Table 3 Summary of studies

Deutsch et al. (2004) [45]

Pilot study, V

Chronic stroke Experiment 1 – 69-year old, 10 months post-stroke (n = 1) Experiment 2 – 1-8 years post-stroke (n = 3), age not given

Exp 1- Right middle cerebral artery CVA Exp 2 – not specified

Exp 1) VR-based measures including power, torque, accuracy,strength and range of motion Exp 2) Walking speed; 6-minute walk test (6MWT); Berg Balance score (BBS); foot strength

Exp 1) One grade increase in the strength of ankle evertors. Increased accuracy on the VR simulations (from 58% to 88%). Exp 2 –Increase in strength increase in 2 to 4 muscle groups and increase in walking speed (80 ft/min to 95 ft/min).

Jaffe et al. (2004) [28]

Randomized controlled trial, II

Chronic stroke (3.8 ± 2.2 years post-stroke), 60.7 ± 2.3 years; virtual object group (n = 10); real object group (n = 10)

Not specified

Percentage improvement on: 1) Balance tests (natural stance, natural stance eyes closed, on toes, tandem stance, tandem stance eyes closed, left leg only, right leg only) 2) Walking tests (walking velocity, cadence and stride length at self-selected and fast walking) 3) Obstacle test 4) 6-minute walk test (6MWT)

1) Results related to the Balance tests are not reported. 2) Greater percent improvements in walking speed (20.5% vs. 12.2%) and stride length for fast walking with VR 3) Obstacle crossing showed greater percent improvements with virtual obstacle protocol. Greater than 95% retention in all key gait parameters at 2 week follow-up 4) 6MWT: No significant changes post-treatment observed in both groups.

You et al. (2005) [29]

Experimenter blind randomized study - RCT, II

Chronic stroke (VR group – n = 5, 54.6 ± 3.0 years, 18.2 ± 2.3 months post-stroke control group – n = 5, 54.6 ± 3.4 years, 19.4 ± 4.3 months post-stroke)

Thalamic hemorrhage (n =2) Corona radiata hemorrhage (n =3) Corona radiata infarct (n = 5)

1) fMRI – Laterality Index (LI) 2) Functional Ambulation Category (FAC) 3) Modified Motor Assessment Scale (MMAS)

1) Bilateral activity in the sensory-motor cortex as seen before treatment either disappeared or decreased on the ipsilateral side indicating cortical reorganization. 2) All patients in the VR group were able to achieve a change of at least one level on the FAC as opposed to the control group. 3) Significant changes in MMAS.

Betker et al. (2006) [46]

Single-subject design: case report, V

n = 3, case 1 – 20-year old case 2 – 58-year old post-stroke (onset not provided) case 3- 14-year old

Case 1- cerebellar tumour excision with ataxia Case 2 – cerebrovascular accident infarction Case 3 – closed TBI (case 1 and 3 will not be discussed in the paper)

1) Tasks that involved maintaining static and dynamic balance 2) CoP excursion and sway path 3) Number of falls

1) Successful completion of oscillating head rotations, trunk bending and trunk rotation tasks post-exercise as opposed to failure pre-exercise. 2) Variable findings on CoP excursion and CoP sway path with increase in excursion and sway path in some and decrease in other tasks. 3) 5 falls pre-exercise as opposed to 2 falls post-exercise

Fung et al. (2006) [10]

Feasibility study, V

Sub-acute stroke (n = 2, 49 & 61 years, 2 and 4.5 months post-stroke) Control – healthy elderly (n = 1, 64 years)

Not specified

Ability to complete the locomotor task within the time constraint and without any collisions with the virtual obstacles.

Both subjects as well as the control were able to increase the walking speed to complete level 1 in all VE’s. The subjects were also able to eventually complete level 2 despite a considerable reduction in gait speed while negotiating platform movements. Both subjects did not reach level 3 to avoid obstacles successfully. The precise increase or decrease in gait speed is not specified.

