OUP user menu

Decreasing Phantom Limb Pain Through Observation of Action and Imagery: A Case Series

Geneviève Beaumont BA, Catherine Mercier PhD, Pierre-Emmanuel Michon PhD, Francine Malouin PhD, Philip L. Jackson PhD
DOI: http://dx.doi.org/10.1111/j.1526-4637.2010.01048.x 289-299 First published online: 1 February 2011


Background. Phantom limb pain is often resistant to treatment. Techniques based on visual-kinesthetic feedback could help reduce it.

Objective. The objective of the current study was to test if a novel intervention combining observation and imagination of movements can reduce phantom limb pain.

Methods. This single-case multiple baseline study included six persons with upper or lower limb phantom pain. Participants' pain and imagery abilities were assessed by questionnaires. After a 3–5-week baseline, participants received a two-step intervention of 8 weeks. Intervention 1 was conducted at the laboratory with a therapist (two sessions/week) and at home (three sessions/week); and Intervention 2 was conducted at home only (five times/week). Interventions combined observation and imagination of missing limb movements. Participants rated their pain level and their ease to imagine daily throughout the study.

Results. Time series analyses showed that three participants rated their pain gradually and significantly lower during Intervention 1. During Intervention 2, additional changes in pain slopes were not significant. Four participants reported a reduction of pain greater than 30% from baseline to the end of Intervention 2, and only one maintained his gains after 6 months. Group analyses confirmed that average pain levels were lower after intervention than at baseline and had returned to baseline after 6 months. Social support, degree of functionality, and perception of control about their lives prior to the intervention correlated significantly with pain reduction.

Conclusions. Persons with phantom limb pain may benefit from this novel intervention combining observation and motor imagery. Additional studies are needed to confirm our findings, elucidate mechanisms, and identify patients likely to respond.

  • Phantom Limb
  • Pain
  • Motor Imagery
  • Intervention


Phantom limb pain (PLP) affects 60% to 80% of amputees [1] and is recognized as difficult to treat [2]. It was demonstrated that, after the amputation of a hand, the representations of the body parts adjacent to the hand in the cortical somatotopy expands in motor [3] and somatosensory [4] cortices, and it was suggested that the amount of reorganization is associated with the magnitude of the PLP [2–5]. This led to the idea that PLP might be underlined by a maladaptive plasticity [5–7] and triggered efforts to develop new rehabilitative interventions that could reverse or reduce this plasticity. Ramachandran and Rogers-Ramachandran[8] showed that when amputees place their intact upper limb inside a mirror box, the reflection of this limb creates the impression that the missing limb is still there. Then the movements of the intact limb generate illusory movements of the missing limb, which appear to alleviate PLP in some patients. Two randomized controlled studies have shown that a similar mirror intervention with lower limb amputees can improve significantly their capacity to move their phantom leg [9] and reduce PLP [10]. Other techniques, such as virtual reality, seem to be effective in decreasing PLP either alone or in combination with motor imagery [11]. Giraux and Sirigu [12] proposed that afferent information, provided by the visual system, restores the coherence with both motor command and motor activation in the cortex contributing to the voluntary access to the motor representation of the missing limb. The importance of having access to both visual and kinesthetic feedback is supported by the observation that providing visual feedback helps to produce kinesthetic feedback and that both can restore voluntary control over phantom movements [13]. The observation and imagination of movements also activate the cerebral areas involved in their execution [6,14], which can potentially prime the voluntary production of phantom movements. Such voluntary and controlled phantom movements could then provide an imagery-based coherent feedback which carries both visual and kinesthetic components. MacIvert and collaborators [15] found a significant reduction in unpleasantness and intensity of constant pain and exacerbations, with a corresponding reduction of cortical reorganization. In contrast, Chan and collaborators [10] showed that observation of the missing limb in movement through mirror therapy is effective but not the imagination of movement. Although Moseley [16] successfully used observation of images prior to motor imagery to decrease PLP, no study used a combination of movement observation on video followed by imagination. Moseley [16] showed the efficacy of graded motor imagery on pain, including PLP, after a 6-month follow-up assessment, but because only 11 out of the 51 patients in the sample had PLP, it is difficult to conclude on the effectiveness for this specific population.

