There are innumerable physical therapeutic approaches used to improve motor control in persons recovering from stroke. These therapies aim to improve the biomechanical performance of lost or limited movements,1 enhance the neuroplasticity mechanisms that occur after central nervous system lesions,2 avoid disuse, and promote function. Within this context, the discovery of mirror neurons in the 1990s led to the development of new therapeutic techniques and approaches based on cognitive strategies, such as the use of virtual reality, mirror therapy, action observation, and motor imagery (MI).3,4
Motor imagery entails the cognitive task of imagining the performance of a given movement or specific task without physically executing it.5,6 During MI, also known as “mental practice,” the mental imagery of the movement or task to be learned is systematically repeated.5 Studies using noninvasive brain stimulation techniques have allowed observation of cortical activation in motor areas during MI,7 and neuroimaging studies have shown that there is cortical reorganization after MI in persons with stroke.8
Over the past decade, there has been growing interest in observing the effects of the MI on the recovery of neurologic deficits such as those observed after stroke. Many clinical trials,8–15 systematic reviews,16–19 and meta-analyses6,20 have also highlighted the viability and benefits of MI for functional recovery when combined with conventional physical exercise protocols.19 However, a wide variety of intervention protocols exist in the literature, differing on aspects such as duration, frequency of MI exposure, and movements or tasks trained.19 In addition, few clinical trials using MI have strong methodological design despite recognition of the importance of methodological rigor and need for consensus related to the use of therapies in rehabilitation after stroke.1
Although previous systematic reviews and meta-analyses have assessed clinical trials of MI interventions, no meta-analysis globally assessing the topic with different outcomes such as gait, balance, activities of daily living, and recovery in the upper limb is available. Likewise, to our knowledge, this is the first attempt to describe the methodology used in clinical trials subdivided according to outcome measure investigated in poststroke recovery, facilitating understanding of the clinical impact and scientific relevance of MI. Considering the promising clinical relevance of MI, our research question was as follows: “In individuals after stroke, does MI training have efficacy for improving balance, activities of daily living, and upper limb and lower limb functions when compared with a control intervention?”
Identification and Selection of Studies
The present study constituted a systematic review and meta-analysis of randomized clinical trials investigating the efficacy of MI for recovery after stroke. Since efficacy can only be demonstrated when there is comparison with another intervention, all studies included were randomized trials. A qualitative analysis of randomized clinical trials selected using a systematic review based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) was carried out. The study was registered on PROSPERO under registration CRD42016041326.
In January 2016, a search, with no restriction on the language or publication date of articles, was conducted using several databases relevant in the context of the present scientific clinical investigation, namely, PubMed, ISI Web of Knowledge, Scopus, Cochrane, and Physiotherapy Evidence Database (PEDro). The key words related to the subject were selected on the basis of a review of the literature. The key words were then grouped into a single Boolean expression as follows:
(imagery OR “mental imagery” OR “kinesthetic imagery” OR “motor imagery” OR “visual imagery” OR “mental practice” OR “mental training” OR “mental rehearsal”) AND (stroke OR “cerebrovascular accident” OR “stroke rehabilitation” OR “stroke recovery” OR “hemiparesis” OR paresis OR ischemia) AND (randomized OR randomised OR trial OR “controlled trial”).
The search tool available for the PEDro database precluded the use of the Boolean expression.21 The methodological strategy employed in this case was to perform searches associated with the terms “clinical trial” with “imagery,” “mental practice,” “mental training,” and “mental rehearsal.”
Assessment of Characteristics of Studies
In Phase 1, searches for studies in the selected databases using the Boolean expression were carried out by 2 different researchers. Duplicate studies were then excluded, and the preliminary analysis based on titles and abstracts was performed. Subsequently, all studies representing entries in annals of congresses or book chapters, those not relevant to topic, observational studies, reviews, studies not investigating MI poststroke, and studies investigating other cognitive intervention strategies such as observation of movement or cognitive training were excluded. A manual search based on the references of other articles was also carried out.
Phase 2 entailed searching for the full texts of articles available for download. We excluded protocols of clinical trials, interventions with robotics or computer-brain interface, clinical trials with no control group or statistical analysis, and nonrandomized clinical trials.
