non-VR gait training. Balance was measured by a non-standardized 7-item balance test adapted from the Performance-Oriented Assessment of Mobility and the Physical Performance Test, at post-treatment (2 weeks) and at follow-up (4 weeks). No significant between-group differences were found at either time point.

The second fair quality RCT (Yang et al., 2011) randomized patients to receive VR treadmill gait training or conventional treadmill gait training. Balance in quiet stance (center of pressure [COP] displacement in medial-lateral direction [COPML], posterior/anterior direction [COPAP], total path excursion [COPE], sway area [COPA], and symmetry index [SI]), sit-to-stand transfer (COPML, COPAP, COPE, COPA, SI /paretic limb), and level walking (paretic limb stance time, step number, contact area) was measured by a pressure mat system at post-treatment (3 weeks). There was a significant between-group difference in only one measure of balance (quiet stance: COPML) favoring VR treadmill gait training vs. conventional treadmill gait training.

The third fair quality RCT (Kim et al., 2015) randomized patients to receive VR treadmill gait training or time-matched physical therapy; both groups received conventional physical therapy. Static balance was measured using the Balancia Software system and Wii FitTM balance board to assess postural sway path length (anterior/posterior, mediolateral, total) and postural sway speed at post-treatment (4 weeks). Significant between-group differences were found in both measures of static balance, favoring VR treadmill gait training vs. time-matched physical therapy.

Conclusion: There is strong evidence (Level 1a) from three high quality RCTs and one fair quality RCT that VR treadmill gait training is more effective than comparison interventions (conventional treadmill gait training, physical exercises, and time-matched physical therapy) in improving balance in patients with chronic stroke.
Note: One of the high quality RCTs found that VR treadmill training was not more effective than conventional treadmill training; two fair quality RCTs found that VR treadmill gait training was not more effective than comparison interventions (conventional treadmill gait training, non-VR gait training) in improving balance in patients with chronic stroke. Differences among studies including outcome measures used, duration of interventions and comparison interventions could account for difference in findings.

Two high quality RCTs (Yang et al., 2008; Kim et al., 2016) and one fair quality RCT (Jung et al., 2012) investigated the effect of VR-based training on balance confidence in patients with chronic stroke.

The first high quality RCT (Yang et al., 2008) randomized patients to receive VR treadmill gait training or conventional treadmill gait training. Balance confidence was measured by the Activities-Specific Balance Confidence (ABC) Scale at post-treatment (3 weeks) and at follow-up (1 month). No significant between-group differences were found at either time point.

The second high quality RCT (Kim et al., 2016) randomized patients to receive VR treadmill gait training, time-matched community ambulation training, or no additional training; all groups received physical therapy. Balance confidence was measured by the ABC Scale at post-treatment (4 weeks). There were no significant differences between VR treadmill training and community ambulation training. Significant between-group differences were found in favour of VR treadmill gait training vs. no additional training.
Note: Significant between-group differences were also found favoring community ambulation training vs. no additional training.

The fair quality RCT (Jung et al., 2012) randomized patients to receive VR treadmill gait training or conventional treadmill gait training. Balance confidence was measured by the ABC Scale at post-treatment (3 weeks). Significant between-group differences were found, favoring VR treadmill gait training vs. conventional treadmill gait training.

Conclusion: There is strong evidence (Level 1a) from two high quality RCTs that VR treadmill gait training is not more effective than comparison interventions (conventional treadmill gait training, community ambulation training) in improving balance confidence in patients with chronic stroke.
Note: However, one of the high quality RCTs found that VR treadmill gait training was more effective than no additional training; a fair quality RCT found that VR treadmill training was more effective than conventional treadmill training.

Three high quality RCTs (Cho & Lee, 2013; Cho & Lee, 2014; Kim et al., 2016) and one fair quality RCT (Jaffe et al., 2004) investigated the effect of VR-based training on gait parameters in patients with chronic stroke.

The first high quality RCT (Cho & Lee, 2013) randomized patients to receive VR treadmill gait training or conventional treadmill gait training; both groups received conventional rehabilitation. Gait parameters (velocity, cadence, step length, stride length, single limb support) were measured by the GAITRite system at post-treatment (6 weeks). Significant between-group differences were found for 2 of 5 gait parameters (velocity, cadence), favoring VR treadmill gait training vs. conventional treadmill gait training.

The second high quality RCT (Cho & Lee, 2014) randomized patients to receive VR treadmill gait training or conventional treadmill gait training; both groups received conventional rehabilitation. Gait parameters (gait speed, cadence, step length, stride length and single/double limb support) were measured by the GAITRite system at post-treatment (6 weeks). Significant between-group differences were found for all gait parameters, favoring VR treadmill gait training vs. conventional treadmill gait training.

The third high quality RCT (Kim et al., 2016) randomized patients to receive VR treadmill gait training, time-matched community ambulation training, or no additional training; all groups received physical therapy. Gait parameters (velocity, cadence, paretic step and stride lengths) were measured by the GAITRite system at post-treatment (4 weeks). No significant between-group differences were found between any group.