The primary aim of this study was to test whether combining the observation and the imagination of movement could reduce PLP. This novel intervention consisted in training patients to evoke and control mentally the images and sensations of movements performed with their missing limb. We expected patients to show reduced levels of pain and improved ability to imagine and move their phantom limb after intervention. A secondary aim was to identify patient characteristics associated with PLP reduction in order to identify those responding best to this intervention.



Seven persons with PLP were recruited (all male, age range 32–65 years, mean 52 years, standard deviation [SD] 11 years). Were included persons with a traumatic unilateral amputation of upper (above wrist), or lower (above ankle) limb (Table 1) who had been suffering from PLP daily (visual analog scale [VAS] = or > 3/10) for at least 6 months (at least one cumulated hour per day and interfering with daily activities). Persons with amputations from a vascular origin, a congenital amputation or limb agenesis, and patients with neurological or psychiatric disorders were excluded. Time since amputation ranged between 0.6 and 27.9 years (median = 1.7; range = 27.3). All participants were able to mentally move (a least partially) their phantom limb. Participants provided written informed consent which was approved by the Ethics Committee of the Institut de réadaptation en déficience physique de Québec, Québec, Canada. Participants were instructed not to modify their medication, including analgesics, or their prosthesis use during the duration of the study, and to inform the researchers in case of any change (prescribed or not) to their usual treatments, as the potential impact of these changes on the level of pain experienced by the participants would introduce an important confound. One of the seven participants stopped taking one of his regular analgesic medications (without medical advice), had an important change in prosthesis, and engaged in several new physical activities over the baseline and intervention phases of the study, and was thus excluded from the study, leaving a final sample of six participants.

View this table:
Table 1

Clinical characteristics of participants

PAgeDuration (Years)Upper/LowerLevel/LesionSideProsthesis (% Use in Daily Activities)Pain Before Amputation (Duration)Previous/Current Treatment
Ex59 1.4LUpper leg/amputationLMechanic (75% and +)NCurrent: anticonvulsant, anti-inflammatory, hypnotic agent, opioid analgesics
Previous: massage, mirror box, prosthesis
148 0.6UShoulder/amputationRAesthetic (25% or −)NCurrent: antidepressants
Previous: mirror box
252 5.8UUpper arm/amputationRMechanic (25–50%)NCurrent: anticonvulsant
Previous: TENS
332 1.7UElbow/BPA + amputationLMechanic (25% and +)Y (1 year +)Current: anticonvulsant
Previous: mirror box, cortical implant, spinal implant.
46027.9UWrist/amputationRNoNCurrent: ibuprophen
Previous: stellar infiltration, stump electrode, nerve removal.
565 8.4LKnee/amputationLMechanic (50–75%)NCurrent: opioid analgesics
Previous: acupuncture, hypnosis, relaxation/music
651 1.1LLower leg/amputationRMechanic (75% and +)NCurrent: ibuprophen
Previous: antidepressants
  • BPA = brachial plexus avulsion; Ex = excluded; L = left; L = lower; N = no; P = participant; R = right; TENS = transcutaneous electrical nerve stimulation; U = upper; Y = yes.

  • Characteristics of participants concerning age, amputation, pain level, prosthesis use and treatments previously tried are presented.

Study Design

Given the exploratory nature of this study, due in part to the scarcity of the population and the known inter-individual variability, a single-case multiple baseline ABA design was used to assess the individual effects of the intervention. Phase A was a baseline condition (3, 4, or 5 weeks). Phase B, the Intervention, was divided in two steps: Intervention 1 (laboratory and home; 4 weeks) and Intervention 2 (home only; 4 weeks). The second Phase A was the follow-up assessment performed 6 months post-Intervention 1 (FU-6 months). This design allowed the examination of group effects associated with the intervention as well as individual changes in pain levels that with such a small sample might be masked by group averages.

Material and Measures

Pain Scale

The main outcome was pain which was assessed with a 100-mm VAS. Participants reported their daily average pain in their logbook by marking the continuous horizontal line, ranging from “no pain” (extreme left) to “worst pain imaginable” (extreme right).