Given the diversity of the studies retrieved, in Phase 3, the studies were subdivided according to the outcome measures assessed: Balance group—balance-related measures, ADL group—measures related to performance of activities of daily living (ADL), Lower Limb and Gait (LL/Gait) group—measures of motor performance of affected lower limb and gait performance, and Upper Limb (UL) group—measures of motor performance of affected upper limb. Some studies appeared in more than 1 subgroup because they reported data on more than 1 outcome. After subdivision, only studies containing mean and standard deviation data for the outcomes investigated were included. The PRISMA flowchart adopted is depicted in Figure 1. The data extracted from each study were as follows: (1) sample size; (2) time since stroke; (3) intervention protocol (type, frequency, duration, and follow-up); (4) scales or measuring instruments; and (5) result of interventions.
Given the type of intervention being investigated, the studies are not considered double-blind, as even when the assessor is blinded (ie, single-blind), participants are aware of the group to which they are allocated. Therefore, the risk of bias of each study was assessed using the Cochrane Back Review Group Criteria List for Methodological Quality Assessment,22 comprising 11 items investigating methodological aspects considered relevant in the quality of clinical trials. We selected this quality assessment tool, as we deemed it to be most applicable for studies having the methodological limitation of an absence of double-blinding. The quality analysis was initially performed by one assessor and then independently by another assessor. Any disparities in the conclusions of the quality analysis were then jointly examined by the 2 assessors, who came to consensus regarding the most appropriate score according to the criteria from the scale. On the basis of previous studies, a minimum cutoff score of 6 points was established, wherein studies with a score 6 or more points were considered to have the most adequate methodology.23 A list of methodological quality based on the score obtained on the PEDro scale is provided (see Supplemental Digital Content 2, http://links.lww.com/JNPT/A189).
The clinical trials included in our systematic review investigated the effect of MI on motor and sensory recovery of individuals clinically diagnosed with stroke; no restrictions on the type or time since neurovascular event were imposed.
The intervention protocol with MI had to entail individuals performing mental imagery of specific movements or tasks, without restriction on MI strategy (such as visual or kinesthetic imagery), or number of repetitions. Also, no restrictions were placed on the presence or absence of strategies promoting imagery, such as films, audio tapes, or images related to the movements or tasks to be imagined, among others.
The primary outcomes investigated in this review were the effects of MI both on performance of ADL and on motor function of balance, gait, and upper limbs.
The meta-analysis was performed using the RevMan 5.2 program of the Cochrane with comparison between the experimental (or intervention) group and the control group (other intervention or waiting list). In studies involving more than 1 control group, the analysis was done separately on the data from the experimental group with the respective control group.
The effect size for outcomes was retrieved from mean, standard deviation, and group size. Given the wide range of outcomes found, the meta-analysis was performed using 3 different analyses as follows. First, in Analysis 1, a meta-analysis including all the studies irrespective of quality and scale used, divided into major outcomes Balance, ADL, LL/Gait, and UL, was performed. Second, in Analysis 2, the studies were further subdivided according to the scale used (ie, Timed Up and Go [TUG], Action Research Arm Test [ARAT], Fugl-Meyer Lower Limb [FMLL] subscale. Finally, Analysis 3 included only studies with adequate methodological quality (≥6 points on the Cochrane scale) for each of the 4 outcome groups analyzed.24 Random-effects models were employed considering the possible heterogeneity in the scales used by the studies.
For the meta-analyses that compiled different scales, the size effect was calculated by the standardized mean difference (SMD) with its 95% confidence interval (CI). This approach enabled us to include different outcome measures in the same synthesis. For the meta-analyses assessing studies that used the same scales (ie, meta-analysis assessing studies that used the ARAT), the effect size was calculated by the raw mean difference, which is appropriate when all studies in the analysis used the same scale.25
Some scales measuring gait or upper limb function indicate improvement by increases in values (eg, gait speed), whereas other scales indicate improvement by decreases in values (eg, TUG). To adjust for the different scale directions, for those scales that report improvement with decreasing values, we transformed the values by multiplying the values by −1.26 Finally, funnel plots were produced to provide a more in-depth analysis of the heterogeneity of studies and publication bias (see Supplemental Digital Content 2, http://links.lww.com/JNPT/A189). In the funnel plots, there is an notable asymmetry. Although there are several causes of asymmetrical funnel plots, these findings may reveal the presence of bias or heterogeneity.
