An investigation into the effects of a virtual reality system on phantom limb pain: a pilot study

Abstract

Background:

There is no first-line treatment available for phantom limb pain (PLP). For some years, there has been interest in the use of mirrors and other techniques based on visual feedback. Unfortunately, up until now, all published studies have had methodological weaknesses with two recent systematic reviews concluding that therapies of this kind need more evidence to support their use.

Aim:

To evaluate the effects of a virtual reality (VR) activity on PLP.

Methods:

This was a prospective pilot study of upper limb amputees using questionnaires to evaluate a VR system. Eleven participants were recruited, with nine completing all three sessions of VR. Participants undertook three sessions of VR, one a month for 3 months. Outcome measures were PLP pain intensity using an 11-point numerical rating scale (NRS), number of PLP episodes and duration of the PLP episodes. All participants were also asked for their judgement of change. Open-ended questions captured the qualitative experience of all aspects of the VR experience.

Results:

The mean PLP pain score following three VR sessions reduced (6.11 v 3.56) but this was not a statistical difference (t = 2.1, df = 8, p = 0.07). No statistical difference was found for the number of PLP episodes (Pearson chi-square = 3.43, df = 2, p = 0.18) or the duration of each PLP episode (Pearson chi-square = 22.50, df = 16, p = 0.13). Three groups emerged: those whose pain reduced (the majority), those whose pain remained the same (small number) and one those whose pain increased slightly.

Discussion:

There is insufficient evidence from these results to identify an effect of VR on PLP; however, this is a small group and qualitatively most were content with the treatment and wanted a longer trial.

Introduction

Phantom limb pain (PLP) is experienced by 65–85% of limb amputees.^1–4 In addition, a range of other sensations can originate in the missing limb including phantom limb awareness, phantom sensations (such as itching and pins and needles), and kinetic and kinaesthetic sensations (movement and position in space). Traditionally the treatment target has been PLP; however, it has become clear over many years that treatment options have limited success.⁵

In perhaps the most comprehensive review of treatments for PLP, it was found that despite most therapies having been tried, there is no first line treatment available.⁶ Since Ramachandran’s innovative use of mirrors,⁷ there has been interest in the effect of treatments based on visual feedback. Unfortunately, up until now, all published studies have had methodological weaknesses, with two recent systematic reviews concluding that therapies of this kind have insufficient evidence to support their use.^8,9

One potential reason for this is that we do not yet have clarity on the ‘active ingredient’ inherent in the various techniques that have been employed. For many mirror therapy studies, an element of movement has been included^8,10,11 and it is impossible to untangle the visual and movement effects. To this end, a collection of therapies which utilise cognitive, visual, imagery and movement aspects have been explored. There is significant variation within the studies and the techniques, which makes the evaluation of the effects challenging.⁹

We have previously explored the use of virtual reality (VR) techniques on PLP with some promising results¹²; however, it remains unclear who would be the best target group and where VR therapy would be placed in the priority list of treatments. The high cost (over £60k) and physical fragility of the VR equipment used in our original work made further experiments unrealistic; however, in recent years – driven primarily by the growth of the video games industry – VR devices and computers with the graphical capability to drive them have become consumer-grade items costing a few hundred pounds rather than tens of thousands. Though the cost of VR equipment has dropped while the quality of the environments that can be generated has increased, the principles of generating virtual environments using a combination of computer-generated images and motion tracking remain the same. This follow-up study was a pilot undertaken to help elucidate some of the methodological issues and help to identify which amputees may be the target for such therapies.

Methods

Aim

To evaluate the effects of a VR activity on PLP.

Objectives

1. Recruit upper limb amputees to test the VR equipment.

2. Evaluate the effect of undertaking the VR activity on PLP intensity.

3. Report any effects of the VR activity on the frequency and length of PLP episodes.

4. Assess whether the effects of the VR activity are maintained after completing the activity.

Design

This was a prospective pilot study of upper limb amputees using questionnaires to evaluate a VR system.