Videos of Movements

Forty-eight videos of movements including 16 different movements performed at three different speeds (movement performed within 4, 6, or 8 seconds) were created for each limb (right/left, lower/upper). The characteristics for upper and lower limb movements were matched as much as possible in terms of proximal vs distal involvement and of complexity (speed, objects use, etc.). Four different DVDs were created, one for each limb, so that participants could take them home. Videos and intervention instructions are available to researchers by contacting the corresponding author.

Imagery Rating of Videos

This numerical scale, used before and after Intervention 1, assessed the quality and ease to perform of imagery from 0 “unable to imagine” (no images) to 10 “I feel as if I was doing the movement” (see and feel movement).


The following questionnaires were administered before the baseline to describe the individual characteristics of the participants:

The Groningen Questionnaire [17,18]. This semi-structured interview designed for amputees was used to document the patients' pain and to describe individual differences concerning demographics, causes and level of amputation, the frequency and the type of suffering caused by PLP, the treatment received, and the prosthesis use. It was also administered at the end of the Intervention 2 to document potential changes in the frequency and the level of suffering associated to their PLP. The following question was added to this interview “Outside of the treatment periods, do you think that you execute more movements with your phantom limb now than before the intervention?” If the answer was positive, we asked to explain in more details how and when they performed phantom limb movements.

West Haven–Yale Multidimensional Pain Inventory Version 3.0 (WHYMPI) [19,20]. The WHYMPI assesses the general adjustment to chronic pain through nine scales: 1) the pain severity; 2) interference linked to pain; 3) life control perceived; 4) affective distress; 5) social support (response of the significant person to pain including); 6) punishment; 7) solicitous; and 8) distracting; the level of participation in activities of daily living like; 9) general activity level. Two composite scores––10) Dysfunctional and 11) Interpersonal Distress––are also produced.

Kinesthetic and Visual Imagery Questionnaire (KVIQ) [21]. This questionnaire measures imagery skills using 10 movements that participants are asked to perform both physically and mentally. Two subscales assess the performance during imagery regarding the clarity and vividness of the images (visual score) and the intensity with which the subject can feel the imagined movement (kinesthetic score). A total score is also computed to assess the global skill.

The Pain Catastrophizing Scale (PCS) [22]. The PCS was used to assess how people tend to perceive their painful situation as a disaster. It assesses rumination, magnification, helplessness, and also produces a total score.

The Pain Self-Efficacy Questionnaire (PSEQ) [23]. Participants indicated how confident they are when they perform activities or daily functions, despite the presence of pain as well as how they are coping with pain without medication.


At the study onset, the participants received a logbook in which they recorded their average daily pain levels, using the VAS, throughout the duration of the study: Baseline, Intervention 1, Intervention 2, and 1 month prior to FU-6.


Participants were randomly assigned to one of three baseline groups: 3 weeks (N = 2), 4 weeks (N = 3), or 5 weeks (N = 2). These variable baseline durations were used to ensure that any changes associated with the introduction of the intervention were not simply because of the amount of time participants had spent in the study group.

Intervention 1

Laboratory Intervention. During the first session, the participants observed the video of each of all 48 movements (corresponding to the amputated limb) on a 24” television screen. For each movement, they imagined performing the movement and then rated their difficulty imagining it using the Mental Imagery Scale. Based on these individual ratings, the 10 easiest movements were selected for intervention. This selection took into account the variety of levels (distal/proximal), speed, and presence or absence of objects. Laboratory interventions were conducted twice a week (weekdays, never on two consecutive days) for 4 weeks, for a total of eight sessions. For the first week, the intervention included four different movements and subsequently two new movements were added each week (up to 10 movements from week 4 to the end of the intervention) to increase task difficulty and number of movements performed (much as in a conventional motor training). This gradual increase was implemented to avoid tiring patients and reduce the short-term exacerbation of the pain that the intervention could produce. For each movement, the participant was first asked to observe it carefully while simultaneously performing it with his phantom limb. This active observation was repeated twice. Then, the participants had to imagine consecutively 10 times the same movement with their eyes closed while counting the number of repetitions aloud. Note that participants were not explicitly told to refrain from moving their stump which occurred slightly in some participants (see Discussion). After completion of the first iteration of all selected movements and a 5-minute break, the same procedure was performed again using the same movements but in the reverse order. Each laboratory intervention session lasted approximately 30 minutes.

