At least 50% of people who undergo amputations suffer from phantom-limb pain (PLP) . PLP can be related to a certain position or movement of the phantom and may be elicited by physical factors like changes in weather or pressure on the amputation stump, and psychological factors like emotional stress . Stump pain is positively associated with PLP . Several brain imaging studies have confirmed what Melzack hypothesized in 1990, that the brain processes generating the experience of the whole limb endure following amputation , , . Central changes in numerous brain regions including somatosensory and motor areas seem to be a major determinant of PLP, with both peripheral and psychological factors contributing to the alterations , . It is argued that cortical similarities exist between PLP, complex regional pain syndrome type-1 (CRPS-1), and brachial plexus avulsion. Cortical neglect of the affected limb leads to changes in cortical mapping .
In 1992 Ramachandran introduced the use of mirror visual feedback (mirror therapy, M) for the treatment of chronic pain of central origin after stroke . However, review studies conclude that despite support for mirror therapy in the peer-reviewed literature, the bulk of positive data derives from anecdotal reports, constituting weak evidence at best . Tactile treatment (desensitization) has been used in CRPS cases by the present authors and reportedly in other clinics . Desensitization protocols vary, and to our knowledge there are no guidelines or controlled trials on tactile treatment. Because PLP and CRPS seem to have similar alterations in cortical function, tactile treatment was used as an active treatment control in the present trial.
One main reason for surgical amputations in poor countries who have experienced war is the presence of numerous landmines left behind. Landmine victims in low-income remote communities experience PLP and stump pain, with poverty and emotional stress also being contributing factors . In underprivileged communities where the burden of trauma is highest, feasible treatment options are needed. In 2014 a pilot study was done in the catchment area of the present study. By convenience sampling, 18 land mine victims with below-knee amputations, all with phantom limb pain, used mirror therapy for 5 min twice a day. All participants bar two reported a reduction of phantom pain, less headache, improved sleep, and improvement of function at the conclusion of a 3-week treatment period. Three patients also reported improved functional control of the prosthetic limb: “now it feels as if my toes really touch the ground”.
The primary aim of this study was to examine the effect of mirror therapy on phantom and stump pain in patients with traumatic trans-tibial amputation, with particular reference to low-income communities. A secondary aim was to study the duration of the treatment effect for up to 3 months.
2.1 Study population and design
The study is registered at ClinicalTrials.gov, ID NCT02912975. The study sample was composed of adults who had developed phantom pain secondary to trans-tibial amputations after landmine trauma in rural Cambodia. The study aimed to examine low-cost treatment alternatives for amputees with phantom limb pain (PLP) and amputation stump pain in a low-income community. The study was designed to examine the effects of mirror therapy as monotherapy (M), tactile treatment as monotherapy (T), and combined mirror+tactile treatment (M+T). The effects of the three treatment arms were compared in an open, randomized, semi-crossover study. The duration of all treatment periods was 4 weeks. Response to treatment was defined as a 33% reduction in VAS-rated PLP. Non-responders to the first-round tactile treatment were crossed over for secondary mirror therapy, and initial non-responders in the mirror group crossed over for secondary tactile treatment. Non-responders to combined M+T treatment during the first-round treatment were excluded from further treatment (Fig. 1). The semi-crossover design was applied because it is claimed to increase the probability of giving the best treatment when the therapies under study are poorly documented , .
The responders were observed for 3 months after the end of treatment. Following treatment they kept the treatment equipment in their home (the mirror and/or a box with utensils for desensitization). During this 3-month observation period they were encouraged to repeat the treatment they found most successful if they experienced increasing pain. The study was closed after an end-point evaluation 3 months after completed treatment.
2.2 Study sample, recruitment and randomization
The study was conducted from May to August 2016 in a remote rural mine-infested area, Samlot, in Battambang Province, Cambodia. Study patients were chosen by the following criteria: age >16 years; unilateral trans-tibial amputation after landmine trauma more than 12 months before entering the study; suffering from phantom limb pain with or without stump pain. Patients were excluded from the study if they had amputation stump anomalies requiring surgical reconstructions such as chronic infections, neuroma or major soft tissue deformities; chronic alcoholism or drug abuse; loss or deformities of limbs other than the present amputation; or mental and/or cognitive disorders rendering self-rating of health unreliable. The study aim and design were publicized in the catchment area by the local health authorities. Trained local physicians screened potential participants regarding inclusion and exclusion criteria.