Flow of Studies Through Review
Of the 1156 articles found in the database, a total of 32 articles were included in the systematic review and analyzed for methodological quality.8–15,27–50 Details related to selection of articles are included in the PRISM flowchart (Figure 1).
Description of Studies
The general characteristics of the studies included are shown in Table 1. Of the 32 studies included, the majority (22 studies; 68.8%) were published between 2011 and 2015.
Quality: Risk of Bias in Individual Studies
The methodological quality, together with final scores, of each article is given in Table 2. Although none of the studies attained a maximum score of 11, owing to the fact that none employed double-blinding, 13 (40.6%) articles had a score of 6 or more, indicating adequate methodological quality.* The highest score (9 points) on the Cochrane scale was achieved by the study of Verma et al.15
The total sample comprised 955 persons with post–ischemic or hemorrhagic stroke seen in the acute/subacute (59.3% with <6 months) phase of stroke. Two studies did not report the time elapsed since neurovascular injury.31,32 The population studied showed motor (29 studies)8–15,28–45,47–49 and sensory/perceptive (3 studies)27,46,50 impairments after neurovascular injury.
In most of the studies, participants performed MI in association with conventional or traditional physical and/or occupational therapy. Only the study by Dickstein et al12 used MI alone as an intervention for the experimental group and did not describe combining MI with any other intervention.
Large differences in exposure times of MI protocols were noted. A total of 11 (34.4%) studies used 30 minutes of MI,† whereas the minimum time was 5 minutes (in only 1 study).34 There were also differences in the number of intervention sessions in each study. The most common number of sessions reported was 12, adopted in 9 (28.1%) studies.‡ In 12 (37.5%) studies, persons with stroke received the intervention for 4 weeks,§ whereas in 9 (28.1%) studies, the intervention period was 6 weeks.¶ With regard to the type of MI applied, 13 (40.6%) studies∥ used internal imagery perspective of movements or specific tasks, 11 (34.4%) studies used both internal and external perspectives,# 1 study used the external imagery perspective alone,43 and 7 studies did not specify perspective.9,13,15,36,37,41,47
Regarding the outcome measures investigated, 5 studies involving a total sample of 272 persons with stroke sought to assess the effects of the intervention on performance of ADL.10,13,30,36,37 Seven studies aimed at investigating the Balance group before and after the intervention, involving a total sample of 206 persons with disabilities after stroke.9–12,29,34,48 Twelve studies investigated motor performance of lower limb and/or gait in an overall sample of 343 persons.** After dividing all studies into subgroups, 9 studies were found to assess two or more outcomes (e.g. gait and balance).9–12,29,30,34,36,37
Effect of Intervention
Concerning the domains of effects investigated by the studies, the domains of activity and of body structure and function were covered, reflecting a large variability in the measures adopted by the studies. Five (15.6%) studies assessed the effects of the intervention on performance of activities of daily living,10,13,30,36,37 12 (37.5%) studies†† assessed the effects on motor recovery of lower limb and/or gait, and, finally, and 20 (62.5%) studies assessed performance and motor recovery of upper limb.‡‡
With regard to the outcomes of the included studies, positive results (ie, the MI group with better outcomes than the control group) were found in 50% of the studies from the subgroup for activities of daily living, 85.7% of the Balance subgroup, 92.3% from the LL/Gait subgroup, and 80.9% of the studies from the UL group. Three studies reported both positive and neutral outcomes, whereas none of the studies had results where measures were worse following the MI intervention.
Conflicts of Interest of the Studies Included
Of the 32 studies included, only 9 (28.1%) studies§§ declared no conflicts of interest, whereas 23 studies did not describe conflicts of interest in statement.¶¶ Tables containing the conflicts of interest and financing of the studies included and the PEDro and Cochrane ratings are provided as supplementary material (see Online PDF, Supplementary Digital Content 2, http://links.lww.com/JNPT/A191).
In the meta-analysis, 9 studies were excluded from the UL subgroup for not containing the necessary data on statistical tests (mean and standard deviation). Two studies were excluded from each of the other subgroups, ADL,13,37 Balance,12,48 and LL/Gait,13,38 for the same reason13 excluded studies.13 On the calculation of effect size, a positive sign indicated when the experimental group showed greater effects than the control group, based on the particulars of the specific outcome measures used by each study.
