Introduction

Technological developments offer new avenues for neurorehabilitation. One such development is virtual feedback (VF). VF uses virtual reality displayed on large screens or on head-mounted displays to provide patients with interactive, multimodal sensory stimuli and biofeedback, and may be applied alone or in combination with physical or cognitive interventions. For example, ‘virtual walking’, offering an illusion of normal gait, may improve motor functions1,2 and may reduce neuropathic pain3, 4, 5 after spinal cord injury (SCI).

The observation that a single type of VF intervention may impact on motor functions and pain simultaneously6 is particularly relevant given the context of intensive neurorehabilitation after SCI during which pain is reported in up to 81% of patients,7 is difficult to treat with conventional therapies,7 and interferes with the capacity to perform high-intensity physical activity.8 In addition, pain has been shown to directly interfere with sensorimotor functions, such as the capacity to learn locomotor9 and upper limb motor10 tasks. As such, simultaneous treatment of motor functions and pain would be preferred over conventional treatment, that is suboptimal as evidenced by the high prevalence of chronic pain.7 Sequential treatment would be undesirable given the overlap in timing of neurorehabilitation and pain onset7 and their negative interaction.

Still, the optimal parameters for VF and their technical implementation are currently unknown. For example, should VF be interactive? What is the optimal duration and frequency of VF sessions? Should VF be combined with other types of therapy, and if yes, how? What patients are likely to benefit from VF? How should we deal with potential adverse effects, for example, pain provocation?11

This narrative review paper provides an overview of interventions that have used VF to improve motor functions or to reduce pain after SCI. In addition, this paper addresses potential working mechanisms underlying the therapeutic effects of VF interventions, identifies knowledge gaps and formulates future directions for clinical research into VF interventions targeting motor functions and pain after SCI.

VF interventions for patients with SCI: an overview

A literature search was performed to identify original papers, conference proceedings and case studies using VF and/or components of VF (for example, visual feedback or movement observation) on outcome parameters related to lower limb motor functions (including balance) and/or pain after spinal cord injury. Both the SCOPUS and Sciencedirect databases were searched. Search terms included ‘spinal cord injury’ and ‘virtual reality’, or ‘virtual feedback’, ‘visual feedback’, ‘virtual walking’, ‘visual illusion’, ‘movement observation’, ‘action observation’, ‘passive observation’, ‘active observation’, ‘imagery’ or ‘mental practice’. In addition, the reference lists of selected studies were searched for additional relevant papers. Studies involving a brain computer interface (BCI) to control VF were only included when the BCI required motor imagery matching the VF task (that is, walking imagery for walking VF). A total number of 17 papers were selected, describing a total number of 13 different VF systems (Table 1). Studies were generally explorative in nature, employing small numbers of patients (1–20 patients per experimental group), and taking place in a clinical setting. VF interventions and assessments were focused on improving motor functions in patients that were pain-free (or in whom pain was not reported; n=12)1,2,12, 13, 14, 15, 16, 17, 18, 19, 20, 21 or on reducing pain in patients with different levels of motor dysfunction and neuropathic pain (n=4).4,5,22,23 Only one study explicitly assessed the effects of a VF intervention on both motor dysfunctions and neuropathic pain.6

Table 1 VF interventions focusing on the improvement of motor functions and/or the reduction of neuropathic pain in patients with SCI: an overview

