This study evaluates the effects of mirror therapy on upper-limb motor impairment in stroke patients early after their cerebrovascular accident. In recent years mirror therapy has been used in stroke rehabilitation, both to ease motor (e.g., upper limb impairment) and cognitive (e.g., spatial neglect) recovery. To note, mirror therapy is a simple and inexpensive treatment that patients can practice independently and with no significant side effects. However, a recent review concluded that the currently evidence available is not enough to determine about the actual effectiveness of mirror therapy in stroke survivors. Moreover, at our knowledge, the majority of studies recruited chronic stroke patients while only a few trials recruited patients within few weeks after stroke. Therefore, further research is encouraged particularly early after stroke.

In mirror therapy patients exercise their sound hand while it is reflected by a mirror placed at right angle to the patient's trunk. With this gambit, patients see two hands moving: their sound hand (i.e., the hand that is voluntarily moved) and the "avatar" of their impaired hand (i.e., the sound hand reflection in the mirror). In this assessor-blinded, randomized controlled trial half of participants receive mirror therapy .The other half receive sham therapy, in which the mirror is flipped so that the opaque surface face the sound arm. Mirror therapy and sham therapy are added to conventional rehabilitation.

In the current work, we investigate the efficacy of mirror therapy on upper-limb recovery in early post-stroke patients.

An assessor blind, randomized controlled study was conducted in the Casa di Cura Del Policlinico (CCP), an inpatient rehabilitation clinic in Milan. Stroke patients were referred to the study by their physician during their inpatient stay.

In addition to the intervention or control treatment, all patients participated to a conventional rehabilitation program consisting of physiotherapy (45 minutes per session, twice daily, five days per week) and occupational therapy (45 minutes per session, once a day, two to five days per week according to the physician prescription). Speech and language therapy and neuropsychological therapy were provided as needed. All treatments were one on one sessions.

Patients were measured at baseline and after five weeks, when treatments end. The assessors were blinded to group allocation.

Eligible patients were randomly assigned to either the intervention group or the control group. We used block randomization so that the number of participants was similar in the two groups (four patients per block, 1:1 ratio). A computer generated the randomization list and when a new eligible patient was recruited, a researcher contacts the person that allocate patient in mirror therapy or sham therapy group according to the randomization list. Patients were unaware of the group assignments (intervention vs. control) and of the alleged active treatment. All patients gave their informed written consent to participate in the study.

Patients of the intervention group received mirror therapy and patients of the control group received sham therapy, added to a conventional rehabilitation program. During mirror therapy, the patient was sitting on a conventional chair and placed her/his forearms on a table. A mirror (45 cm × 40 cm) was positioned between the two arms, at right angle with the patient's trunk. The reflective surface was oriented so that the participant could easily see the mirror image of his/her sound arm. During sham therapy, the mirror was flipped so that the opaque surface faced the sound arm.

Intervention and control group patients exercised the very same movements. In particular, movements were organized into three classes (simple, complex and functional movements). Examples of simple movements are the flexion-extension of the elbow with the pronated forearm or flexion-extension of the wrist. Complex movements were simple movements performed with the elbow flexed at 45° or simple movements performed with the elbow flexed at 45° and lifted from the table. Functional movements consisted in reaching, grasping and moving or using different objects (e.g., a pen, a tennis ball, a coin) Patients were asked to move their sound arm while looking the mirror reflective surface (intervention group) or the opaque surface (control group). Patients were also asked to stay still with the impaired arm.

Both mirror therapy and sham therapy consisted in one on one sessions (one therapist treated one patient), lasting 30 minutes each and administered once daily, five days per week for 30 days. From day 1 to 10, from day 11 to 20 and from day 21 to 30, patients practiced simple, complex and functional movements, respectively. In each session, ten different movements were practiced. Mirror therapy and sham therapy were administered in a quiet room close to the rehabilitation gym.

The study was powered to detect a clinically important difference of the main outcome. Sample size was calculated choosing a large effect size (Cohen's d = 0.9) and type 1 and type 2 error probabilities equal to 0.05 and 0.2, respectively. These parameters return a total sample size of 40 patients (13). Therefore, we planned to recruit 20 patients in each treatment group.

Counts, mean and standard deviation (SD) were used as descriptive statistics. Differences between the baseline characteristics of the intervention and control groups were tested using the two sample t-test and the Fisher's exact test (nominal data). The 0.95 confidence interval (0.95 CI) for matched samples was used for comparing the main and secondary outcomes before baseline and at week five (within group difference). The 0.95 CI for independent samples was used for testing differences in the main and secondary outcomes between the intervention and the comparison groups (mirror therapy vs. sham therapy, between groups difference).

A type 1 error probability equal to 0.05 was chosen. An intention-to-treat analysis was performed using the last observation carried forward method. In addition, the significance analysis was repeated after drop out removal. Statistical analyses were done in R 3.3.0 (R: A Language and Environment for Statistical Computing) with the ggplot and cowplot packages.

Inclusion Criteria:

• first ischemic or hemorrhagic stroke causing right or left hemiplegia or hemiparesis;
• enrolling in the trial within four weeks from the stroke;
• Mini Mental State Examination (MMSE) score ≥ 24, in order to exclude patients with significant cognitive decline;
• Token Test score < 40, in order to exclude patients with severe verbal comprehension deficits.

Exclusion Criteria:

• significant visual impairment despite glasses,
• cognitive deficits that could prevent patients from understanding the therapist instructions,
• an additional neurological or orthopedic disease (e.g., Parkinson's disease, limb amputation) known to cause a motor impairment for itself.