Home Intervention. After the first intervention session, a DVD containing the videos and a list of movements to be practiced was given to the participants. A home intervention was proposed for the three weekdays during which they did not come to the laboratory. Participants had to repeat on their own the same procedure as during the intervention. Participants continued to keep track daily of their pain levels in a logbook as during baseline. They were asked to write comments about the effect of each movement on their phantom limb and rate their ease to imagine. This logbook was thus helpful to insure adherence to the proposed intervention as well as to gather additional clinical information.

Intervention 2 (Home Only)

After Intervention 1, participants were asked to continue the home intervention for 5 days a week for a period of four consecutive weeks.

Follow-up 6 Months

After Intervention 2, participants were asked to refrain from practicing the movements and from using the DVD until FU-6 months, but they were requested to rate their pain level daily for 1 month prior to the follow-up assessment which was scheduled 6 months after the end of Intervention 1 (FU-6).


Pain Ratings

Descriptive Analysis. The change in pain level was normalized against baseline using the following formula: ([post-intervention pain – pre-intervention pain]/pre-intervention pain) × 100. The pre-intervention pain was the average VAS rating (converted on a 100 points scale) for the last 7 days of the baseline period while the post-intervention pain was the average VAS rating for the last 7 days of Intervention 1, Intervention 2, or FU-6 months. The threshold of efficacy for this intervention was set at 30%, which was considered superior to the maximal placebo effect (<25%) observed in a non-pharmacologic randomized double-blind clinical trial conducted on PLP [24].

Intervention Time Series Analysis (ITSA). For each participant, ITSA using the autoregressive residual was carried out on data from the baseline, Intervention 1, and Intervention 2. Data from the FU-6 months could not be submitted to the same analysis because no continuous data were available between the end of Intervention 2 and this period. These analyses were performed, with the Autoreg procedure of SAS [25], to determine whether changes in pain were because of time and/or treatment. ITSA controls for the serial dependency of the measures repeated daily. Level and slope effects (sudden and gradual changes) were estimated as recommended by Huitema and McKean [26]. Extreme scores (outliers) were identified and their contribution to error variance was removed. To ensure that the residuals of the final models were normally distributed, they were reviewed visually and statistically.

Group Analyses. Because of the small sample size, we used non-parametric tests to examine changes in pain ratings. We first performed a Friedman test to determine whether changes in pain ratings occurred between baseline, Intervention 1 and Intervention 2. We also used Wilcoxon tests for post hoc comparisons between these three time points. An additional Wilcoxon test was used to determine if pain ratings at follow-up were significantly different from those obtained at baseline.


The quality of the imagery for each movement type (16) was computed before and after Intervention 1, by taking the average of the rating for the three different speeds of each movement. Wilcoxon tests were done, for each participant, to compare the rating of the 16 movements between the first and the last session of Intervention 1.

Predisposing Variables

In order to examine how individual differences contributed to explain the outcome following intervention, Spearman correlations between the total change in pain level (%) at the end of Intervention 2 and scores from the questionnaires at the beginning of the study were computed.


Pain Ratings

Descriptive Analysis

The largest decreases in PLP (ranging from 32% to 43%) were observed at the end of Intervention 2 for participants #2, #4, #5, #6. However, only participant #4 maintained his gains (34%) at FU-6 months (Figures 1 and 2). Participants #2 and #5 showed no clinically significant change and participant #6's pain increased (51%).