Sample

Upper limb amputees were recruited from a large Specialised Ability Centre (SAC) in the North of England. The SAC has more than 3000 patients registered to it and approximately 70 attend each day to be reviewed by members of the multidisciplinary team. The pilot nature of this study meant that no power calculation was undertaken, but we targeted a minimum of 10 participants which matches the number included in the case series reported in the literature previously.^13,14 Ethics committee approval was granted by the Greater Manchester East Research Ethics Committee 15/NW/0897.

Inclusion criteria

• Unilateral upper limb amputee

• Aged over 18

• More than 12 months since amputation

• PLP present in the last month

Exclusion criteria

• Lower limb amputee

• Amputation less than 12 months ago

• Inability to give informed consent

• Visual impairment thought to render the test invalid (people who need to wear glasses day to day are not excluded)

• Medical history of photophobia, epilepsy or any other light sensitivity

Procedure

Upper limb amputees with planned appointments were contacted by letter in advance with details of the study and the participant information sheet; others were recruited via a poster campaign within the SAC. On arrival if they wished to participate they signed a consent form. In a quiet room within the SAC, the participants were fitted with a VR headset (Oculus Rift). Software which visualised an avatar upper limb to represent the amputated limb was commenced once the participant was familiarised with the equipment. Once settled, a 3D ‘ball game’ visualisation was commenced which required the use of the ‘missing limb’ within the visualisation. The movement of the participant’s remaining limb was tracked and transposed in reflection to control the virtual ‘missing limb’. Figure 1 gives an image of the virtual experience of the participant. Each session consisted of three 10-minute blocks in the VR environment with 10-minute gaps and a 15-minute reflection at the end.

Figure 1.

Image of participant view in the virtual world.

Each participant undertook three sessions, one a month for 3 months.

Questionnaire/measures

Prior to session one, participants completed a questionnaire to capture demographic details and their normal PLP status. Then prior to and after each session, they completed questionnaires to record their PLP before, during and after each VR session. All participants who completed the three sessions were then asked to complete the questionnaires again after 1 year. PLP intensity was captured using a numerical rating scale (NRS) where 0 is no pain, 1–3 is mild pain, 4–6 is moderate pain and 7–10 is severe pain. As well as scoring the actual values for PLP pain intensity, number of PLP episodes and duration of the PLP episodes at each point, all participants were also asked for their judgement of change in each of these parameters.

Open-ended questions captured the qualitative experience of all aspects of the VR experience. Examples of these questions are ‘Please let us know what you particularly liked about using the VR equipment’ and ‘Please let us know what you feel could be changed to improve the VR experience’.

Analysis

Descriptive statistics plus paired t-tests and chi-square analyses were utilised to compare groups dependent upon the data collected. In the main, visual interpretation of the data was used to identify groups and to best represent the effects of the VR.

Qualitative data from the open-ended questions were analysed thematically.

Results

Eleven participants were recruited; however two did not complete the study due to health or social circumstance changes between recruitment and the first VR session. Nine participants completed all three VR sessions. Seven of the completers were male and the ages ranged from 46 to 80 years old. Only one had their amputation in the last 2 years; the other nine had been amputated for more than 3 years. The level of amputation can be seen in Table 1.

Table 1.

Amputation level of the participants.

Amputation	Frequency
Left transhumeral	2
Left transmetacarpal	1
Right forequarter	2
Right shoulder disarticulation	1
Right transhumeral	2a
Right transradial	2
Right transmetacarpal	1

^aBoth right transhumeral amputees did not complete.

All 11 were originally right handed, and as seen in Table 1, the majority were right arm amputations. The small number of left arm amputations made it impossible to make comparisons. Seven (63.6%) identified kinaesthesia prior to the VR sessions and only two identified telescoping of the phantom. The most common description of the PLP was ‘like electric shock’ (72.7%) and ‘spasm’ (72.7%). Although two participants said the PLP was predictable, the majority said that it followed no pattern (n = 9, 81.9%).

