Initially, sixty-four patients were selected for the study, out of whom nine were excluded based on the exclusion criteria; fifty-five patients were admitted to the study undertaken in the Clinical Hospital of Psychiatry and Neurology in Brașov (Romania). They were all inpatients in the Neuro-rehabilitation Department. The inclusion criteria were:
Stroke survivors after the acute phase, at least six weeks post-stroke; mild impairment (FIM ≥ 73, FMUE ≥ 13), minor cognitive impairment (Cognitive FIM ≥ 25);
Stroke survivors within no more than four years after a stroke, at least 30-degree flexion and scapulohumeral abduction against gravity, and at least 30-degree elbow flexion against gravity.
The exclusion criteria were: severe cognitive impairments, global or transcortical sensory aphasia, anemia, atrial fibrillation, NYHA class IV heart failure, other dysfunctions in the upper extremity such as surgery, fractures, scapulohumeral periarthritis, or moderate-severe pain. These parameters of the physical and cognitive assessment may limit the possibility to apply the VR exergames with MIRA technology.
During the two-week follow-up period, three study participants, one from the chronic experimental group and two from the chronic control group, were removed from the research because they developed anemia and atrial fibrillation.
The study was conducted over nine months, from July 2019 to March 2020, and the patients were introduced to the therapy one by one (asynchronously). The Research and Ethics Committee approved the study of the Clinical Hospital of Psychiatry and Neurology in Brașov (no. 12534/18 July 2019). Informed consent was obtained from each patient after acknowledging his/her valuable participation in increasing the quality of life of future patients. Besides, they were all informed in advance about the possibility of drop out of the study at any moment, and that their participation in the research represents their consent to use and process personal data, according to the European legislation. The study was registered in clinicaltrials.gov
, with no. NCT04436770.
2.2. Outcome Measures
Our research methodology consisted of the evaluation undertaken with four psychometric scales: Functional Independence Measure (FIM), Modified Rankin Scale (MRS), Modified Ashworth Scale (MAS), and Fugl Meyer Upper Extremity Assessment (FMUE). Manual Muscle Testing (MMT) and Active Range of Motion (AROM) were used to assess muscle strength and range of motion. The assessments were carried out by two experienced physiotherapists specially trained for this research and aimed to register data on stroke severity, activities of daily living, degree of spasticity, motor function and functionality, and the active range of motion. Their presentation and reliability are expressed below.
MRS was used to assess stroke severity. The reliability and validity of MRS have been assessed for test-retest reliability, and the weighted kappa statistic was excellent (kappa w = 0.95) [21
]. Inter-rater reliability of the MRS was found to be excellent intraclass correlation coefficient (ICC) = 0.95 [22
]. The concurrent validity criterion of the MRS was found excellently related to the Functional Independence Measure (FIM) (r = −0.089) while convergent/discriminant validity was 87% with the Barthel Index [23
FIM was used for the assessment of ADLs. The psychometric properties of the FIM scale have been widely studied: (1) Internal Consistency proved to be excellent, with Cronbach’s alpha of 0.93 at admission and 0.95 at discharge. (2) Test-retest reliability was also excellent (ICC = 0.90 for Motor FIM and ICC = 0.80 for Cognitive FIM) [24
]. (3) Intra-rater and inter-rater reliability showed excellent indices compared to the Barthel Index and among various raters. Content and criterion validity also proved that FIM is an adequate instrument to assess functional independence [24
MAS was used to assess the degree of spasticity. Intra-rater and inter-rater reliability of MAS has proved to be an efficient tool for spasticity evaluation, especially for upper extremities [26
]. Construct convergent validity was proven to be excellent compared to Fugl–Meyer Assessment (r = −0.94) [27
The motor function and functionality were measured with a short Romanian version of FMUE [28
] and MMT. FMUE reliability and validity were excellent for upper extremities (ICC = 0.95 for reliability and one factor for validity) [29