Figure 1

Mean of daily pain. The mean pain level was assessed every day, in the evening on a 100-mm visual analog scale. Duration of the baseline (B) period varied from 3 to 5 weeks. Note that y-axis scales are different for each patient. All patients received about 8 weeks of training with observation and imagination of movements on weekdays. For the first 4 weeks (Intervention 1 = I-1) the intervention was practiced 2 days at the laboratory and 3 days at home. For the last month (Intervention 2 = I-2), the intervention was practice 5 days only at home. Then the participants stopped rating their pain. One month before the 6-month follow-up (FU-6), they began to rate their pain daily. The FU-6 was performed 6 months after Intervention 1.

Figure 2

Change in pain (%) between baseline and other phases of the study. Only one participant had a reduction of 30% or more between baseline and Intervention 1. Four patients out of six exhibited a reduction on pain of 30% or more from baseline to the second phase of the intervention (Intervention 2). Six months after the first phase of intervention (FU-6), only one participant maintained his pain reduction. P = patient; FU-6 = 6-month follow-up.

Intervention Time Series Analysis (ITSA)

Table 2 illustrates the results of the ITSA, the number of outliers removed, the variance, and the degrees of freedom. Non-standardized regression coefficients in ITSA are interpreted much like coefficients of a linear regression and can be interpreted directly as changes in pain/100. The statistical modelization of the daily data explained on average 66.72% of the variance.

View this table:
Table 2

Results of the intervention times series analysis

Time0.245 (ns)1.482***0.006 (ns)−0.049 (ns)0.209 (ns)0.582***
Level I-1−2.192 (ns)−11.771**0.309 (ns)5.379**−10.556**1.674 (ns)
Level I-2−3.784 (ns)−7.365*2.351**1.250 (ns)1.108 (ns)−2.993 (ns)
Slope I-1−0.289 (ns)−1.673***−0.126*−0.530***−0.728**0.390**
Pain/week I-1−2.023−11.711−0.882−3.710−5.0962.730
Slope I-20.107 (ns)−0.076 (ns)0.175**0.048 (ns)0.734**0.200*
Pain/week I-20.749−0.5321.2250.3365.1381.400
  • * P = 0.05 < P< 0.1;

  • ** P< 0.05;

  • *** P< 0.01.

  • Intervention 1 = I-1; Intervention 2 = I-2; ns = not significant; R2 = variance; DF = degree of freedom.

  • These analysis produced independent results for each participant. These results represent the change of pain in mm on a 100 mm visual analog scale. The time effect is the daily tendency of the pain before the intervention. The level effect at Intervention 1 is the sudden change in pain level when the Intervention 1 began. The level effect at Intervention 2 is the sudden change in pain when the Intervention 2 began. The slope effect at Intervention 1 is the daily change of pain or the daily tendency of the pain during the Intervention 1. These results are transformed on a weekly tendency of pain. The slope effect at Intervention 2 is the daily tendency of the pain during the Intervention 2. These results are also transformed on a weekly tendency of pain. The R2, DF, and the number of outliers or extreme data are presented for each participant. The results show that three participants rated their pain gradually lower every day (or −3.7 to −11.7/100 every week) at Intervention 1.


Data from two participants (#2 and #6) displayed a significant time effect suggesting that, already at baseline, their pain tended, respectively, to increase and decrease each day.

Intervention 1

Two participants (#2 and #5) showed a significant negative level effect, which means that their pain decreased, when Intervention 1 began. Then, three participants (#2, #4, and #5) had a negative slope effect during Intervention 1, suggesting that participants rated their pain gradually lower, between 3.7 and 11.7/100 (mm) each week. Participant #3's pain tended also to decrease (P< 0.1). One participant (#6) showed a significant positive slope effect at Intervention 1, which means that the decrease in pain observed during Intervention 1 was inferior to the decrease in pain observed during baseline.

Intervention 2

When Intervention 2 started, one participant (#2) displayed another negative level effect where his pain decreased, and one participant (#3) had an increase of his pain. Then two participants (#3 and #5) had a gradual increase, two participants (#2 and #4) had no supplementary slope effect (but their pain continued to decrease), and one participant (#6) showed a positive slope effect because the slope was less important than at baseline, but his pain continued to decrease upon visual inspection (see Figure 1 and Table 2).