Flynn et al. (2007) [47]

Case report, V

Chronic stroke (n = 1, 17 months post-stroke, 76-year old)

Cerebro-vascular accident, area not specified

Only those concerned with balance and mobility are listed here: 1) BBS 2) Dynamic Gait Index (DGI) 3) Timed Up-and-Go (TUG) 4) 6MWT 5) Functional Reach test (FRT)

Only those concerned with balance and mobility are listed here: 1) BBS increased from 51/56 pretest to 54/56 post-test 2) DGI – 16 pre-test to 21post-test 3) TUG - 12.73 s pre to 11.68 s post 4) 6MWT–1282 ft pre to 1337 ft post 5) FRT 11.33 in pre to 11.50 in post

Yang et al. (2008) [30]

RCT, II

Chronic stroke: VR group (n = 11, 55.5 years; 5.9 years post-stroke) Control group (n = 9, 60.9 years, 6.1 years post-stroke)

Not specified

1) Walking speed 2) Community walking time 3) Activities-specific Balance Confidence scale (ABC) 4) Walking Ability Questionnaire (WAQ) score

1) Insignificant change in walking speed but a trend towards increase. 2) Significant improvement in community ambulation time (6.14 ± 5.53 s); 3) ABC scores (8.86 ± 10.10points); 4) WAQ scores (3.45 ± 5.11points); All gains retained at 1 month follow-up.

Dunning et al. (2008) [48]

Case report, V

Sub-acute stroke (n = 1, 51 years; 9 months post-stroke)

Ischaemic infarct in the posterior limb of the left internal capsule

1) Modified Emory Functional Ambulation Profile (mEFAP) 2) Lower extremity portion of the Fugl-Meyer (FM) scale 3) Gait velocity: self-selected & fast 4) Temporal distance gait parameters 5) Lower extremity joint kinematics and kinetics

1) mEFAP – 4.4 s (11.2%) decrease post-intervention 2) FM – 5 point (23.8%) increase with largest changes for obstacles (17.2%) and stairs (13.7%) 3) Gait velocity increase post-intervention -Slow – 0.29 m/s (27.7%) fast – 0.26 m/s (16.4%) 4) Temporal distance gait parameters such as step length, cadence, stance and swing times also showed minor improvements post-treatment. 5) Ankle plantarflexor moment during push off increased 29.5% and 13.3% for self-selected and fast walking speeds respectively Greater hip extension at push off, improved knee extension at heel strike and greater ankle plantarflexion at push-off.

Mirelman et al. (2009; 2010) [31;33]

Randomized controlled trial, II

Chronic stroke Robotic VR group (n = 9; 61.8 years, 37.7 months post-stroke), robotic group (n = 9; 61 years, 58.2 months post-stroke)

Not specified

1) 6MWT 2) Community-based walking as measured by the PAM (no. of steps/day, average daily distance walked, speed, cadence, walking strides, maximum walking speed, longest consecutive locomotion period in minutes and longest consecutive distance travelled). 3) Self selected walking speed (SSWS) 4) Joint kinetics – ankle moments during stance and pre-swing, knee flexor moment during stance and push-off, hip flexor moment at initial swing, power at the ankle, knee and hip joints 5) Joint kinematics – ROM of the ankle and hip joints during the gait cycle

1) 6MWT – 21% increase in the Robotic VR group; 0.5% increase in the robotic group 2) Community based ambulation – significant differences seen in distance walked, no. of steps per day, average speed and maximum speed. All changes were retained in the robotic VR group at 3 months follow –up. 3) SSWS –24% increase (from 0.65 to 0.81 m/s) in the VR group vs. 2% (0.67 – 0.68 m/s) in the NVR group. 4) Joint kinetics- Ankle moment (barefoot walking) – VR group (from 0.74 ± 0.24 Nm/kg to 0.90 ± 0.31 Nm/kg, 21%) vs. NVR group (0.68 ± 0.17 Nm/kg to 0.67 ± 0.08 Nm/kg, 1.5%). Ankle power (barefoot walking) – significant increase in the VR group (0.63 ± 0.28 W/kg to 0.91 ± 0.45 W/kg; 44%) as opposed to the NVR group (0.5 ± 0.27 W/kg to 0.52 ± 0.26 W/kg; 4%). Retention at follow-up was seen in the VR group. 5) Kinematics –Barefoot walking Ankle ROM – increase from 29.3 ± 7.4 degrees (19.5%) in the VR group and from 32.6 ± 13.4 to 36.7 ± 3.2 degrees (3.3%) in the NVR. Both changes were reported to be statistically significant. Knee ROM – significantly greater increases (stance-34%, swing – 15.7%) in the VR group as compared to the NVR group (stance – 7.2%, swing – 3.9%). Both ankle and knee ROM gains were preserved at follow-up. Onset of push-off – improved from 55% of gait cycle in both groups to 57.7% of the gait cycle in the VR group only.