Figure 2 illustrates the outcomes of Analysis 1. In these forest plots, lines that fall on the right-side of the graph, represent participants who received the intervention in one particular study and showed significant positive changes compared to control participants. The “black diamonds” are the average effect size of all trials and should be interpreted the same way. Significant differences were found between the MI group and the control group for 3 outcomes: Balance (SMD = 1.78 [95% CI, 0.47-3.10], P = 0.0008, I 2 = 93%) (Figure 2A), LL/Gait (SMD = 0.70 [95% CI, 0.23-1.17], P = 0.003, I 2 = 82%) (Figure 2C), and UL (SMD = 0.36 [95% CI, 0.16-0.55], P = 0.0004, I 2 = 30%) (Figure 2D). The outcome of the ADL group was not significant (SMD = 0.35 [95% CI, −0.24 to 0.95], P = 0.24, I 2 = 79%) (Figure 2B). However, the studies were highly heterogeneous, as evidenced by the I 2 values.
In Analysis 2, statistically favorable differences were found for the MI group relative to the control group on the balance assessment using the Functional Reach Test (FRT) (MD = 8.82 [95% CI, 6.16-11.49], P < 0.00001, I 2 = 0%) but not the Berg Balance Scale (MD = 4.98 [95% CI, −4.36 to 14.32], P = 0.30), I 2 = 93%). On the assessment of performance in gait, a significant difference was observed on the assessment by the TUG (MD = −4.43 [95% CI, −7.66 to −1.19], P = 0.007, I 2 = 57%) and gait speed (MD = 0.49 [95% CI, 0.09-0.89], P = 0.02, I 2 = 0%) but not for the 10-Meter Walk Test (MD = 0.06 [95% CI, −3.38 to 3.50], P = 0.97, I 2 = 49%). Regarding the instruments used in the studies for assessing motor performance of the UL group, a statistical difference was detected for the ARAT (MD = 4.80 [95% CI, 2.47-7.13], P < 0.0001, I 2 = 0%) and the Fugl-Meyer Upper Limb (FMUL) subscale (MD = 3.94 [95% CI, 0.76-7.12], P = 0.02, I 2 = 0%), along with a lower heterogeneity.
In Analysis 3, for the balance outcome was not possible because only one study, by Braun et al,10 had a quality score of 6 or more on the Cochrane scale. However, all the studies included in this review that investigated the effects of MI on activities of daily living had adequate methodological quality and the Analysis 3 found no significant difference between the intervention group and the control group (SMD = 0.35 [95% CI, −0.24 to 0.95], P = 0.24, I 2 = 79%) (Figure 3A). Similarly, for the other outcomes, no significant differences were found between the experimental group (MI) and the control group: LL/Gait (SMD = 0.67 [95% CI, −0.20 to 1.54], P = 0.13, I 2 = 85%) and UL (SMD = 0.17 [95% CI, −0.07 to 0.40], P = 0.16, I 2 = 27%) (Figures 3B and 3C).
The findings of our systematic review and meta-analysis revealed a wide range of MI protocols and a large amount of heterogeneity in the methodological quality of the studies. The study samples included participants at different times after stroke and used a large variety of assessment methods. While the studies included in the systematic review showed predominantly positive results of the MI interventions, the overall meta-analysis of studies (which excluded lower-quality studies) found no significant differences compared with controls (Analysis 3). However, when including low-quality studies (Analysis 1) or on subanalyses (Analysis 2), positive results were found for both lower limb/gait (shorter performance time on the TUG and increased gait speed) and the upper limb (improved motor performance on the ARAT and FMUL) as well as balance (balance gains on the FRT).
Our results are comparable with those of the previous meta-analyses, which found differences for the ARAT subanalysis.6,20 When only high-quality studies were included, our results are also consistent with another meta-analysis that found no changes in upper limb function in the overall analysis.18 Akin to previous meta-analyses,6,18,20 a relatively small number of studies compared with those initially identified can be included in the final analysis when methodological quality and date availability are taken into account.