Group Results

Median and range are reported here for pain ratings obtained at baseline (median = 58.1; range = 56.7), Intervention 1 (median = 48.3; range = 56.0), Intervention 2 (median = 41.7; range = 66.5), and FU-6 months (median = 46.5; range = 42.2). The Friedman test revealed a significant main effect of time (χ2(2) = 6.33, P= 0.042). Post hoc Wilcoxon tests showed a significant reduction in pain from baseline to Intervention 1 (Z = −2.201, P= 0.028), and from baseline to Intervention 2 (Z = –1.992, P= 0.046). Pain ratings at follow-up were not significantly different from baseline (Z = −0.106, n.s.).


All participants, but one (#6), showed a significant improvement (ratings before Intervention 1 were lower than after Intervention 1) in the ease of imagery between pre- and post-Intervention 1 (P< 0.001) suggesting a generalized improvement in imagery after Intervention 1 for most of the 16 movements (Table 3).

View this table:
Table 3

Comparison between the imagery rating of movements before and after Intervention 1 and number of movements associated to each comparison

ParticipantNumber of Movements (N = 16)Imagery Rating/10Wilcoxon's Test
Post < PrePost = PrePost > PreMpreMpost?SEZP
67 451.310.440.880.521.6120.107
  • ? = Mpost – Mpre; SE = standard error.

  • The average of each of the 16 different movement types was computed by combining the rating for the three different speeds. Wilcoxon's test was done, for each participant, to compare the rating for the 16 movements between the first (pre) and last session (post) of Intervention 1. This table also shows the number of movements (out of a total of N = 16) for which the rating at the end of the intervention (post) was superior to that reported at baseline (pre). For the imagery ratings, the mean before the intervention (Mpre) and post-Intervention (Mpost), as well as the difference between the two (?) are indicated. The result of the Wilcoxon's test (Z) and the level of significance (p) are also presented. Five participants showed a significant difference between pre and post Intervention 1, suggesting a generalized improvement in imagery after 4 weeks of intervention for most of the 16 movements.

Predisposing Variables

A significant negative association was found between baseline score on the Life Control Scale and the decrease in pain after Intervention 2 (r= −1.00; P< 0.01) of the WHYMPI questionnaire suggesting that subjects who perceived having control over their life benefited more from the intervention. Then, a significant negative association was found between the baseline score on the Support Scale and the decrease in pain after Intervention 2 (r= −0.97; P< 0.01) which could mean that there is an association between the benefits of the intervention and the support that participants reported receiving by a significant person. In addition, a significant positive association was found between baseline score on the Dysfunctional Composite Score and the reduction in pain after Intervention 2 (r= 0.83; P< 0.05), suggesting that patients who are more functional and autonomous than the average person with chronic pain seemed to benefit more from this intervention. There was a trend showing that participants who were less affected by some Affective distress before the intervention displayed decrease of their PLP at Intervention 2 (r= 0.77; P= 0.07). The PCI, PSEQ, and KVIQ were not correlated significantly with the change in reported pain level.

Qualitative Results

Table 4 compares the frequency of the pain and the level of suffering associated with PLP, before baseline and after Intervention 2, and also indicates, based on documented clinical observations, whether the ability to move the phantom limb was improved following Intervention 2. Three participants (#2, #4, and #6) reported that the frequency of the PLP decreased and three (#1, #3, and #5) reported no change following Intervention 2. The global level of suffering related to the PLP was reduced for three participants (#1, #2, and #4), and remained the same for the others after Intervention 2. Thus overall, four of the six participants reported a decrease in the frequency or the suffering associated to their PLP after Intervention 2. In addition, all participants reported that they improved their ability to move their phantom limb at the end of the treatment.

View this table:
Table 4

Comparison of self-reported qualitative measures before baseline and after Intervention 2

ParticipantFrequency of the PainLevel of SufferingAbility to Move PL
Before BaselineAfter Intervention 2Before BaselineAfter Intervention 2After Intervention 2
1Few times per hourFew times per hourVery muchModeratelyImproved
2AlwaysFew times per hourMuchModeratelyImproved
3AlwaysAlwaysVery muchVery muchImproved
4AlwaysFew times per dayMuchModeratelyImproved
5AlwaysAlwaysVery muchVery muchImproved
6Few times per dayFew times per weekModeratelyModeratelyNot improved
  • PL = phantom limb.