]. Although controversial, MMT reliability and validity proved to be a useful instrument to assess muscle motor force [30
AROM was also assessed for both control and experiment groups. Test–retest reliability scores of active ROM were excellent even among unskilled examiners [31
]. The Functional Reach Test (FRT) was used to measure reaching distance [32
Fifty-five participants were assigned to the experimental and control groups, using simple randomization. To avoid bias within our sample (n
= 55), we used Graph Pad QuickCalcs to generate numbers that set patients into four groups. The allocation was performed using sealed opaque envelopes with the group name, which were placed in a plastic container in numerical order. The randomization procedure was performed by different individuals who were not involved in the intervention [33
]. During the research, three participants were identified as having different health conditions which did not permit them to continue participating in the study. Therefore, fifty-two patients took part in the entire research program. The participants were divided into four groups: (1) Subacute Experimental (SE) enrolled patients less than six months since stroke (n
= 6), and (2) Chronic Experimental (CE) enrolled patients more than six months since stroke (n
= 20), both groups received VR therapy; (3) Subacute Control (SC) enrolled patients less than six months post-stroke (n
= 5), and (4) Chronic Control (CC) enrolled patients more than six months since stroke (n
= 21), both groups received conventional physiotherapy. All data regarding group allocation are presented in Figure 1
, the CONSORT flow diagram. The duration of participation in the study for every patient was of 10 working days for two consecutive weeks. All four groups received a 60-min therapy session for upper extremities for ten days. The control groups benefited daily from a standard physiotherapy protocol of exercises such as self-passive mobilization, bilateral active mobilization, and active mobilization with resistance, task-specific functional exercises to increase the ability to perform ADLs, and dexterity exercises for the hand (occupational therapy exercises), for a total time of 60 min. The program of the experimental groups included 20 to 40 min of VR therapy (this duration was set according to the patient’s capacity) associated with dexterity exercises (occupational therapy exercises), so overall every patient performed one hour daily of upper extremity training. The protocol for occupational therapy exercises for the hand included the use of the Canadian plate, thick and thin grip training, lateral and palmar pinch, as well as wrist extensor strengthening tasks, for both groups.
2.4. Virtual Reality Software, Devices, and Exergames
The technology used consisted of a 55-inch TV screen, a computer running MIRA Rehab Limited, London, UK (a software for virtual reality therapy), and a Microsoft Kinect sensor that allows the detection of the human body, joints, and movements on all three axes. MIRA is a software (eHealth) telerehabilitation tool, which improves the effectiveness and convenience of physiotherapy for patients in recovery. The software uses the Kinect sensor to calibrate the patient’s position at the beginning of each VR session, or during exergames, if necessary. The technology used includes an evaluation tool for the AROM assessment, also performed through the Kinect sensor [34
]. Through the MIRA program and the sensor, the patient receives feedback regarding the correctness of motion and posture during VR therapy sessions. During the research, the MIRA software was updated from version 126.96.36.199 (released on 16 July 2019) to version 188.8.131.52 (released on 19 December 2019).
The technology requires the presence of the therapists to assess the patient’s AROM, to determine the exergames tolerance level, and to establish the types of motion and exercise for the patient’s VR therapy session, according to the patient’s functional and motor capacity, at least at the beginning of the therapy (first two sessions). One example is the adjustment of the tolerance levels for motion from 0–100%. The lower the tolerance, the higher the software’s feedback on the correctness of movement, warning the patient that he/she is not performing the task accurately. The maximum level of tolerance used in our research was 20%, according to the manufacturer’s recommendations.