Kim et al. (2009) [32]

RCT, II

Chronic stroke: Experimental group (n = 12; 52.4 years, 25.9 months post-stroke); Control group (n = 12; 52.4 years, 25.9 months post-stroke)

Intra-cranial hemorrhage in the thalamus, putamen, basal ganglia (n = 13) Infarcts in the deep cerebral white matter, basal ganglia, putamen and pons (n = 11)

1) BBS 2) Modified motor assessment scale 3) 10 m walk test 4) mean CoP sway area; sway path and maximal sway velocity; antero-posterior (AP) and medio-lateral (ML) sway angles 5) Temporal distance gait parameters: cadence, velocity, step time, stance time, swing time, single/double support time, step/stride length.

1,2,3) Significant improvement in scores 4) reduction in sway area and maximal CoP velocity; AP and ML sway angles increased 5) cadence, step time, step length showed significant increases

Walker et al. (2010) [43]

Pre-post design, IV

Sub-acute to chronic stroke (3 weeks to 1 year post-stroke) N = 6, 53.4 years (49-74 years)

Ischaemic stroke with left-side hemiparesis

1) FGA 2) BBS 3) Overground walking speed Also reported pre-post treadmill walking speed, treadmill walking duration and percentage body-weight supported

Significant increases in 1) FGA (30%); 2) BBS (10%); and 3) overground walking speed (38%, change of 0.19 m/s pre-post). Treadmill walking duration – pre -10 min, post- 19.83 min Treadmill walking speed – pre – 1.31mph, post – 1.7 mph Weight support (% Body weight) – pre- 31.67, post – 18.33.

Shin et al. (2010) [39]

Pre-post, III

Chronic stroke (Control group, n = 16, 60.7 ± 9.2 years, 71.5 ± 33.9 months post-stroke; Game exercise group, n = 16, 60.8 ± 7.5 years, 69.2 ± 36.4 months post-stroke)

Control group – 11 ischemic and 5 hemorrhagic stroke; Game exercise group – 10 ischemic and 6 hemorrhagic stroke

1) 10 m walk test 2) 6MWT

1) Significant improvements in gait speed (0.86 ± 0.32 m/s pre-training to 1.10 ± 0.34 m/s post-training) in the game-exercise group as compared to controls. 2) Significant improvements were also seen in distance walked in the 6MWT (225.87 ± 60.16 m pre-training to 268.79 ± 56.42 m post-training) and was significantly larger than that seen in control group.

Yang et al. (2011) [34]

Randomized controlled trial, II

Chronic stroke (control group, n = 7; 65.7 ± 5.9 years, 16.3 ± 10.4 months post-stroke Experimental group, n = 7, 56.3 ± 10.2, 17.0 ± 8.6 months post-stroke).

Not specified

1) Standing eyes open -Maximum CoP displacement in the ML (CoPML) & AP (CoPAP) directions, excursion (CoPE), sway area (CoPA), & bilateral limb-loading symmetrical index (SI). 2) Sit to stand (STS): those included above + CoP excursion for the paretic foot (CoPE/P). 3) Walking: Stance time of the paretic limb (ST/P), number of steps of the paretic limb & contact area of the paretic foot.

1) Standing eyes open: No statistically significant change in any parameters. A tendency towards increase in CoPML, CoPAP, CoPE, SI and CoPA in the experimental group. 2) STS: a significant improvement in SI and CoPE/P in the experimental group. 3) Walking: Significant increase in stance time of the paretic limb in both groups. Significant increase in the contact area of the paretic foot in the experimental group.