  • Self-reported measures concerning the daily frequency of pain and the average level of suffering, which are taken from the French adaptation of the Groningen Questionnaire, are compared before baseline and after Intervention 2. After Intervention 2, an additional question concerning the perceived ability of participants to move their phantom limb was asked. Four of the six participants had a decrease of the frequency or the suffering associated to their PLP after the Intervention 2. All participants perceived that they improved their ability to move their PL.


The results of this study show that some persons with PLP may benefit from a new intervention consisting in repetitively imagining phantom limb movements previously observed in a video. Group statistics show that this intervention is associated with a significant decrease in pain intensity, with individual analyses showing pain reductions of up to 43% after 8 weeks. Although the group effect was not significant between the two intervention phases, individual analyses showed that for most participants, the decrease in pain level was more important after 8 weeks than after 4 weeks. The absence of a significant difference in pain ratings between baseline and follow-up suggests that on average, improvements were not maintained after the end of treatment. In fact, only one participant maintained his gains after 6 months. The Intervention Time Series Analysis indicated that three of our six participants rated their pain as gradually and significantly lower during Intervention 1. In fact, it revealed that decreases were more often gradual than sudden. The qualitative analysis of a semi-structured interview indicated that after 8 weeks of intervention, four of the six participants reported a decrease in the frequency or the level of suffering associated with their PLP. For one participant (#5), the qualitative data seemed inconsistent with pain ratings, as this participant reported no improvement in level of suffering despite reporting reduced pain levels. This may be because of the fact that the level of suffering presented in Table 4 reflects a global measure of suffering, including both intensity and frequency of pain, as well as level of functioning and psychological distress. The level of suffering reported by this patient may have reflected his psychological as well as his physical suffering.

These results are most interesting given that participants who had PLP for periods ranging from 5.8 to 27.9 yearshad tried several treatments without success (Table 1). The finding that participants with PLP may benefit from this novel intervention suggests that other factors or mechanisms than those targeted in other interventions are involved.

Adherence to treatment (imagining phantom movement) could explain why some benefited and others did not. Based on our results, however, most participants showed some improvement (statistically significant in all but #6) in the quality of their imagery whether or not they benefited from the intervention in terms of pain reduction. This suggests that they practiced the intervention as prescribed even if they had doubts about its success and while the pain ratings did not change, they did improve their imagery and execution of phantom movement. It is important to note, however, that as in the case of participants 1 and 3, an improvement in imagery abilities may not be sufficient to produce a significant decrease in pain intensity, frequency, or level of suffering. Some other variables are likely to be involved and may interfere with this ability. Additionally, a general improvement in imagery abilities may not be necessary for this intervention to be beneficial. Indeed, although participant #6 did not report an improvement in imagery, he still experienced a decrease in pain. Additional studies will be needed to better understand the specific mechanisms involved in this intervention.

Some individual factors were identified that could explain why this novel intervention was more beneficial for some participants than others. Participants who had a higher perceived control on their life, a better support of a significant person, and who were less dysfunctional benefited more from this intervention. According to clinical observations, it is possible that the quality of the efforts put in this kind of intervention can, like for most type of rehabilitation treatment, contribute to its outcome. Some of our preliminary findings, together with our clinical observations, suggest that personal factors of a psychological nature play a role in this type of intervention. Further studies should take these variables into consideration.

Other factors such as the type of approach and the duration of therapy could also be responsible for the high response rate (67%) in our study in comparison with the response rate observed (33%) in the Chan study [10]. Indeed, in the present study, we began to observe the beneficial effects more systematically after 8 weeks (30 minutes/day; total practice time: 20 hours) which is almost three times the amount of training received after 4 weeks of visualization (15 minutes daily; total practice time: 7 hours) in the Chan study. Further support for the importance of treatment duration or intensity comes from the MacIvert and collaborators [15] study that showed not only a high rate of PLP decrease (69% or 9/13) but also a higher decrease in pain levels (>50% vs 32–43%) using a visual and kinesthetic imagery intervention that was more intensive (40 minutes daily; total practice 28 hours/6 weeks) compared with ours (30 minutes; 20 hours/8 weeks). Thus, their findings suggest intensity and frequency of practice as potential contributing factors. Likewise, our results showed that only one participant maintained his gains after 6 months which suggests that either the intervention needs to be maintained for gains to persist or perhaps “booster” or maintenance sessions on a regular basis would be enough to maintain lowered pain levels.