MIRA Rehab software has two categories of exergames based on types of motion, respectively a set of upper-limb analytical movements (see Supplementary Material
), as well as a functional and complex set of exergames which involve muscle control, movement coordination, isometric contraction, and multiple directions of motion. The program features (through a bar on the left of the screen) types of motions that can be selected and (on the right of the screen) types of exergames, thus offering a multitude of possibilities for creating VR therapy programs. As for the functional exergames, the software features two types. One type of exergame can be used for coordination and movement control, for example, Firefly, Follow, Catch, Spaceship, and Move, which require the patient to follow an already designed complex path with three difficulty levels and another type of exergame with progressive difficulty sublevels. The software adjusts automatically the level of difficulty based on the patient’s assessed progress (from one session to another). MIRA VR therapy technology also assigns performance points for every exergame, according to the quality of the execution and the number of repetitions (within a set time), allowing patients to self-assess their progress and enabling them to receive proper feedback on their performance (quality of execution). The software provides a scoring system gathered at the end of the session, in addition to the feedback received during the execution of the exercises. Through this type of feedback, NVIR offers the patient the opportunity to see his progress through a computerized system, from the perspective of the quantitative evaluation of the progress, concretely counting on the results of therapy, and not just verbally.
For the study design, besides the assessment scales, to determine increased patient compliance in the use of VR, we considered two essential parameters, namely AROM and MMT. Firstly, we examined AROM in patients, both through VR technology and manually with the goniometer, where the muscle strength was below 3. In the cases of muscle strength above 3, AROM was measured in the antigravitational position and through MIRA technology. The tolerance level for AROM assessment was set to 20%. Within the first 10 patients, we assessed AROM with MIRA technology and the goniometer concurrently, to identify the level of tolerance needed for an accurate assessment and motion. All the directions of motion from the shoulder, elbow, and wrist were assessed. All reported AROM values are in degrees. For the personalized VR therapy, we adjusted the amplitudes of motion required for every of the chosen movement directions, according to the initial AROM assessment and to other assessment results included in the research (see Supplementary Material
for additional information).
To encourage and motivate the patient to perform the tasks, we adjusted the tolerance level of the movement through the software at the beginning of the first two sessions.
The games used were customized according to the functional capacity of patients, divided into three groups according to their AROM and MMT capacity: limited, low, and high. For patients with low AROM and MMT capacity (less than 50 degrees and MMT between 2 and 3), we adjusted the virtual reality exergames by working bilaterally with both upper extremities (see Supplementary Material
). If they had a limited AROM value, meaning that shoulder flexion and abduction were below 90 degrees (and MMT less than 3), the exergames were adjusted so that the subjects could do, as recreational activities, analytical movements of flexion, extension, abduction, and shoulder rotation, as well as analytical elbow movements, against gravity. For the patients with high shoulder mobility, with at least 90-degree flexion and abduction (and MMT at least 3), exercises involving complex movements were adapted. The multiple movements of the shoulder joint, involving the entire upper limb, were performed in the frontal-anterior, sagittal, and transverse plane. The elaborate motions were also adjusted to increase control and coordination so that movements could be performed diagonally, vertically, and randomly (see Supplementary Material
Twenty percent of the exergames chosen (mostly, the same used for rehabilitation of movement coordination and kinesthetic proprioception) aimed at maintaining the upper extremity in isometric contraction during the exercise (1–3 min). For patients with MMT of at least 3, we increased the exergames difficulty level by attaching weights of 0.5–1 kg to the wrist, according to the patient’s strength. Throughout the virtual reality therapy sessions, the patients received specialized supervision and were initially guided verbally by a physiotherapist until they learned how to perform the exergames correctly. Patients were not seated unless they showed signs of imbalance (n = 7).
Due to the movement tolerance that can be adjusted in MIRA therapy, patients were warned to maintain the correct posture of the segment, for example, if they compensated with elbow flexion while they should have performed shoulder abduction. Also, patients received feedback at the end of every exercise, whereas in previous sessions/exergames, they had only received points.
Besides, all cooperation between patients and therapists was based on verbal and nonverbal communication. In the beginning, the therapy program was explained to patients in detail by therapists. The patients received clear learning-by-doing explanations and were also given a chance to exercise both specialized-assisted and self-automated NIVR therapies, the latter including no further human touch or correction from the therapists. In the first and second therapy sessions, patients needed verbal and physical guidance, but also an example of the movements they have to perform. Starting with the third session, most patients became familiar with the therapy used and needed verbal guidance, specifically only when the level of difficulty of the exergame changed. In the last sessions of VR therapy, patients no longer needed any guidance in exercise performance.