With regard to MI protocol used, it was not possible to determine the number of repetitions of imagery applied for the movements or specific tasks to be imagined in the studies because only the total duration of the intervention was reported. Given that MI requires orientation time and a facilitation strategy for the movement or task to be imagined, which in some studies led to the use of videos or audio, failure to describe the time or exact number of repetitions of imagery performed by the volunteer hampers the development of more structured protocols for use of MI.
Beyond the number of repetitions, another noteworthy deficit of the studies is the lack of monitoring of participants during imagery of the movements. Although MI is a cognitive task that requires concentration and whose training can produce autonomic changes,51 the studies dedicated little attention to monitoring parameters such as heart rate and respiration frequency of participants during MI, which could demonstrate whether the imagery was being performed correctly. In view of evidence showing differences in imagery ability among individuals irrespective of neurologic impairments,52 and that synaptic compromise can affect imagery ability,53 it is notable that not all studies took this criterion into account by using specific questionnaires. This fact might explain the neutral results of some studies and also the findings of the studies analyzed in the present meta-analysis.
With regard to time elapsed since stroke, our data were derived from studies that included participants of all phases poststroke, although most were in the acute and subacute phases. The predominance of studies in the acute and subacute phases was also found in the meta-analysis by Cha et al20 investigating the effect of MI combined with functional task training. In previous systematic reviews, such as that of García Carrasco and Aboitiz Cantalapiedra17 and meta-analyses including those by Barclay-Goddard et al,16 Kho et al,6 and Machado et al,18 the studies of participants in the chronic phase of stroke predominated.
With regard to the quality of the studies, methodological weaknesses still prevail in this area, where 59.3% (19) of the studies failed to attain an adequate score on the Cochrane scale. Some outcomes such as balance were associated with lower-quality studies than other outcomes.9,34 Although evidence exists showing that the PEDro scale provides a better measure of methodological quality than the Jadad scale54 and the former has been used in previous meta-analyses and systematic reviews, we chose the Cochrane scale over other scales because it is less influenced by the double-blinding issue.
Our findings reveal that there remains a lack of standardization in the MI study protocols and other aspects of this area of study, which may be responsible for some of the conflicting results between trials and meta-analyses. Another unresolved question is the timepoint at which MI should be introduced. Several trials started MI in early phases poststroke whereas others included participants in the chronic phase, which could also be responsible for differences in findings. Likewise some meta-analyses have included more studies of participants with acute/subacute stroke than others, depending on their search methods and inclusion and exclusion criteria. Motor imagery may have more value in acute and subacute stroke, but at present, there is insufficient evidence to determine this. Overall, the evidence suggests that MI may have value for persons recovering from stroke, particularly for specific outcomes. However, a larger number of rigorous studies are needed.
The present study has several limitations including the small number of studies included in the analysis. The analysis included few studies because methodological quality and data availability were inclusion criteria, leading to the exclusion of all studies with measures investigating balance. Another limitation was the selection of studies whose abstract or title were in English, where the key word/descriptors chosen for the search may have excluded studies that met the other inclusion criteria from the analysis. Finally, some article databases such as EMBASE were not searched, where this may have led to the exclusion of some studies.
There is a high heterogeneity among studies investigating MI for recovery of persons poststroke, in terms of both protocols and intervention time. The majority of studies included in the systematic review showed benefits of the MI, albeit with a large proportion of low-quality studies. The meta-analysis of all studies regardless of quality revealed significant differences on overall analysis for 3 outcomes (balance, lower limb/gait, and upper limb). However, when only high-quality studies were included, no significant difference between the MI and control groups was found. On the subgroup analysis, MI was associated with shorter performance time for the TUG, increased gait speed, improved motor performance on the ARAT and FMUL, as well as balance gains on the FRT, relative to control groups. Further high-quality studies and greater standardization of these interventions are needed in this field of research.
1. Winstein CJ, Stein J, Arena R, et al Guidelines for adult stroke rehabilitation
and recovery: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2016;47(6):e98–e169. doi:10.1161/STR.0000000000000098.
2. Murphy TH, Corbett D. Plasticity during stroke recovery: from synapse to behaviour. Nat Rev Neurosci. 2009;10:861–872.
3. McEwen SE, Huijbregts MP, Ryan JD, Polatajko HJ. Cognitive strategy use to enhance motor skill acquisition post-stroke: a critical review. Brain Inj. 2009;23:263–277.