The fair quality RCT (Jaffe et al., 2004) randomized patients to receive VR treadmill gait training or non-VR gait training. Gait parameters (at fast and comfortable speed: velocity, cadence, stride length, step length) were measured by the Stride Analyzer gait analysis system, at post-treatment (2 weeks) and at follow-up (4 weeks). Significant between-group differences were found for 2 measures (fast speed: velocity, stride length) at post-treatment, favoring VR treadmill gait training vs. non-VR gait training. Differences did not remain significant at follow-up.

Conclusion: There is conflicting evidence (Level 4) regarding the effect of VR treadmill gait training on gait parameters in patients with chronic stroke. One high quality RCT found VR treadmill gait training to be more effective than conventional treadmill gait training, whereas another high quality RCT found it was not more effective than community ambulation training or no training. A third high quality RCT and a fair quality RCT found mixed results when comparing VR treadmill gait training with conventional treadmill gait training and non-VR gait training (respectively).

One high quality RCT (Yang et al., 2008) investigated the effect of VR-based training on functional ambulation in patients with chronic stroke. This high quality RCT randomized patients to receive VR treadmill gait training or conventional treadmill gait training. Functional ambulation was measured by the Community Walk Test (CWT) and the Walking Ability Questionnaire (WAQ) at post-treatment (3 weeks) and at follow-up (1 month). There was a significant between-group difference in one measure of functional ambulation (CWT) at post-treatment, favoring VR treadmill gait training vs. conventional treadmill gait training; this difference did not remain significant at follow-up. However, there was a significant between-group difference in the other measure of functional ambulation (WAQ) at follow-up, favoring VR treadmill gait training vs. conventional treadmill gait training.

Conclusion: There is moderate evidence (Level 1b) from one high quality RCT that VR treadmill gait training is more effective than a comparison intervention (conventional treadmill gait training) in improving functional ambulation in patients with chronic stroke.

Four high quality RCTs (Kang et al., 2012; Cho & Lee, 2013; Cho & Lee, 2014; Kim et al., 2016) and one fair quality RCT (Jung et al., 2012) investigated the effect of VR-based training on mobility in patients with chronic stroke.

The first high quality RCT (Kang et al., 2012) randomized patients to receive VR treadmill gait training with optic flow, conventional treadmill gait training or physical exercises; all groups received conventional physical therapy. Mobility was measured by the Timed Up and Go test (TUG) at post-treatment (4 weeks). Significant between-group differences were found, favoring VR treadmill gait training with optic flow vs. conventional treadmill gait training; and favoring VR treadmill gait training with optic flow vs. physical exercises.
Note: There were no significant differences between conventional treadmill gait training and physical exercises.

The second high quality RCT (Cho & Lee, 2013) randomized patients to receive VR treadmill gait training or conventional treadmill gait training; both groups received conventional rehabilitation. Mobility was measured by the TUG at post-treatment (6 weeks). Significant between-group differences were found, favoring VR treadmill gait training vs. conventional treadmill gait training.

The third high quality RCT (Cho & Lee, 2014) randomized patients to receive VR treadmill gait training or conventional treadmill gait training; both groups received conventional rehabilitation. Mobility was measured by the TUG at post-treatment (6 weeks). Significant between-group differences were found, favoring VR treadmill vs. conventional treadmill gait training.

The forth high quality RCT (Kim et al., 2016) randomized patients to receive VR treadmill gait training, time-matched community ambulation training, or no additional training; all groups received physical therapy. Mobility was measured by the TUG at post-treatment (4 weeks). Significant between-group differences were found, favoring VR treadmill gait training vs. no additional training. There were no significant differences between VR treadmill training and community ambulation training.
Note: There were no significant differences between community ambulation training and no additional training.

The fair quality RCT (Jung et al., 2012) randomized patients to receive VR treadmill gait training or conventional treadmill gait training. Mobility was measured by the TUG at post-treatment (3 weeks). Significant between-group differences were found, favoring VR treadmill gait training vs. conventional treadmill gait training.

Conclusion: There is strong evidence (Level 1a) from four high quality RCTs and one fair quality RCT that VR treadmill gait training is more effective than comparison interventions (conventional treadmill gait training, physical exercises, no additional training) in improving mobility in patients with chronic stroke.
Note: Results from one high quality RCT showed that VR treadmill training was not more effective than community ambulation training.

One fair quality RCT (Jaffe et al., 2004) investigated the effect of VR-based training on obstacle clearance performance in patients with chronic stroke. This fair quality RCT randomized patients to receive VR treadmill gait training or non-VR gait training. Obstacle clearance performance was measured by the height of the longest obstacle successfully negotiated at post-treatment (2 weeks) and at follow-up (4 weeks). No significant between-group differences were found at either time point.

Conclusion: There is limited evidence (Level 2b) from one fair quality RCT that VR treadmill gait training is not more effective than a comparison intervention (non-VR gait training) in improving obstacle clearance performance in patients with chronic stroke.

Two high quality RCTs (Kang et al., 2012; Kim et al., 2016) and one fair quality RCT (Jaffe et al., 2004) investigated the effect of VR-based training on walking endurance in patients with chronic stroke.

