2.1. Study Design
We began the study in April 2018 and completed the evaluation of the three-month follow-up in March 2020. The study followed the standards of the Declaration of Helsinki and was approved by the Research and Medicines Ethics Committee (CEIm) of the Integrated Area of Talavera de la Reina (protocol code: 12/2018). It is registered in the ISRCTN trial registry (ISRCTN27760662) [26
All participants received verbal and written information about the study and gave their written informed consent.
This randomized controlled trial compared the conventional rehabilitation of physiotherapy and occupational therapy (control group) and the combination of conventional rehabilitation with the use of specific virtual reality (SVR) technology (experimental group), following the Consolidated Standards of Reporting Trials (CONSORT) guidelines [27
] and CONSORT-artificial intelligence extension [28
]. Change in upper limb motor skills and its impact on ADLs (baseline, post-intervention, and three-month follow-up) were used as primary outcome. The evaluation of the post-intervention variables was completed three weeks after the start of treatment in both groups (after 15 intervention sessions).
The participants were recruited from the neurology and rehabilitation units of the University General Hospital of Talavera de la Reina, Spain. They were randomly assigned to the control or experimental group by a researcher who did not participate in the intervention and the evaluation process (allocation ratio of 1:1). The conventional rehabilitation therapists were blinded to the study, but neither the participants nor the therapist who applied the VRET could be blinded to the intervention.
2.3. Outcome Measures
The primary outcome variables for this study were upper limb motor function and the impact of stroke diagnosis on ADL involving the use of the upper limb. To quantify these variables, we used the Fugl-Meyer Assessment for upper limb (FMA-UE), the Modified Ashworth Scale for the evaluation of muscle tone and the Stroke Impact Scale (SIS 3.0).
A large number of international guidelines and research in the field of neurorehabilitation and brain damage suggest that the FMA-UE is a valid instrument, given its excellent psychometric properties and its adequate scale to assess the functionality and motor function of the upper limb after a stroke. Furthermore, its use has been validated with virtual reality technology, specifically with the Kinect sensor, which is widely known and was part of the evaluation and intervention processes of this study [29
]. The full version has 113 items, while the subscale for the evaluation of the upper limb examines 63 items (55.75%). Regarding the characteristics of the upper limb, 33 items (29.20%) evaluate motor function, 6 items (5.31%) the sensitivity and proprioception, and the last 24 points (38.09%) correspond to pain and joint mobility. Each item on the evaluation scale responds to an ordinal level of 0 to 2 points: 0 corresponds to an inability to carry out movement and 2 to a capacity to carry it out completely and adequately [29
]. We applied the Spanish version of the FMA-EU [33
] with a Spearman coefficient of 0.946 (p
= 0.000) for the domain of the upper limb, excellent reliability (ICC of 0.987; p
= 0.000), and a Cronbach’s alpha of 0.98 for motor balance of the upper limb.
The Modified Ashworth Scale measures resistance to passive movement according to a scale of 0 to 4, in which 0 corresponds to no increase in muscle tone and 4 to the affected part being rigid in flexion or extension (Kendall W of 0.765; p
= <0.001 for elbow and a reliability of 0.4 to 0.75 for 95% of the assessments) [34
The SIS 3.0 contains 59 items that conceptually evaluate eight important domains: strength, hand function, ADLs and instrumental activities of daily living (IADLs), mobility, communication, emotion, memory, and thinking and participation [35
]. The new structure of four domains (physical, cognitive, emotional, and social participation) has conferred the SIS 3.0 a good reliability of internal consistency (Cronbach’s alpha of 0.98 for the physical domain) and test-retest (ICC of 0.79 for global recovery from stroke), concurrent validity, and responsiveness, which recommend its use in clinical practice and research [36
The motor function of the upper limb, the impact of stroke on ADLs and muscle tone were evaluated and recorded before the start of treatment (baseline), at three weeks (post-intervention) and three months after its completion (follow-up). The entire evaluation process was carried out by the same researcher in both groups (an experienced occupational therapist trained for this research). In addition, we recorded sociodemographic and clinical data, such as age, sex, time elapsed since diagnosis, location of the lesion, risk factors, dominance, pain, self-perceived quality of life, or hemineglect syndrome.
All study participants received 15 treatment sessions lasting 150 min per session and distributed over five consecutive days a week. In total, the intervention lasted three weeks per participant. The patients assigned to the experimental group combined conventional upper and lower limb strength and motor training (50 min physiotherapy and 50 min occupational therapy; administered by the hospital’s physiotherapy and occupational therapy team) with SVR technology devices (50 min), while participants from the control group received only conventional training in physiotherapy (75 min) and occupational therapy (75 min).