Cikaljo et al. (2012) [44]

Pre-post (follow-up only for the experimental group), IV

Sub-acute stroke (control group: n = 22, age: 61.0 ± 7.4 years) Experimental group: n = 6, 58.5 ± 12.1 years)

Not mentioned

1) BBS 2) TUG 3) 10 meter walk time 4) Single leg stance on affected (SAE) and unaffected (SAU) side. 5) Additional outcomes for the experimental group: VR performance time and number of collisions.

1, 2, 3, 4) No statistically significant between group differences on any clinical outcome. Both groups improved over time on the clinical tests. 5) The experimental group showed improved performance and decrease in the number of collisions post- intervention.

Cho et al. (2012) [35]

RCT, II

Chronic stroke (experimental group: n = 11, age: 65.3 ± 8.4 years, 12.54 ± 2.58 months post-stroke; control group: n =11, age: 63.3 ± 6.9 years, 12.6 ± 2.5 months post-stroke)

Not described

Static balance using postural sway velocity (PSV) in the AP and ML directions with eyes open (EO) and eyes closed (EC) 1) PSV – APEO, 2) PSV – MLEO, 3) PSV – APEC, 4) PSV-MLEC. Dynamic balance 5) BBS score, 6) TUG (s)

Static balance PSV measures (1,2,3 and 4) did not show statistically significant changes pre and post treatment in both experimental and control group. 5) BBS: BBS scores improved in both groups post treatment; greater increase in the experimental (VR) group. 6) TUG: A significant decrease in TUG times in both groups, decrease significantly larger in the experimental VR group.

Feasel et al. (2011) [49] & Lewek et al. (2012) [50]

Feasibility study; case report, V

Feasel et al. – Hemiparesis due to various reasons (n = 5). Of these 2 were chronic stroke; 32 and 48 months post-stroke respectively. Lewek et al. Chronic stroke (n = 2; 18 and 21 months post-stroke)

Feasel et al. - Not described Lewek et al.- Patient 1 – right internal carotid artery stroke Patient 2 – embolic stroke to the left middle cerebral artery

Feasel et al. 1) Gait velocity 2) Overground gait symmetry before and after training for stance time, single support time and step length at comfortable and fast walking speeds. 3) Patients’ comments about usability (ease of learning, what was easy or hard, what they liked and did not like about the experience) Lewek et al.: 1) Gait speed (comfortable (CGS) and fast walking (FGS)), 2) Step length asymmetry ratio, 3) Stance time asymmetry ratio

Feasel et al. 1) Gait velocity in the first 5 min of treadmill walking was comparable to overground walking 2) No significant difference in gait symmetry between the first and the last minute of walking. The ‘best minute’ of symmetry was significantly more symmetric than the first minute. No significant differences in average gait symmetry before and immediately after training. 3) Participants reported the task to be mentally taxing. However, positive comments about having visual feedback (walking in the VE) and having a visual goal (keeping the walking path straight) were received. Lewek et al.: 1) Patient 1 improved the CGS from 0.49 m/s (using a large-base quad cane) to 0.84 m/s post-training (without using a cane). This improvement was preserved over the follow-up. The FGS also improved from 0.56 m/s pre-training to 0.95 m/s post-training. Similar results were seen with patient 2, CGS improved from 1.02 m/s pre-training to 1.28 m/s post-training (retained on follow-up) while FGS improved from 1.71 to 1.88 m/s (retained on follow-up). 2) Step-length symmetry improved from 1.52 to 1.32 (with cane) and 1.18 (without cane) in patient 1. The positive effects were maintained on follow-up. Patient 2 did not demonstrate step-length asymmetry to begin with. 3) Stance-time asymmetry did not improve in Patient 1, but improved from 1.11 pre-training to 1.04 post-training. These effects were lost to some degree on follow-up.