VF modalities and tasks

Improving motor functions

VF interventions focused on walking (n=4), balance (n=3) and leg muscle (n=2) training. Several studies (n=5) involved healthy control subjects to obtain normative data on outcome parameters (for example, electromyography). None of the studies used a VF control intervention. Different VF modalities were used, including combinations of movement execution or motor imagery with interactive VF or movement observation, and simple movement observation (without overt movement execution). VF interfaces included a (TV) screen, LCD monitor or mirror. Only one VF system used a head-mounted display.1,13 Interactive VF was mediated by a variety of different sensors (for example, movement, tilt, force, video-capture and electroencephalography). VF tasks included displacement in an environmental scene from a first-person perspective, controlling virtual limbs from a first-person perspective, controlling a virtual person (avatar) from a third-person perspective, observing one’s own movements in a mirror, or the displacement of an object on the screen. Several studies made use of supportive devices,1,2,13,14 and one study used neuromuscular electrical stimulation,21 to provide active or passive support. A gradation of task difficulty accommodated patients with different levels of baseline functions, and allowed for standardized therapy progression. Therapy could for example be progressed by increasing the required speed or number of targets to be reached, and by decreasing levels of support. Some studies introduced competitive elements, although these were not necessarily associated with additive effects.1,2,13 Multi-session interventions applied a total number of 5–48 sessions over a period of 2–16 weeks.12,15,16,19,21 The duration ranged from 10 to 60 min per session.

Reducing neuropathic pain

VF interventions focused on reducing at-level and below-level neuropathic pain. Only one study used a VF control intervention,5 and one other study included both healthy and pain-free SCI patients as control subjects to obtain normative data on outcome parameters (for example, pain thresholds).22 Non-interactive video displays and/or a mirror were used. VF tasks included either the observation of video-taped walking legs, synchronized with an upper body mirror-image and (moving) upper limbs to provide an illusion that the patient was walking, or traditional mirror therapy during which movements of the pain-unaffected lower limb had to be observed in a mirror, providing the illusion that the pain-affected limb was moving. Two studies combined VF with transcranial direct current stimulation over the primary motor cortex.5,22 Multi-session interventions4,5,22 applied a total number of 10–15 sessions over a period of 2–3 weeks. The duration ranged from 10 to 15 min per session.

Improving motor functions in patients with neuropathic pain

Only one study systematically assessed the effects of a VF intervention on both motor functions and neuropathic pain in patients with SCI having incomplete lesions.6 No control intervention was applied. Interactive VF was displayed on a large screen from a first-person perspective. The movements of the patients were detected by sensors at the feet, and were used to control a pair of virtual legs. By adjusting the sensitivity of the sensors, the actual movements combined with simultaneous observation and imagery of the virtual legs gave patients the illusion that they could use their limbs normally. Several games involving repetitive lower limb movements could be played, such as the juggling of a ball between the two virtual feet. Task progression was based on motor performance. The study applied a total number of 16–20 sessions of 45 min each over a period of 4 weeks.

Outcome parameters and results

Motor functions

A variety of outcome parameters have been used to assess motor functions, including VF task performance, activity scores, clinical motor tests, gait speed (10-m walking test), balance (Berg Balance Scale), muscle strength (motor score), mobility (Spinal Cord Independence Measure, Walking Index for Spinal Cord Injury II), locomotion (kinematic gait analysis), global motor effects (Patient Global Impression of Change Scale), electromyography and activity in cortical motor areas.

Single sessions of VF involving walking may lead to increased activity scores,1,13 self-confidence and motivation,1,13 leg muscle EMG2 and brain activity in motor cortical areas.12,14,15 Single and multiple sessions of VF for balance may improve task performance,16, 17, 18, 19 static and dynamic sitting balance16, 17, 18, 19 and may increase practice volume and attention span.16 Single and multiple sessions of VF for leg muscle functions may increase muscle strength, endurance and ankle joint range of motion,21 as well as improve gait patterns, gait speed, muscle strength, balance, ankle dorsiflexion (reducing foot drag) and mobility up to 12–16 weeks after training.6 Although, VF activated motor cortical brain areas, the consistency of the activated regions differed across studies.12,15,20 One study showed different brain activation for patients with complete and incomplete lesions.14

Neuropathic pain

Outcome parameters included both clinical (for example, pain intensity, unpleasantness, quality, interference, Patient Global Impression of Change Scale) and experimental assessments (for example, pain thresholds, evoked potentials).

