This systematic review is registered in PROSPERO with the code CRD42018102548. We used the most important evaluation items for systematic reviews and meta-analysis guides. We carried out a series of searches following PRISMA [19] recommendations for systematic reviews in 5 scientific databases: (1) PubMed, (2) Web of Science, (3) Scopus, (4) The Cochrane Library, and (5) MEDLINE via EBSCO. The searches were carried out between June 2018 and September 2019. Complete articles in English or Spanish were required. The following keywords from the Medical Subject Headings tree were used for the search: physical therapy, physiotherapy, upper limb, VR, new technologies, Duchenne, Duchenne muscular dystrophy, physical therapy modalities, virtual reality, and virtual reality exposure therapy. See Figure 1 for a complete list of the search strategy and terms used.

The research question was: “Is the use of virtual reality in rehabilitation of the upper limbs effective in children with Duchenne muscular dystrophy?”

The following was derived using the PICO model [19]:

When searching the databases, a series of filters were chosen to limit the searches and select the articles. These were:

After searching different keywords in the aforementioned list of databases and sorting articles by title and summary, relevant articles were identified for complete reading, duplicate articles were eliminated, and the inclusion and exclusion criteria were applied to the sample of definitive data (Figure 1).

We evaluated the methodological quality and internal validity of the studies using the PEDro scale. The PEDro scale (0-10) is based on the Delphi list developed by Verhagen et al [20]. Two independent evaluators (RMV and RBB) used the PEDro checklist to score each study. A study with a score of 4-5 was considered poor or acceptable, where a score below 4 was considered to indicate low methodological quality. Studies with a score below 6 were considered as having low or level 1 evidence, where a study with a score of 6-8 was considered good and a study with a score of 9-10 was considered excellent.

A total of 43 articles were identified for review. Duplicate articles were removed, leaving a total of 32 articles. Twelve of these were excluded as they were considered irrelevant. A total of 20 articles were screened and assessed for eligibility after reading the full text. Thirteen articles were excluded during the eligibility assessment phase: 8 because they showed an incorrect concept, 2 because they did not show the full text, 1 because it did not focus on Duchenne muscular dystrophy subjects, 1 due to incorrect context, and 1 because it did not display upper limb rehabilitation in the variables. The final complete quality assessment included 7 articles.

After deleting duplicates, 32 articles were identified through the keyword search, 20 of which were selected for screening and assessed for eligibility. After a full-text reading, only 7 were included in the qualitative analysis (Figure 1).

The 7 selected studies were published between 2009 and 2019. Of the 7 studies, 6 were carried out in Brazil and 1 in the Netherlands. Using the PEDro scale, 3 studies received a score of 7 and 1 received a score of 8. These studies were considered “good.” Two studies received a score of 6 and 1 study received a score of 5. These studies were considered “acceptable.” Criteria and scoring are depicted in Table 1.

Collectively, the studies enrolled individuals with Duchenne muscular dystrophy aged 5 to 34 years. Individuals with Duchenne muscular dystrophy were ambulatory or wheelchair-dependent. Control groups included those with typical development. The number of individuals in the intervention groups varied. The largest included 60 participants [17], and the smallest included 19 [21]. Only 2 studies compared 2 groups of individuals with Duchenne muscular dystrophy during the intervention and control periods [21,22].

Total study length ranged from 1 day to 20 weeks. Heuntick et al [21] reported training with the virtual reality task for 20 weeks at home. Correa et al [18] reported 12 cumulative days of weekly intervention. Five studies measured the time it took participants to complete tasks they had been assigned in each of the following 3 phases: (1) acquisition, (2) retention, and (3) transfer [5,17,22-24]. These 5 studies agreed with the number of attempts in each phase, with attempts being higher in the acquisition phase (20-30 trials) than retention phase (5 trials) and transfer phase (5 trials).