The first high quality RCT (Kang et al., 2012) randomized patients to receive VR treadmill gait training with optic flow, conventional treadmill gait training or physical exercises; all groups received conventional physical therapy. Walking endurance was measured by the 6-Minute Walk Test (6MWT) at post-treatment (4 weeks). Significant between-group differences were found, favoring VR treadmill gait training with optic flow vs. conventional treadmill gait training; and favoring VR treadmill gait training with optic flow vs. physical exercises.
Note: There were no significant differences between conventional treadmill gait training and physical exercises.

The second high quality RCT (Kim et al., 2016) randomized patients to receive VR treadmill gait training, time-matched community ambulation training, or no additional training; all groups received physical therapy. Walking endurance was measured by the 6MWT at post-treatment (4 weeks). No significant differences were found between VR treadmill gait training and community ambulation training, or between VR treadmill gait training and no training.
Note: There was a significant between-group difference in walking endurance, in favour of community ambulation training vs. no training.

The fair quality RCT (Jaffe et al., 2004) randomized patients to receive VR treadmill gait training or non-VR gait training. Walking endurance was measured by the 6MWT at post-treatment (2 weeks) and at follow-up (4 weeks). No significant between-group differences were found at either time point.

Conclusion: There is conflicting evidence (Level 4) regarding the effect of VR treadmill gait training on walking endurance in patients with chronic stroke. One high quality RCT found that VR treadmill training was more effective than conventional treadmill training and physical exercises, whereas a second high quality RCT and a fair quality RCT found it was not more effective than no training and non-VR gait training.

Two high quality RCTs (Yang et al., 2008; Kang et al., 2012) investigated the effect of VR-based training on walking speed in patients with chronic stroke.

The first high quality RCT (Yang et al., 2008) randomized patients to receive VR treadmill gait training or conventional treadmill gait training. Walking speed was measured by the 10-meter walking test at post-treatment (3 weeks) and at follow-up (1 month). Significant between-group differences were found at post-treatment, favoring VR treadmill gait training vs. conventional treadmill gait training; differences did not remain significant at follow-up.

The second high quality RCT (Kang et al., 2012) randomized patients to receive VR treadmill gait training with optic flow, conventional treadmill gait training or physical exercises; all groups received conventional physical therapy. Walking speed was measured by the 10-meter walking test at post-treatment (4 weeks). Significant between-group differences were found, favoring VR treadmill gait training with optic flow vs. conventional treadmill gait training; and favoring VR treadmill gait training with optic flow vs. physical exercises.
Note: There were no significant differences between conventional treadmill gait training and physical exercises.

Conclusion: There is strong evidence (Level 1a) from two high quality RCTs that VR treadmill gait training is more effective than comparison interventions (conventional treadmill gait training, physical exercises) in improving walking speed in patients with chronic stroke.

One quasi-experimental design study (Cikajlo et al., 2012) investigated the effect of VR-based training on balance in patients with stroke. This quasi-experimental design study allocated patients with subacute/chronic stroke to receive VR balance training or non-VR balance training; both groups received conventional rehabilitation. Balance was measured by the Berg Balance Scale and timed stance (affected and non-affected limbs) at post-treatment (3 weeks). No significant between-group differences were found.

Conclusion: There is limited evidence (Level 2b) from one quasi-experimental design study that VR balance training is not more effective than a comparison intervention (non-VR balance training) in improving balance in patients with stroke.

One quasi-experimental design study (Cikajlo et al., 2012) investigated the effect of VR-based training on mobility in patients with stroke. This quasi-experimental design study allocated patients with subacute/chronic stroke to receive VR balance training or non-VR balance training; both groups received conventional rehabilitation. Mobility was measured by the Timed Up and Go Test at post-treatment (3 weeks). No significant between-group differences were found.

Conclusion: There is limited evidence (Level 2b) from one quasi-experimental design study that VR balance training is not more effective than a comparison intervention (non-VR balance training) in improving mobility in patients with stroke.

One fair quality RCT (Mao et al., 2015) investigated the effect of VR-based training on pelvis control in patients with stroke. This fair quality RCT randomized patients with acute/subacute stroke to receive VR body-weight supported treadmill training or over ground walking training. Pelvis control (tilt, obliquity, rotation) was measured by the Vicon Motion Capture System + Plug-in-Gait model at post-treatment (3 weeks). No significant between-group differences were found.

Conclusion: There is limited evidence (Level 2b) from one fair quality RCT that VR treadmill training is not more effective than a comparison intervention (over ground walking training) in improving pelvis control in patients with stroke.

One quasi-experimental design study (Cikajlo et al., 2012) investigated the effect of VR-based training on walking speed in patients with stroke. This quasi-experimental design study allocated patients with subacute/chronic stroke to receive VR balance training or non-VR balance training; both groups received conventional rehabilitation. Walking speed was measured by the 10 Meter Walk Test at post-treatment (3 weeks). No significant between-group differences were found.

Conclusion: There is limited evidence (Level 2b) from one quasi-experimental designs study that VR balance training is not more effective than a comparison intervention (non-VR balance training) in improving walking speed in patients with stroke.