In all studies, patients reported overall reduced pain intensities and/or symptoms after a VF intervention.3, 4, 5, 6,23 In addition, an increased duration of pain relief,4 reduced interference of pain,5 increased pain thresholds3 and reduced evoked potentials3 were reported. Pain reductions were observed up to 12 (see Moseley4 and Soler et al.5) and 16 (see Villiger et al.6) weeks after therapy termination. In addition, the effects of VF were associated with particular pain qualities (for example, continuous versus paroxysmal pain).5 Interestingly, the combination of VF and transcranial direct current stimulation led to significantly more pain reduction than VF or transcranial direct current stimulation alone.5

Feasibility and adverse effects

Feasibility issues such as VF immersion, adherence, drop-out and adverse effects, were generally not systematically assessed. As studies were all clinical lab-based and exploratory, they generally involved the presence of a therapist to ensure a safe therapy session (for example, prevent falls). In those studies that did report adverse effects, these seemed to be occurring in a minority of patients and were considered as relatively mild and comparable to conventional treatment (for example, fatigue, distress and transient pain).1,2,4,6,13,18

Summary of VF studies in patients with SCI

Several important steps have been made regarding the use of VF interventions to improve motor functions and/or to reduce pain after SCI. Effects were demonstrated even after single session interventions and occurred relatively independent of the precise technical implementation and dose, which were heterogeneously employed. However, when comparing multi-session studies, the dose was usually higher for studies aiming to improve motor functions (up to 45 min per session, up to 5 × per week, for up to 6 weeks) as compared with studies aiming to reduce pain (up to 15 min, up to 5 × per week, for up to 3 weeks). Importantly, in the single study that assessed both motor functions and pain, higher therapy dose was not associated with an increase in pain.6 Providing different gradations in task difficulty seems essential to treat patients with different levels of functions and progression and might be particularly relevant when dealing with concomitant pain (that is, to avoid additional increases in pain or fatigue).

Still, the variability in VF systems, the lack of controlled studies, as well as the relatively small sample sizes, prevent an appropriate evaluation of the efficacy of VF interventions as compared with conventional therapeutic approaches for motor dysfunctions and pain. Moreover, it remains unclear for whom (a particular type of) VF might be most effective. A better understanding of the mechanisms underlying the effects of VF may provide additional guidance for VF research and development and for tailoring VF interventions to individual patients.

Potential working mechanisms underlying the effects of VF on motor functions and pain

Several mechanisms have been proposed to underlie the effects of VF interventions on motor function improvement and pain reduction, including sensorimotor and cognitive-emotional mechanisms. It is assumed that these mechanisms coexist in parallel (rather than operating selectively on motor, sensory or cognitive-emotional functions), and that their relative importance could be related to the design and implementation of VF as well as to SCI characteristics.

VF interventions modulate cortical sensorimotor integration

In its simplest form VF can be considered as a sensory stimulus which may provide (additional) information about interactions of the body with the environment. For example, the displacement of a virtual object may provide additional information on body positioning and applied force.16,19 Moreover, VF interventions may involve movement observation and/or may induce or facilitate motor imagery, which are known to activate a network of brain areas commonly known as the ‘mirror system’.24 This network of brain areas, including the pre-motor and primary sensorimotor cortices, is active during the observation, imagery and execution of movements,24 and its (repetitive) activation is thought to impact on motor preparation and motor control,25 as well as on nociceptive processing26 and pain.27