When the study sample was identified, computer-generated random numbers were used for simple randomization to the M (n=15), T (n=15) and M+T (n=15) subsamples. Randomization was done without stratification regarding the severity of baseline pain. For several reasons, the study could not be performed blinded: The patients’ families and local health workers were mobilized both in treatment and collection of data. There is anyway close contact within rural communities in daily work and social life. Thus the types and effects of treatment could not be hidden between the study patients and the support staff.
2.3 Variables and factors
Before the intervention, the study population was examined for signs of complex regional pain syndrome (CRPS) by expert clinical examination, using the criteria of the International Association for the study of Pain , . The result variables were estimated by self-rating, using a visual analogue score for estimates of phantom limb pain and stump pain . Participants were asked to mark with pen on a horizontal 10 cm line their estimates of mean pain during the previous week. The VAS scores were registered with one decimal. The study sample was categorized in three groups according to the self-rated severity of pain, severe pain defined as VAS >6 cm, moderate pain as VAS 3–6 cm, and mild pain as VAS <3 cm. Data on gender, age, time since the amputation, and the level of the amputation, were collected.
2.4 The interventions
A pain specialist (LD) instructed the Khmer investigators (HSO, YVH) in clinical pain examinations and in the implementation of treatment protocols. All study patients were told that there were changes in the cerebral cortex associated with PLP, and that external sensory stimuli might modify brain imaging due to the plasticity of cortical function. Careful instructions regarding the details of the interventions was given to participants.
Mirror therapy: the patient sits on a chair, both lower limbs bared. A mirror measuring 30 cm×80 cm is placed between the legs along the trans-tibial amputation stump so that the patient can see the uninjured limb in the mirror while the amputated limb is hidden behind the mirror screen. For 5 min every morning and night the patient fully concentrates on performing slow repeated movements of the foot from a neutral position to maximum dorsal flexion while closely observing the reflected image of the uninjured limb in the mirror.
Tactile treatment: The patient lies on a bed, not watching the stump, just concentrating on feeling the tactile stimuli, while for 5 min every morning and evening a close family member carefully exposes the skin of the medial, frontal, lateral, and dorsal parts of the amputation stump to five different stimuli: a stone, a wooden stick, a soft brush, a soft cloth, and a soft feather. The same sequence of tactile stimuli is applied in all treatment sessions.
Combined mirror and tactile treatment: The mirror and the tactile treatments go on serially, with 5 min for each treatment. If the patient has the mirror therapy before the tactile treatment in the morning, the tactile treatment is done before of the mirror therapy at night.
2.5 Sample size and processing of data
The sample size calculation was based on the distribution of self-rated PLP. A change of VAS rating of 33% was considered to be relevant. Given an assumed standard deviation of VAS rating of 10%, power at 80%, significance level at 5%, and with a semi-crossover design, 15 patients were included in each of the three treatment arms. Local expert staff monitored compliance by weekly interviews in the Khmer-Khmer language at the home of each study patient. The compliance rate was estimated as the rate of actual treatment periods by required treatment periods (two times per day for 28 days). Baseline data for the outcome variables were collected within 1 week before the commencement of the treatment period, and follow-up data 1 week after the conclusion of the treatment period. For the first-round non-responders to M or T, a second-round treatment of 4 weeks with the alternative treatment started within a month after ending the initial treatment. A “drop-out” was a patient who decided to leave for reasons not related to the study and its implementation. There was only one drop-out in the study.
2.6 Statistical platform and ethical considerations
The study sample was analysed at three stages. Firstly, the randomized subsamples were compared at the conclusion of the first-round treatment. Secondly, based on the assumption that a patient exposed to mirror or tactile therapy remains primed by this treatment when later exposed to the alternative treatment, the responders to the second-round of treatment were re-assigned to the M+T group, and the three treatment arms compared in the reclassified study sample. Finally, the differences between baseline and end-point pain after 3 months were compared between the three subsamples. Continuously and symmetrically distributed variables are expressed by mean values with 95% confidence intervals (95% CI) by using the Student procedure. Proportions are described using exact 95% CIs . Analysis of Variance (ANOVA) was performed for changes within and comparison between the subsamples. Estimates were considered significantly different if the 95% CI for the difference did not contain zero.