In addition to therapy intensity, the novel approach used to promote the representation of phantom limb movements possibly contributed to the reduction in pain. Indeed, the observation of videos illustrating a variety of movements performed at specific speeds and specific range of movement prior to the imagining of the phantom limb movement likely contributed to generate a more vivid (both visually and kinesthetically) representation of movement during motor imagery than when no specific guidance is provided during the observation of still images. In addition, specific instructions to “imagine seeing and feeling the movement” being provided at every laboratory intervention session further ensured consistency of the imagery across intervention sessions. In fact, our intervention is closest to the mirror treatment condition of Chan and collaborators [10], which included the observation of the limb corresponding to their missing limb on a mirror or television (visual feedback) and the execution of the phantom movement (kinesthetic feedback). Finally, participants were allowed to contract stump muscles during phantom movements if it helped them to imagine movements. It was a clinical goal because we know that observation, execution, and imagination activate similar cerebral areas [6,14], and some patients reported that moving their stump helped them imagine the movement. After the intervention, participants generalized their ability to imagine or execute new movements with their phantom limb. This improvement of visual and kinesthetic feedback of their phantom limb movements during the intervention could reinforce the consistency between afference (internally seeing and feeling the limb) and the efferent motor command (to move the limb). This in turn could reactivate the motor representation of the missing limb, which may be responsible for a decrease in pain [12]. Further functional brain mapping experiments are necessary to corroborate this hypothesis.

One of the main limitations of this study was the absence of a control group or a placebo condition that would have allowed us to draw more definitive conclusions about the effectiveness of the intervention. Given the exploratory nature of our study and the difficulties associated with recruitment of a rare clinical population; however, we thought it would be preferable, at this point in the development of our intervention, to have all patients participate in the intervention. Further studies should, however, replicate our findings with a larger group, ideally in a randomized controlled trial.

Another limitation of this study is the fact that all participants were able to move their phantom limb to some extent at the beginning of the study, so it is possible that this intervention might only be effective in patients with this ability. However, even participants for whom it was difficult to move their phantom limb seem to have benefited from the intervention. For instance, participant #6 reported a low ability to imagine most of the 16 movements tested at the beginning of the study (Table 3), but still reported a decrease in pain after the intervention. This suggests that patients who have difficulty imagining movements can benefit from this intervention. This may be explained by the fact that the intervention is tailored to each patient by targeting the easiest movements for them to imagine. Further studies will be needed to examine this issue and to determine whether certain PLP patients might benefit from this intervention more than others.

In conclusion, some individuals with PLP seem to have benefited from a new imagery-based intervention that aimed at improving the motor representation of the missing limb. The effect was more important after 8 weeks but did not seem to last after a period of 6 months without practice, suggesting that this type of intervention needs time to be integrated and requires an active participation to maintain effect. The fact that this method could be used almost anytime and anywhere as it is a relatively simple and inexpensive method that patients learn quickly makes it a good potential adjunct to current treatment methods. More studies, including a randomized controlled trial, will be necessary to determine the exact nature of the mechanisms through which this intervention operates and to explore individual characteristics including those tested here that could have an impact on this intervention.


This study was made possible thanks to a grant from the Fonds de la Recherche en Santé du Québec (FRSQ) to PLJ and FM, including a scholarship to G.B, and salary grants from the FRSQ to PLJ and CM.

Conflicts of Interests

None of the authors have any financial or other relationship that might lead to a conflict of interest regarding this manuscript.


The authors are grateful to the following people for their precious help in data collection: Jean-Nicolas Carrier, Martine Fortin and Alexandra Gosselin. Special thanks to the participants who took part in this study.


View Abstract