After an SCI, structural and functional neuroplastic changes occur both in the spinal cord and brain, and these changes impact on sensorimotor organization and processing (for extensive reviews see Freund et al.28 and Kokotilo et al.29). Interestingly, somatosensory reorganization was found to be more pronounced in SCI patients with below-level neuropathic pain.30 As such, the additional sensory information as well as ‘mirror system’ activation associated with VF interventions may compensate for lacking or altered sensorimotor information after SCI.23 In turn, this may contribute to neuroplastic changes that promote a normalization of sensorimotor processing, leading to improvements in motor functions and to reductions in pain.25, 26, 27 The importance of cortical sensorimotor mechanisms underlying the effects of VF seem to be supported by the cumulative rather than independent pain-reducing effects of the combination of VF and transcranial direct current stimulation as observed in patients with neuropathic pain after SCI,5 as well as by its simultaneous impact on motor functions and pain.6 In addition, motor imagery abilities20,31, 32, 33, 34 and cortical sensorimotor activity during movement observation,20 motor imagery,31,34, 35, 36 or attempted movements 31,34, 35, 36, 37 have been shown to be intact after SCI, at least for simple movements. Still, these may not be preserved for complex movements.38 Moreover, reports of decreased activation levels and delayed timing,35 as well as of additional brain regions being activated,31,34 suggest that sensori-motor tasks may require more attention after SCI. A better understanding of the cortical sensorimotor mechanisms underlying the effects of VF in patients with SCI is therefore warranted.

VF interventions engage and motivate and may distract from effort and pain

Functional motor recovery has shown to be mainly dependent on therapy intensity and active (goal-oriented) movement repetition.39 This requires that attention and motivation are maintained over sustained periods of time in single and multiple training sessions. VF environments and associated games and tasks that transform simple repetitive exercises to goal-oriented movements may contribute to increased and sustained attention levels, intrinsic and extrinsic motivation and an associated sense of reward. In addition, VF immersion may distract patients from potential negative aspects of exercise and from the perception of pain by creating a state of focused concentration on the VF task at hand. This may coincide with a loss of the awareness of oneself, a sense of control over one’s activity and an altered sense of time.2 Not surprisingly, higher levels of immersion or ‘presence’ in a virtual environment have been associated with increased distraction and analgesia.40 Although distraction is not expected to have long-term effects on the clinical (neuropathic) pain complaints often associated with SCI, it might contribute to the prevention of additional increases in pain and fatigue and as such to increased therapy intensity. Indeed, SCI patients may show increased effort when training in a VF environment,17 with interactive VF leading to particularly high levels of enjoyment, motivation and attention.2,6

Future directions: towards closing knowledge gaps

On the basis of the overview of studies that aimed to improve motor functions and/or reduce pain in patients with SCI, as well as considering our present understanding of potential mechanisms underlying the effects of VF, the following knowledge gaps have been identified (see Figure 1).

Figure 1
figure 1

Knowledge gaps relating to the application of VF targeting motor functions and pain in patients with SCI. LCD, liquid crystal display.

Knowledge gap 1: The optimal VF system

The acquisition or development of a VF system involves considering different interfaces, sensors and additional feedback modalities. Although it is generally assumed that VF systems inducing higher levels of VF immersion are associated with higher effectiveness,2,6,40 this was not explicitly demonstrated by any of the retrieved studies. Rather, it was shown that even a simple mirror may do the trick. As such, more knowledge is needed on the (cost-) effectiveness of different VF systems that aim for similar effects. Still, not only VF immersion could make a particular VF system superior over another. Most VF systems were operated only in the clinical setting and required additional safety precautions (for example, to prevent falls or fatigue). As such, the (cost-) effectiveness of a particular VF system is likely to be dependent on the degree to which it can be integrated with, added to, or replace the conventional practice, while remaining safe.

Knowledge gap 2: VF modalities and tasks

Feedback can be presented from different perspectives, can be a simulation of real-life or involve a non-realistic game environment, and can involve object displacements and/or movement observation. It is likely that particular types of feedback could result in increased immersion and effectiveness, but this remains currently unclear. Interaction is not required per se, since effects on motor functions and pain have also been demonstrated using non-interactive VF. Still, increased levels of VF interaction have been associated with increased muscle activation,2 with reduced pain,40 and with high levels of enjoyment, motivation and attention.2,6 Adding or removing competitive elements or performance feedback could be useful for task progression, but the actual effectiveness of adding these elements has not yet been demonstrated in patients with SCI.1,2,13 Theoretically it could be argued that a graded application of VF protocols would be important to avoid additional increases in pain and fatigue,41 however, there is currently no evidence that this is indeed the case.