The study was approved The Cambodian National Ethics Committee for Health Research, ref. no. 081/2016 NECHR. The data were stored and processed according to ethical permission from the Norwegian Social Science Data Service, ref. no 2015/2193/REK-North.
3.1 Study population and randomization
The mean age of the study patients was 55.7 years (SD 6.7); all but one was male. Traumatic amputations had occurred years before (mean 23 years, range 15–32 years). Most patients had undergone only one primary surgical operation, six patients had experienced two primary operations, and one patient three primary operations. The level of tibial amputation varied: 14 patients had amputations through the proximal third of the tibia, 15 had mid-shaft amputations, and 16 patients amputations through the distal third. Pre-intervention clinical screening of the study sample did not reveal any cases of CRPS. The mean baseline levels of pain were high: with PLP, VAS had a mean of 7.2 (SD 2.0); with stump pain, VAS had a mean of 8.1 (SD 1.5). Two outliers were identified. One patient reported mild phantom pain but severe baseline stump pain (VAS 8.6). Another patient reported mild phantom and stump pain. The two pain variables were checked across treatment allocation, age, time since the amputation injury, and level of amputation (Table 1).
3.2 The first round of treatment
Compliance rates during the first-round treatment were high, with a mean of 89.9% (SD 16.6). In the second week of the initial treatment period, one patient in the treatment group M+T developed a severe soft tissue infection at the amputation stump and dropped out of the study. This left a study sample of 44 patients who completed the initial intervention. At the conclusion of the 4-week treatment period, reduced PLP and stump pain were observed in all of the three treatment arms, except for one patient in the M group and one in the T group. The mean reduction in VAS ratings for phantom and limb pain in all three treatment arms was >50%. No significant differences were observed between the three subsamples (Table 2). Reductions in VAS scores were similar for the patients with severe compared to moderate pain.
3.3 The second round of treatment, reclassification of subsamples
Nine non-responders were identified from first round treatment and assigned to a second round of treatment with the alternate therapy. Three of the non-responders were in the M-subsample and five in the T-subsample. Two non-responders refused to undertake further treatment, so the second round of treatment was given to seven patients. The mean delay between the conclusion of round one and start of round two was 33 days (range 19–53). The compliance rate during the round two treatment was 100%. All initial non-responders reacted to the second-round treatment with a reduction in VAS rating of >90% for phantom as well as stump pain.
The second round patients who had undertaken the two monotherapies sequentially were reclassified to the M+T category, thus making a study sample of 14 M patients, 10 T patients and 20 M+T patients for the main analysis of treatment effects. Table 3 demonstrates a tendency toward better effect of combined mirror-tactile treatment compared to the monotherapies as estimated by percentage reduction in VAS scores. The 95% CI for the difference in percentage PLP reduction between T and M+T was 2.8–20.3; between M and M+T 10.0–8.6; and between M and T −11.5–31.0. Also regarding stump pain the combined treatment had a slightly better effect than the monotherapies as estimated by percentage VAS reduction, the 95% CI for the difference between T and M+T being 5.0–15.7; between M and M+T 4.9–22.8. No significant difference was found between the monotherapies, the 95% CI for the difference between the M and T subsample regarding percentage VAS reduction being −10.0–17.0.
3.4 Duration of treatment effects
None of the study patients applied tactile or mirror therapy during the post-treatment observation period. All forty-four study patients estimated the levels of pain 3 months after the conclusion of the treatment by VAS scales. The end-point ratings demonstrated that the intervention had a sustained effect. The changes in VAS rating from the end of the last intervention to evaluation 3 months later were minimal: for PLP the mean difference in rating was 0.9 (SD 0.8), for stump pain the mean difference was 1.0 (SD 0.9). No significant differences between the three treatment arms were observed regarding how long the treatment effects lasted.