The most commonly used variable was Motor Function Measurement, which measures limb functionality in 3 dimensions. The scores of Dimension 1 (D1: standing and transfer), Dimension 2 (D2: axial and proximal limb), and Dimension 3 (D3: distal limb) can predict functionality improvements [5,17,24]. However, some of the authors whose articles were selected for review do not believe that the Motor Function Measurement score is predictive of improvement, but instead believe that the time to complete the tasks assigned in each phase is predictive of improvement [22,23]. The relation between performance and Motor Function Measurement-D1 indicates that the use of virtual technology like smartphones is partly reliant on the muscles responsible for standing and transfer, as these muscles allow the head to look at the screen [5]. The dependent variable reported in studies to compare between phases is the movement time or time to perform [5,22-24]. Only one study included a motivation scale (Likert Scale) [18], and only one included a quality of life variable (Kidscreen-52) [21].

Virtual reality games can be simple, allowing individuals to adapt quickly and perform the tasks without problems [24]. Games used in the studies included in the review included musical games, virtual ball mazes, catching cubes, or labyrinths. Study participants used a variety of devices or virtual interfaces such as computers, webcams, touch screens, Kinect sensors, Leap Motion interfaces, simulated sounds, video consoles (like PlayStation II), and smartphones [5,17,18,21-24]. De Freitas et al [17] compared different interfaces with or without physical contact.

Complete summaries of the included studies and their respective intervention details are presented in Table 2 and Table 3.

Study outcomes showed that the grade of motivation improved when using virtual devices [18]. Furthermore, a better performance was seen with a smartphone if a previous learning phase was used [5]. Improvements in the quality of life and elbow extension were seen if training with virtual reality games was performed at home [21].

Although Quadrado et al [23] found that conducting a timed task in a virtual environment facilitated real-life completion of that same task for individuals with Duchenne muscular dystrophy, Massetti et al [22] found no transference of learning between environments when comparing real-life and virtual tasks. Finally, individuals with Duchenne muscular dystrophy showed better performance when using interfaces without contact (like Leap Motion and Kinect) compared to touch screen interfaces [17]. No adverse effects of the use of virtual reality were described in the studies included in our review.

The evaluation of study participants’ functionality and overall muscle performance was mostly done using psychosocial variables and scores obtained using the Vignos Scale [17,21]. The Motor Function Measure Scale was also used by the majority of authors chosen for the review [17,21,22,24].

Although those with Duchenne muscular dystrophy can benefit from the use of virtual reality technologies, the technologies in question should be selected carefully and with mind to participants’ unique abilities and needs [17,22]. The studies reviewed agree on the importance of choosing a suitable task for individuals with Duchenne muscular dystrophy due to the generalized weakness that these individuals experience, especially in the upper limbs at the level of the forearm and in the antigravity muscles of the shoulder and elbow in advanced stages [17,23]. For this reason, the pattern of movement required by a task will determine the difficulty of that task [5,17,23]. Many studies found that patients with greater muscle weakness and loss of functionality had lower speeds of movement and worse performance during assigned tasks [5,17].

According to Masseti et al [22], De Freitas et al [17] found that individuals with Duchenne muscular dystrophy exhibited better responses in virtual environments than in the real world or when using touch interfaces. On the other hand, there was no relationship between performance of virtual reality tasks and general functionality (Motor Function Measurement-total), nor between performance of virtual reality tasks and functionality of the hands (Motor Function Measurement-D3) [5,17,23]. However, these results are based on a previous study in which training did not significantly improve performance of the upper limb (performance of upper limb=3.4). This training consisted of playing PlayStation II games for 20 weeks, which challenged participants to move their upper limbs with the help of a dynamic arm support (Gainboy) that compensated for gravity [21].

The studies of Heuntick et al [21] and Massetti et al [22] compared individuals with Duchenne muscular dystrophy in both the experimental and control groups. The other 5 studies included in our review compared a Duchenne muscular dystrophy group with a typical development control group [5,17,18,23,24]. The Duchenne muscular dystrophy groups always recorded a slower result and had lower scores (time to perform) in the acquisition phases [5,18,23,24].

Regarding the metrics used and results obtained, De Freitas et al [17] reported that age did not influence participants’ functionality improvements or their ability to complete virtual tasks, even with the muscle deterioration caused by the progress of the disease.