Another element of VF that could impact on levels of VF immersion is the congruency with actual body posture.42 Remaining somatosensory afference in incomplete SCI patients might contribute to visual-proprioceptive conflicts when VF is not congruent with body posture. Still, when targeting pain, virtual walking was associated with pain reduction even though applied while patients were sitting. In addition, the possibilities for and safety of applying VF while standing might be limited. As such, a better understanding of the role of body posture and clever solutions for postural support are needed. Finally, for patients with complete lesions, it could be relevant to assess whether VF is associated with attempted versus imaged movements, since attempted movements might increase pain.11

Knowledge gap 3: Dose

Although a variety of VF doses were able to induce effects on motor and pain parameters, none of the selected studies explicitly explored the effectiveness of different doses of VF training. In relation to dose, an important element that remains to be investigated is the use of VF in patients with motor dysfunctions and concomitant pain, which was only explicitly assessed in a single study.6 Patients with SCI often have multiple types of pain (for example, musculoskeletal pain, neuropathic pain, visceral pain),7 but studies targeting lower limb motor functions have generally performed in patients without (significant) pain, or pain was not assessed or reported. As such, the actual potential of VF protocols to improve motor functions and reduce pain simultaneously remains to be confirmed in future studies.

Knowledge gap 4: Patients likely to benefit from VF interventions

Although some studies have hinted towards potentially different effects of VF for patients with complete and incomplete lesions,14 and for patients with lower and higher levels of pain,43 the relationship between VF effectiveness and patient characteristics remains currently unclear. Although our understanding of mechanisms underlying the effects of VF interventions in patients with SCI is limited due to a lack of mechanistic studies, VF effectiveness is likely to dependent on any factor that could impact on sensorimotor integration, on therapy engagement and motivation and on distraction from effort and pain. As such, it is expected that SCI patients with different age or gender,44 somatosensory profiles,45 body representations,38,46,47 motor imagery abilities48 or multiple types of pain5,49 may require different types of VF.

Closing the gaps

Future work should focus on additional multi-session intervention studies including larger sample sizes. Importantly, these should include control interventions and should systematically assess effects on both motor functions and clinical pain, for which the study by Villiger et al.6 forms an important starting point. Regarding pain assessment, it will be important to distinguish between temporary task-related fluctuations in pain (that could increase with effort or decrease due to distraction) versus long-term effects on clinical pain (which reflects true clinical outcome). Moreover, as VF could impact differently on different types of pain, a simple numeric rating scale, or even the Basic Pain Data Set50 may not be sensitive enough to demonstrate effects. For this purpose, additional tools assessing different types of pain qualities (for example, neuropathic pain scales) and nociception (for example, quantitative sensory testing)3 could be useful. In addition, mechanistic studies are needed to improve our understanding of the role and interaction of cortical sensorimotor integration and cognitive-emotional mechanisms in patients with SCI undergoing VF interventions.

Conclusion

VF may contribute to improve motor functions and reduce pain in patients with SCI, although the evidence so far is of low-level quality. These effects could be mediated by modulations of cortical sensori-motor integration, by increased therapy engagement, and/or by distraction from effort and pain. Considering the high prevalence of concomitant pain and its negative impact on neurorehabilitation outcomes, these results provide an important incentive to further assess the potential of VF interventions to simultaneously improve motor functions and reduce pain after SCI. A better understanding of the underlying mechanisms of VF may guide decisions on VF design and may lead to more effective and tailored VF interventions, which could contribute to better neurorehabilitation outcomes after SCI.

Data archiving

There were no data to deposit.