End of trial (three sessions of VR)

Although there was a drop in mean PLP pain NRS following the three VR sessions (6.11 vs 3.56), this was not a statistical difference (t = 2.1, df = 8, p = 0.07). In addition, no statistical difference could be found for the number of PLP episodes (Pearson chi-square 3.43, df = 2, p = 0.18) or the duration of each PLP episode (Pearson chi-square 22.50, df = 16, p = 0.13).

After the third VR session, groups emerged depending upon each aspect of the PLP. Six (67%) recorded a reduction in pain NRS following the third VR session. When asked whether they had identified a change in NRS, five out of these six were able to articulate that there had been a change and that the pain intensity was reduced. Two participants’ pain intensity was unchanged while one participant scored their pain 7/10 after the third session which was an increase from their first score of 6/10; however, they considered the pain to have reduced (see Table 2).

Table 2.

Pain NRS and change in pain NRS at the beginning, after the third VR experience and at 1 year.

	Participant number
	1	2	3a	4	5	6	7	8a	9	10	11
NRS1	2	8	2	7	7	10	5	5	6	4	6
NRS2	2	5	–	0	6	0	3	–	7	4	5
Actual change 2	↔	↓	–	↓	↓	↓	↓	–	↑	↔	↓
Reported change 2	↔	↓	–	↓	↓	↓	↓	–	↓	↔	↔
NRS3	2	7	–	4	7	10	5	–	7	4	4
Actual change 3	↔	↓	–	↓	↔	↔	↔	–	↑	↔	↓
Reported change 3	↑	↔	–	↓	↔	↓	↓	–	↔	↔	↓

NRS: numerical rating scale; VR: virtual reality; NRS1: pain score at baseline; NRS2: pain score after the three VR sessions; NRS3: pain score after 1 year; ↔: no change from baseline; ↓: Pain reduction from baseline; ↑: Pain increase from baseline.

^aParticipant withdrew after baseline.

Three groups were identified regarding the number of episodes of PLP. At the end of the trial, four participants experienced unchanged numbers of episodes from before the VR sessions, three had fewer episodes and two had more episodes. Similarly, four identified that the duration of each PLP episode remained unchanged (three out of the four overlapped with those with unchanged number of episodes), while another four identified that the duration of each episode reduced in length after the VR sessions. One participant stated that their PLP episodes had increased from 6–10 minutes to 11–30 minutes each.

One-year follow-up

Once again, there was a drop in mean PLP pain NRS at 1 year (6.11 v 5.56), but this was not a statistical difference (t = 1.35, df = 8, p = 0.21).

When reporting their current PLP intensity, three recorded NRS scores to be less than at the start of the study, five reported the same NRS and one reported a slight increase in NRS (6 up to 7). There was a high similarity with their ability to recognise a change as although they were not told what baseline score they gave, four stated that their PLP was slightly less intense, four thought it was unchanged and one felt it was slightly increased (see Table 2).

In addition, no statistical difference could be found for the number of PLP episodes (Pearson chi-square 8.00, df = 3, p = 0.05 but multiple cells below the expected value) or the duration of each PLP episode (Pearson chi-square 20.25, df = 12, p = 0.06) when compared with baseline.

The frequency of PLP episodes had returned to baseline in four participants. Three identified that they were having fewer episodes and two were having more. Regarding the duration of episodes, four had returned to baseline, four identified their episodes as being shorter and one said that each episode had lengthened. When asked if they had recognised a change in number of episodes, 6/9 accurately identified the change (up or down). Similarly seven participants accurately recognised a change in the duration of the PLP episodes.

No link could be found between the effects of VR and the presence of kinaesthesia.