We report for the first time a randomized controlled clinical trial of mirror and tactile therapy for PLP and stump pain in a homogenous sample of amputees. This study documents a significant and sustained reduction of PLP and stump pain after two brief interventions daily for a 4-week treatment period. Treatment effects were as good for patients with moderate PLP as for those who were severe. The majority of the study patients also reported improved well-being and reduced emotional stress (“freshness in the mind”, “improved sleep”, “less headache”). Mirror therapy (M) combined with simple desensitization (T) had a slightly better effect than M or T as monotherapies. The difference was statistically significant, however not considered to be of clinical importance.
4.1 Limitations to the study
External physical strains and climatic conditions may trigger PLP . As the geographical catchment area of the study was narrow and study patients all poor, hard-working farmers making a living from non-mechanized agricultural labour, we assume that this minimizes any significant influence of uncontrolled external physical variables. As noted earlier, emotional stress may trigger PLP  and this was illustrated in the present trial. One of the initial non-responders reacted to a stressful accident to his son with increasing PLP, but 1 month later responded well to M. Even if experienced local health workers monitored the study patients closely, there may still have been emotional stressors escaping our attention at pre-intervention screening and through the observation period. Attention per se, the experience of being seen and cared for, and also getting some insight into the neurophysiology of PLP may well have had some placebo effect. However, the study design made it unlikely that external factors of this kind should have favored one treatment arm over the other.
There were minor deviations from the trial protocol. One study patient responded poorly to the initial tactile treatment, perhaps because he suffered from a peroneus neuroma not identified at pre-intervention screening. This patient should thus not have been included in the study. In the event he did gain by being included and responded positively to the crossover treatment. Another patient with severe stump pain was erroneously included on the study, despite having mild pre-intervention PLP. However this patient also benefitted from being included, reporting significant stump pain relief after the crossover treatment. Despite a rather uncontrolled study context the compliance rate was generally high. Four of the nine first-round non-responders had to interrupt or reduce the prescribed doses of treatment due to job or family obligations, but still reported significant reductions in pain levels. Chronic pain after landmine amputation is a well-known condition in these Cambodian villages, affecting the entire family and a concern also for neighbors and friends. As the research team explained the physiology of chronic pain and options for treatment, the local community and the family responded positively to the trial. This probably enhanced compliance, though possibly also any placebo effect.
The re-allocation of crossover patients into the M+T subsample could be questioned. Is a patient undergoing simultaneous combined mirror and tactile treatment equivalent to a patient using the same two interventions sequentially as monotherapy? Based on their firsthand clinical experience and on brain imaging studies, the authors believe that this is so. It is well documented that the central nervous reorganization of cortical patterns after desensitization or mirror therapy is not erased or “washed out” like an analgesic drug , , , , . Based on this neurophysiological understanding of cortical plasticity, we assume that the treatment-induced alterations in somatosensory patterns after round one of treatment remained even if the active treatment ended. We therefore hypothesize that a patient undertaking, say, mirror therapy remains primed by the initial treatment when the potentially additive effect of another type of cortical stimulation occurs. Hence, there is evidence to support an analytical approach where the main analysis is done on a sample where the crossover responders have been merged with the subsample initially randomized for combined M+T therapy.
Finally, cross-cultural differences should be considered. Does the concept “pain” carry the same meaning in Khmer and English language? “Pain” in Khmer, “chheu”, means physical suffering. However, the term is also used in an abstract sense referring to “the feeling of agony”. The Khmer term thus seems to carry the same double meaning as the English term, comprising both effective pain (pain as immediate sensation) and affective pain (what the pain does to me). When explaining the VAS rating to the participants, care was taken to make them understand that the present rating aimed at the strictly physical pain .
The semi-crossover design is ethically attractive as the study patients are entitled to alternative potentially effective treatments. It is also a powerful method as it generates data both from intra-unit and between-unit comparisons. The design is powerful also in small numbers of study subjects, and it compensates well for minor bias . Despite the rather uncontrolled study conditions, the study was conducted according to the protocol with just one dropout and without any withdrawals.