The authors of the 7 papers we reviewed debated the influence of the Motor Function Measurement variable. De Freitas et al [17] did not find a relation for Motor Function Measurement-D1 in the performance of the task, while Capelini et al [5] did find a connection when using a smartphone. This may be why individuals in the Capelini et al study had a lower mean Motor Function Measurement-D1 score than individuals in the De Freitas et al study (De Freitas et al=12%, Capelini et al=15%) [5,17]. De Freitas et al [17] also reported that participants with more severe motor impairment and fewer motor skills in the proximal musculature (which corresponds to a higher score in Vignos and a lower score in Motor Function Measurement-D2) exhibited a greater capacity to complete homework, and learning was greater. This is, in part, because the score with which the participants started was also lower and, therefore, the range of improvement after training was broader.

Quadrado et al [23] performed a transfer phase with a change in velocity, and only Quadrado et al measured the time of the task. Massetti et al [22] used the variable “time to perform” in the same phases as Malheiros et al [24]. However, Massetti et al used a crossover model with Duchenne muscular dystrophy patients whereas Malheiros et al had an unaffected control group.

In relation to the transfer of tasks, the results of the study by Quadrado et al [23] show that the realization of the task in a virtual environment facilitated the transfer to the real environment, but that the difficulty of the task must be adjusted to be more difficult in the virtual environment in order to facilitate the transfer to the real environment. This may be attributed to the fact that the space-time organization of the virtual task is different from the real one, reducing performance in the virtual environment [22,23]. Massetti et al [22] states that there is no transfer between environments due to the difference in the complexity of tasks. For example, while Quadrado et al [23] required only the movement of the hand to press a key, Massetti et al [22] required the entire upper limb to hit real and virtual cubes. Heuntick et al [21] supported the idea that if the training is transferred from a virtual environment, the effectiveness may be greater than standard exercises with low resistance, which are normally used to work muscles in isolation rather than general movements aimed at functionality.

The variability of the samples in the studies and the differences between the registered variables made it impossible to carry out a meta-analysis to complete the review. The correlation between the type of task assigned to the participants and the participants’ cognitive demand was not measured, nor was the correlation between the type of task and the participants’ quality of movement [5,22,24]. The preparation of the environment or the accuracy of the devices was not specified and may have caused measurement errors [24,25]. The amount of audio-visual stimulus in some cases caused anxiety and did not produce the desired effects in low mobility individuals compared to conventional treatments [18]. The wide range of devices used makes it difficult to compare the results of the studies. The only similar devices were smartphones [5] and computer keyboards [24]; tasks at the motor level were completely different. If we introduce virtual reality as a tool that helps with user motivation and improves the users’ quality of life, these variables should be used in all studies [18,21]. Finally, A great limitation in the use of virtual reality systems is their cost, making it impossible for all users to access.

The application of virtual reality must go beyond measuring the time for task completion and speed of movement. We must provide studies with specific respiratory and cardiac variables that provide data concerning the metabolic effort during and after exercise and the protection of cardiac activity during exercise. Therapy times and intensities should be defined in individuals with high fatigue and difficulty breathing [26]. Virtual reality can be a great tool for physiotherapy in individuals with Duchenne muscular dystrophy.

Improvement should be measured at different levels in the upper limbs, as well as the proximal level and stabilizing muscles and at the distal and fine motor level [15]. In the Shin et al [15] study, the authors reported benefits at the distal and proximal level when they only expected to find benefits at the distal level, as reported by Capelini et al [5].

The treatment of neuromuscular diseases has changed in recent years, from palliative symptom management to preventive methods for capacity building. The use of new technologies such as virtual reality may be necessary in the treatment of progressive diseases involving movement difficulties, as virtual reality technologies can provide freedom and improve capacity training results. Given that new technologies are increasingly accessible, rehabilitation and physiotherapy programs more frequently employ virtual reality environments to improve task performance and promote the transfer of this practice to daily life in the real world, which is essential for people with disabilities. Ultimately, virtual reality can be used for Duchenne muscular dystrophy rehabilitation programs to improve individuals’ training performance.