Qualitative analysis identified three themes: (1) phantom experiences, (2) issues associated with the VR activities and (3) side effects from the VR experience. Phantom experiences included changes in the hand which often became unclenched and more open than previously even up to a year later. In one patient, with right shoulder disarticulation, the clenched phantom hand that was stuck to his chest wall for over 20 years came off the chest and unclenched for 2 weeks after VR input; however, once the VR trial ended it relocated back to the chest wall. For most it appeared that pain had reduced over time, and it was apparently associated with the number of VR sessions. There were requests for more sessions as a result of this. PLP would return relatively quickly after each session but often at a reduced level than before. At the 1-year follow–up, two stated that they did not feel that VR benefitted them; however, the majority wanted more sessions and considered that longer term therapy would be helpful. Indeed, one participant was using their Son’s VR equipment and considered that it was being helpful.

Issues associated with the VR activities included minor injuries due to hitting the head, hand and arm on furniture while following the balls within the game. Some of the balls were too wide or too low in the virtual environment and so were not available to hit and some participants felt that because of this they lost interest. As a result it was suggested that the games have more variety and more options. An example of a game variant suggested by a participant was patting animals.

Side effects included mild motion sickness and tingling/throbbing/heat within the phantom and were identified by individuals.

Discussion

This was a small-scale pilot study focusing on the effects of VR on PLP with an attempt to identify aspects within amputees which may help to decide who may benefit from VR therapy. Based on these results, the best or most appropriate target as an outcome measure appears to be pain NRS rather than the frequency or duration of the PLP episodes. Although we did not find a statistical difference in pain intensity, the majority identified a reduction in NRS after the third session when compared with the original pain score. It is possible that a statistical difference could be found with a larger sample.

Overall, there was a trend towards reductions in pain intensity, the number of PLP episodes and the length of episodes. However, some participants could not identify any effect from VR. No links were found between those who experienced movement prior to the trial and the response to VR or any other aspect of their post-amputation phenomena so it was not possible based on these results to give any clues as to which amputees to target with this therapy. According to the participants, no significant adverse effects were found, as those who measured slight increases in intensity, number of episodes and length of episodes did so within a narrow range which might be interpreted as natural variation of what is a fluctuant condition.⁴ A small number of participants reported hitting their head or their hand while in the VR environment. As they were all in the clinical area, there was a clinician with them at all times. All events were minor and the participants laughed about them afterwards.

It is possible that variations in response to the VR may be due to variables within the participants including the reasons for amputation. However, despite many authors hypothesising that there may be differences, a recent study was not been able to find a difference between traumatic and surgical amputations¹⁵ and it was not possible, in such a small pilot study, to explore any effects this may have had on our results.

The quantitative findings are supported generally by the qualitative results. It is clear that in the main the participants considered VR to have some benefits including a reduction in PLP during and after the VR exercise. Most requested more sessions and by implication, this suggests that we have not identified what might be termed a dose of VR therapy. It does appear that the effects of VR wear off over time as the PLP intensity had almost returned to baseline at the one year point. Our equipment was fixed and only able to be used in the department so future studies should consider supplying equipment and allowing participants to use it at home as and when PLP is present. All the patients indicated that they would have liked to continue using a portable version of Immersive VR set up, at home for a prolonged period.

The unclenching of the hand has been reported previously in mirror therapy studies of upper limbs¹⁶ and our participants seemed to gain benefit from this as well as from changes in PLP, although the two things did not appear to be coincident.

The comments about the gaming software were important. Several hit body parts including their head or arm when playing the game due to the immersive effects of the VR experience. Future software needs to take account of this and work to ensure that amputees are safe especially if lower limb amputees (where the risk of falling will be increased) are to be recruited for upcoming studies.

The repetitive and rather dull nature of the activity performed within the virtual world will also need to be considered if participants are to use this at home and more often. New games and games that develop with user input may need to be considered in later iterations of VR for PLP.

Conclusion

The use of an immersive VR environment may have a small, short-term effect on PLP for the majority of amputees who experience PLP. It is possible that a larger effect might be seen if VR is used for longer or at the points when PLP is most problematic. We intend to undertake a pilot study where participants take home the VR equipment to see if the results are affected by more regular use.