4.2 Clinical implications
Even if mirror and tactile stimulation work well as monotherapies, the study suggests the superior effects of a combined approach. This is in line with previous studies recommending multi-disciplinary treatment strategies for PLP, including enhancement of a cognitive understanding of the painful condition , . In the present study the intervention did not merely comprise the mechanics per se of mirror and/or tactile stimulation. The understanding of malformed patterns in the brain as the source of pain, and evidence that the cortex can be reorganized and “normalized”, were shared with all study patients before and during the treatment period. Further, the weekly visits by research assistants were not value-neutral observations, but allowed trusted and knowledgeable local health workers to encourage the participants and their families to pay careful attention to the therapy sessions and to improve their morale. The Cambodian study was thus not an exercise in “self-delivered home-based mirror therapy” as reported by Darnall . Even if the impact of verbal support and encouragement as an integrated part of the study intervention cannot be validated, the authors hold that such support is an essential condition for effective somatosensory stimulation. Besides, it would be unethical to withhold from the participants our recent and updated understanding of the plasticity of the central nervous system.
The findings in the Cambodian study are especially relevant for low-resource communities. The burden of trauma hits hardest in low- and middle-income countries . The survivors of atrocities in Central Africa, Gaza and Aleppo do not have access to sophisticated therapeutic interventions for PLP or stump pain. Where poverty and war are endemic, it is difficult to control and reduce the “sympathetic discharge” produced by emotional stress, as recommended by Hagenberg and Carpenter . But even in such underprivileged communities it is undoubtedly possible to relieve chronic pain and improve function by mirror and/or tactile therapy with support from a close friend and/or family. The feasibility of mirror and simple tactile therapy is the main asset of these methods.
The present study does not provide evidence regarding the recommended intensity and duration of the interventions. Comparison with a true control group receiving either no treatment or placebo is lacking. In retrospect most participants reported significant improvement after just 2 weeks of treatment, but precise data was not collected on personal trajectories of pain during the intervention. Griffin at al report significant reductions of severe PLP after 28 brief treatment sessions , which is comparable to the Cambodian experience of a successful treatment protocol of 56 brief interventions over 4 weeks. As the follow-up period was short, just 3 months, the long-term effect of the intervention cannot be estimated from the this study. It is well documented, and also reported by several participants in the present study, that stressful events can impact unfavorably on attempts to relieve pain. In order to examine the duration of the treatment effect and identify factors that may trigger PLP relapse, a qualitative study of the participants in this study is now being conducted in Cambodia. The results are pending.
Four weeks practice of mirror therapy and tactile treatment causes a sustained reduction of PLP and stump pain in the majority of trans-tibial amputees. The most efficient method seems to be simultaneous mirror therapy and tactile treatment, or the two interventions serially. The interventions are simple and cheap, thus appropriate to the treatment of PLP in low-resource communities.
We want to thank Dr. Yous Pov at Samlot Operational District, and Drs. Bunreth and Davoeung at Battambang Health Department for facilitating the study. The research assistants Nith Nan, Roth Ny and Sim Sophal in Samlot were instrumental in data collection and the supervision of study patients. Professors Stig Larsen and Eystein Skjerve at the Norwegian University of Life Sciences helped design the study and analyze the data.
Diers M, Christmann C, Koeppe C, Ruf M, Flor H. Mirrored, imagined and executed movements differentially activate sensorimotor cortex in amputees with and without phantom limb pain. Pain 2010;149:296–304. Google Scholar
Flor H. Phantom-pain characteristics, causes and treatment. Lancet Neurol 2002;1:182–9. Google Scholar
Montoya P, Larbig W, Grulke N, Flor H, Taub E, Birbaumer N. The relationship of phantom pain to other phantom limb phenomena in upper extremity amputees. Pain 1997;72:87–93. Google Scholar