Jung et al. (2012) [40]

Pre-post design, III

Chronic stroke (control group: n = 10, age 63.6 ± 5.1 years, 15.4 ± 4.7 months post-stroke; experimental group: n = 11, age 60.5 ± 8.6 years, 12.6 ± 3.3 months post-stroke

Control group: 5 ischemic, 5 hemorrhagic and 6 right, 4 left paretic stroke; Experimental group: 7 ischemic, 4 hemorrhagic and 5 right, 6 left paretic stroke

1) TUG (s) 2) ABC (%)

Both groups improved significantly on outcomes 1 and 2 (TUG change: control: -0.8 ± 0.7 s, stroke: -2.7 ± 1.9 s; ABC change: control: 4.3 ± 3.3%; stroke: 9.5 ± 6.0%). Improvements seen in the experimental group were significantly larger than the control group.

Kim et al. (2012) [41]

Pre-post design, III

Sub-acute to chronic stroke (control: n = 7, age 55.0 ± 13.0 years,12.9 ± 6.1 months post-stroke; experimental: n = 10, age 41.3 ± 6.6 years, 12.6 ± 7.1 months post-stroke

Control group – 3 hemorrhagic, 4 ischemic stroke, Experimental group – 4 hemorrhagic, 6 ischemic stroke

1) Postural assessment scale for stroke patients (PASS), 2) MMAS, 3) Functional Independence measure (FIM)

1 & 2) Both groups showed significant improvements on the PASS and MMAS. The experimental group improved significantly more than the control group. 3) No significant differences were found on the FIM scores both within and between groups.

Cho et al. (2013) [36]

RCT, II

Sub-acute to chronic stroke (control group: n = 7, age 65.1 ± 4.7 yrs; experimental group: n = 7, age 64.6 ± 4.4 yrs)

Control group: 5 ischemic, 2 hemorrhagic; Experimental group: 4 ischemic, 3 hemorrhagic

Walking balance: 1) BBS, 2) TUG (s), Temporal gait parameters: 3) Gait speed (cm/s), 4) Cadence (steps/min), Spatial gait parameters (paretic side): 5) Step length (cm), 6) Stride length (cm), 7) Single limb support (%)

Both groups showed significant improvements on all outcome measure post-test as compared to pre-test (1-7). The change seen post-test was significantly greater in experimental group as compared to the control group in BBS scores, TUG, gait speed and cadence (1-4).

Fritz et al. (2013) [38]

Randomized matched single blind design, III

Chronic community dwelling stroke (control group: n = 13, age 64.5 ± 10.1 years, 3.6 ± 3.2 years post-stroke; Experimental group: n = 15, age 67.6 ± 9.3 years, 2.5 ± 2.6 yrs post stroke)

Control group: 4 left, 9 right paretic stroke; Experimental group: 9 left, 6 right paretic stroke

1) FM

2) BBS

3) DGI

4) 6MWT

5) 3 m walk test

6) 3 m walk test – fast

7) Stroke Impact scale

8) TUG

No between or within group difference on any outcome (1-8). The effect sizes for the outcomes in the VR group were larger than those seen in the control group. These changes were preserved at the 3-month follow-up.

Rajaratnam et al. (2013) [37]

Randomized controlled trial, II

Acute stroke (control group: n = 9, 65.3 ± 9.6 yrs; Experimental group: n = 10, 58.7 ± 8.6 yrs)

Control group: 8 ischemic, 1 hemorrhagic; Experimental group: 8 ischemic, 2 hemorrhagic

1) FRT

2) TUG

3) BBS

4) COP sway

5) Modified Barthel Index (MBI)

Both the control and experimental group showed significant improvements in the TUG (2) and the MBI (5) post-intervention. There were significant improvements in FRT (1) post-intervention in only the experimental group.

Singh et al. (2013) [42]

Controlled pre-post design, III

Chronic community-dwelling stroke (Control group: n = 13, age: 67 ± 8.4 yrs; experimental group: n = 15, age: 65.4 ± 9.8 yrs).

Not specified

1) TUG

2) 30 s sit to stand test (30sSTS)

3) Timed 10 m walk

4) 6MWT

5) Overall balance score (OBS): RMS of the combined AP and ML sway

6) Barthel Index

Both the control and experimental group showed significant gains in TUG (1) and 30s STS (2) tests post-training. However, no difference between control and experimental groups were found post-training on any outcome.