The principle of mirror therapy (MT) is the use of a mirror to create a reflective illusion of an affected limb in order to trick the brain into thinking movement has occurred without pain. It involves placing the affected limb behind a mirror, which is sited so the reflection of the opposing limb appears in place of the hidden limb . A Mirror box is a device which allows the clinician to easily create this illusion. It’ is a box with one mirror in the center where on each side of it, the hands are placed in a manner that the affected limb is kept covered always and the unaffected limb is kept on the other side whose reflection can be seen on the mirror.
The patient places the good limb into one side, and the stump into the other. The patient then looks into the mirror on the side with good limb and makes “mirror symmetric” movements, as a symphony conductor might, or as we do when we clap our hands. Because the subject is seeing the reflected image of the good hand moving, it appears as if the phantom limb is also moving. Through the use of this artificial visual feedback it becomes possible for the patient to “move” the phantom limb, and to unclench it from potentially painful positions.
Evidence on effectiveness
Most of the evidence since the early work has come from case studies and anecdotal data.
- Chan et al (2007)  allocated 22 patients with Phantom Limb Pain (PLP) into a mirror therapy group, mental imagery group and a covered mirror group (control) . They reported that all patients in the mirror therapy group experienced reduced PLP. This was not the case in the other two groups. The study did not control potential biases and its methodology was not described in detail, so weakening the power of its findings.
- A more robust trial investigated two groups of subjects suffering with PLP . A mirror group were compared to a covered mirror group, however, there were no statistically significant reductions in PLP between groups.
- In 2011 a large scale review of the literature on mirror therapy by Rothgangel  summarized the current research as follows:”For stroke there is a moderate quality of evidence that MT as an additional intervention improves recovery of arm function, and a low quality of evidence regarding lower limb function and pain after stroke. The quality of evidence in patients with complex regional pain syndrome and phantom limb pain is also low. Firm conclusions could not be drawn. Little is known about which patients are likely to benefit most from MT, and how MT should preferably be applied. Future studies with clear descriptions of intervention protocols should focus on standardised outcome measures and systematically register adverse effects”.
- A further review  of current approaches in the treatment of PLP concluded that the benefits of mirror therapy appear to be limited to patients who suffer from cramping and muscular-type phantom pain . They noted that despite the findings of one RCT, there was no systematic evidence to support the use of this modality and even some suggestion it could be counterproductive. However, this may be due to study design, choice and size of sample and application of the modality .
- Diers et al (2010)  noted that applying MT as part of a sequence of modalities appeared to produce positive results against applying it in isolation . In an RCT, patients with CRPS1 and PLP showed decreases in pain, and improved function both immediately post treatment and at a 6 month follow-up when using mirror training as part of a sequence of modalities known as Graded Motor Imagery (GMI). Mirrored imagery alone did not, however, activate cortical processes in patients with phantom limb pain. The authors concluded that further research was required to establish the cortical processes underlying MT and motor imagery in order to guide the optimal method of application for these modalities.
- These findings appeared to support earlier suggestions that whilst mirrored movements may expose the cortex to sensory and motor input, the therapeutic effect is magnified if cortical networks were gradually activated using limb recognition, motor imagery and finally mirrored movement  . This sequence of modalities became known as GMI. Using a single blind randomised control design this approach was investigated with patients suffering with PLP, CRPS and brachial plexus avulsion. Whilst the heterogeneity of the sample was acknowledged, it was argued that cortical similarities exist between these conditions causing a cortical neglect of the affected limb leading to changes in cortical mapping. The sample size in each study was small, but both studies showed significant reductions in pain and cortical reorganisation following a six week program.