Campos de Paz Braga LW, Downs HJ. A preliminary functional brain study on amputees. Appl Neuropsychol 2000;7:121–5. Google Scholar
Roux FE, Ibarrola D, Lazorthes Y, Berry I. Virtual movements activate primary sensory areas in amputees: report of three cases. Neurosurg 2002;49:736–41. Google Scholar
Melzack R. Phantom limbs and the concept of a neuromatrix. Trends Neurosci 1990;13:88–92. Google Scholar
Karl A, Diers M, Flor H. Neuroelectric source imaging of steady-state movement-related cortical potentials in human upper extremity amputees with and without phantom limb pain. Pain 2004;110:90–102. Google Scholar
Moseley LG, Gallace A, Spence C. Is mirror therapy all it is cracked up to be? Current evidence and future directions. Pain 2008;138:7–10. Google Scholar
Ramachandran VS, Altschuler EL. The use of visual feedback, in particular mirror visual feedback, in restoring brain function. Brain 2009;132:1693–7. Google Scholar
Barbin J, Seetha V, Casillas JM, Paysant J, Perennou D. The effects of mirror therapy on pain and motor control of phantom limb in amputees: a systematic review. Ann Phys Rehab Med 2016;59:270–5. Google Scholar
Pleger B, Tegenthoff M, Ragert P, Forster A, Dinse HR, Schwenkreis P, Nicolas V, Maier C. Sensorimotor returning in complex regional pain syndrome parallels pain reduction. Ann Neurol 2005;57:425–9. Google Scholar
Husum H, Resell K, Vorren G, Van Heng Y, Murad M, Gilbert G, Wisborg T. Chronic pain in land mine accident survivors in Cambodia and Kurdistan. Soc Sci Med 2002;55:1813–6. Google Scholar
Hektoen L, Løken T, Larsen S. A stratified semi-crossover design used in evaluation in clinical mastitis in dairy cows. Proceedings of the 10th International Symposium on Veterinary Epidemiology and Economics, 2003. Available at: http://www.sciquest.org.nz/node/62997. Accessed: 30 Apr 2018.
Hellings IR, Larsen S. ImproWin® in the treatment of gastric ulceration of the squamous mucosa in trotting racehorses. Acta Vet Scand 2014;13:47–56. Google Scholar
Harden R, Bruehl S. Diagnostic criteria: the statistical derivation of the four criterion factors. In: Wilson PR, Stanton-Hicks M, Harden RN, editors. CRPS: current diagnosis and therapy. Seattle: IASP Press, 2005:45–58. Google Scholar
Harden RN, Oaklander AL, Burton AW, Perez RS, Richardson K, Swan M, Barthel J, Costa B, Graciosa JR, Bruehl S. Complex regional pain syndrome: practical diagnostic and treatment guidelines, 4th edition. Pain Med 2013;14:180–229. Google Scholar
Quesnel-Vallée A. Self-rated health: caught in the crossfire of the quest for ‘true’ health? Int J Epidem 2007;36:1161–4. Google Scholar
Altman DC. Practical statistics for medical research. London: Chapman & Hall/CRC, 1999. Google Scholar
Le Feuvre P, Aldington D. Know pain, know gain: proposing a treatment approach for phantom limb pain. J R Army Med Corps 2014;160:16–21. Google Scholar
Price DD, Harkins SW. The affective-motivational dimension of pain. A two-stage model. APS J 1992;1:229–39. Google Scholar
Hagenberg A, Carpenter C. Mirror visual feedback for phantom pain: international experience on modalities and adverse effects discussed by an expert panel: a Delphi study. PM&R. 2014;6:708–15. Google Scholar
Darnall BD. Self-delivered home-based mirror therapy for lower limb phantom pain. Am J Phys Med Rehabil 2009;88:78–81. Google Scholar
Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med 2006;3:e442. Google Scholar
Griffin SC, Curran S, Chan AWY, Finn SB, Bajer CI, Pasquina PF, Tsao JW. Trajectory of phantom limb pain using mirror therapy: retrospective analysis of two studies. Scan J Pain 2017;15:98–103. Google Scholar
About the article
Published Online: 2018-07-03
Published in Print: 2018-10-25
Research funding: The study was sponsored by a grant from Lancelot Holding AS, Norway, org. no. 993 229 865.
Conflict of interest: The authors report no conflicts of interest related to the study.
Informed consent: Before signing the informed consent forms, local trusted health workers carefully informed all study participants about the study procedure, data management and of the right to withdraw from the study at any time.
Ethical approval: The study was approved by The Cambodian National Ethics Committee for Health Research, ref. no. 081/2016 NECHR. The data were stored and processed according to ethical permission from the Norwegian Social Science Data Service, ref. no 2015/2193/REK-North.
Citation Information: Scandinavian Journal of Pain, Volume 18, Issue 4, Pages 603–610, ISSN (Online) 1877-8879, ISSN (Print) 1877-8860, DOI: https://doi.org/10.1515/sjpain-2018-0042.