The conventional intervention protocol consisted of performing manual therapy techniques (massage), passive and active-assisted mobilizations of the upper and lower limbs, march in parallel, slope and stairs, exercises with and against resistance with balls, elastic bands and dumbbells in therapeutic cage and trellises, active-assisted mobility exercises of the upper limb and fingers in a sitting position, moving objects horizontally on the table, elevation and superposition of objects in the vertical plane, biomechanical tasks that simulated flexion-extension and abduction-adduction of the shoulder and flexion-extension of the wrist and fingers.
The motor training protocol with SVR devices consisted of the application of three systems: (1) HandTutor© glove [40
], 3DTutor© [42
], and Rehametrics© [43
]. All systems are based on intensive and repetitive practice through movement instructions and feedback provided by software with virtual environments and tasks that simulate the movements that the stroke survivor requires for daily life [44
In this work, we will address the clinical and functional effects of the use of Rehametrics© (30 min of intervention/session per participant).
The Rehametrics© software [43
] and Microsoft Kinect sensor [47
] work the upper limb (shoulder and elbow), trunk and lower body. The technology simulates ADLs and mobility in the community with virtual environments and in combination with the use of gamification. It monitors and captures the user’s movement of body, joints in real time through the Kinect sensor. In addition, the sensor calibrates the patient’s position at the beginning of each session and during exercise execution, providing visual feedback for movement and posture correction during treatment sessions. During the study, the Rehametrics© software was updated to the 2019 and 2020 versions.
The software requires the physical presence of the therapist to assess the AROM at the beginning of the session, determine the tolerance level, and adjust and customize the difficulty, duration, range of motion, and number of distracting elements or visual aids.
Rehametrics© has two types of ‘exergames’: (1) analytical exergames that work isolated movements necessary to complete an ADL-inspired flexion-extension of the elbow or abduction and adduction of the shoulder (Figure 1
a) and (2) functional exergames that involve motor control, coordination of movements, contraction dynamics and displacements (Figure 1
b). Before starting the treatment session, the therapist selects the exergames, the duration of each, breaks between exercises, the range of motion for each of the joints, the time pressure for the patient, the number of distractors (night or fog) and the number of visual aids (arrows that indicate if the obstacle appears on one side of the screen or another) (Figure 1
c). The software automatically adjusts the level of difficulty based on the patient’s progress during each exergame. In addition, Rehametrics© provides result graphs to indicate the progress of the patient during the different treatment sessions for a given exergame, the number of failures or the ability to react to obstacles in seconds. This allows patients to visualize their progress and access a quantitative evaluation of failures, successes, and times (Figure 1
The exergames used were personalized according to the functional capacity of the patient, dividing them into low, medium, or high difficulty. In the first sessions, we focus on analytical exergames to increase the joint range of the upper limb. In a second phase, we selected exergames that allowed us to work in several planes and required elaborate and coordinated movements with the trunk and lower limb, lateral, and anteroposterior movements. In the final treatment sessions, elements were incorporated that allowed greater destabilization and high motor control (exergames in a sitting position on bobath balls or in a standing position on a trampoline or destabilizing base). In addition, weights were added to increase muscle strength.
Changes in active range of motion (AROM), patient position, movement correction during activity, and scheduled task execution level were extracted from Rehametrics© software and Microsoft’s Kinect sensor. These changes were not used as an outcome measure in the study. The software automatically stores these variables for each patient and exergames.
2.5. Statistical Analysis
The sample size was calculated with the Epidat 4.2 program. An effectiveness ratio of 90% was estimated for the experimental group and 50% for the control group. Using a power of 80% and a confidence level of 95% (p
< 0.05), a minimum sample size of 20 participants was obtained in each group. The data were analyzed with the IBM SPSS statistical package (version 24.0) (IBM Spain, S.A., Madrid, Spain). To compare the clinical and sociodemographic variables of the groups, Student’s T and chi-square tests were used. Differences in the Ashworth, SIS 3.0 and FMA-UE scales at baseline, post-intervention, and 3-month follow-up were analyzed with inter- and intra-group ANOVA and Student’s t
test. For the FMA-UE, a score of 7.35 in the upper limb subscale was maintained as the minimum detectable change in the three-month follow-up [33
]. Statistical significance was set at a p
-value of less than 0.05.
The analysis of missing data from the control group was carried out with multiple imputation in the analysis (expectation maximization and regression method), with a little’s chi-square statistic of 31,370 (degree freedom = 48; p = 0.970).
The investigator performing the statistical analysis was unaware of the random allocation of participants to the intervention groups.