4. Peters HT, Page SJ. Integrating mental practice
with task-specific training and behavioral supports in poststroke rehabilitation: evidence, components, and augmentative opportunities. Phys Med Rehabil Clin N Am. 2015;26:715–727.
5. Jackson PL, Lafleur MF, Malouin F, Richards C, Doyon J. Potential role of mental practice
using motor imagery in neurologic rehabilitation. Arch Phys Med Rehabil. 2001;82:1133–1141.
6. Kho AY, Liu KP, Chung RC. Meta-analysis on the effect of mental imagery on motor recovery of the hemiplegic upper extremity function. Aust Occup Ther J. 2014;61:38–48.
7. Loporto M, McAllister C, Williams J, Hardwick R, Holmes P. Investigating central mechanisms underlying the effects of action observation and imagery through transcranial magnetic stimulation. J Motor Behav. 2011;43:361–373.
8. Sun L, Yin D, Zhu Y, et al Cortical reorganization after motor imagery training in chronic stroke patients with severe motor impairment: a longitudinal fMRI study. Neuroradiology. 2013;55:913–925.
9. Bae YH, Ko Y, Ha H, Ahn SY, Lee W, Lee SM. An efficacy study on improving balance and gait in subacute stroke patients by balance training with additional motor imagery: a pilot study. J Phys Ther Sci. 2015;27:3245–3248.
10. Braun SM, Beurskens AJ, Kleynen M, Oudelaar B, Schols JM, Wade DT. A multicenter randomized controlled trial to compare subacute “treatment as usual” with and without mental practice
among persons with stroke in Dutch nursing homes. J Am Med Dir Assoc. 2012;13:85.e1–85.e7.
11. Cho HY, Kim JS, Lee GC. Effects of motor imagery training on balance and gait abilities in post-stroke patients: a randomized controlled trial [with consumer summary]. Clin Rehabil. 2012;27(8):675–680.
12. Dickstein R, Deutsch JE, Yoeli Y, et al Effects of integrated motor imagery practice on gait of individuals with chronic stroke: a half-crossover randomized study. Arch Phys Med Rehabil. 2013;94:2119–2125.
13. Liu KP. Use of mental imagery to improve task generalisation after a stroke. Hong Kong Med J. 2009;15:37–41.
14. Page SJ, Levine P, Leonard AC. Effects of mental practice
on affected limb use and function in chronic stroke. Arch Phys Med Rehabil. 2005;86:399–402.
15. Verma R, Arya K, Garg RK, Singh T. Task-oriented circuit class training program with motor imagery for gait rehabilitation in poststroke patients: a randomized controlled trial. Top Stroke Rehabil. 2011;18:620–632.
16. Barclay-Goddard RE, Stevenson TJ, Poluha W, Thalman L. Mental practice
for treating upper extremity deficits in individuals with hemiparesis after stroke. Cochrane Database Syst Rev. 2011;(5):CD005950.
17. García Carrasco D, Aboitiz Cantalapiedra J. Effectiveness of motor imagery or mental practice
in functional recovery after stroke: a systematic review. Neurologia. 2016;31:43–52.
18. Machado S, Lattari E, de Sa AS, et al Is mental practice
an effective adjunct therapeutic strategy for upper limb motor restoration after stroke? A systematic review and meta-analysis. CNS Neurol Disord Drug Targets. 2015;14:567–575.
19. Carrasco DG, Cantalapiedra JA. Effectiveness of motor imagery or mental practice
in functional recovery after stroke: a systematic review. Neurología. 2016;31(1):43–52.
20. Cha YJ, Yoo EY, Jung MY, Park SH, Park JH. Effects of functional task training with mental practice
in stroke: a meta analysis. NeuroRehabilitation. 2012;30:239–246.
21. Pohl S, Zobel J, Moffat A. Extended Boolean retrieval for systematic biomedical reviews. In: Proceedings of the Thirty-Third Australasian Conference on Computer Science. Vol 102. Brisbane/Queensland/Australia: Australian Computer Society Inc; 2010:117–126.
22. Berger VW, Alperson SY. A general framework for the evaluation of clinical trial quality. Rev Recent Clin Trials. 2009;4:79.
23. Candy B, Jones L, Varagunam M, Speck P, Tookman A, King M. Spiritual and religious interventions for well-being of adults in the terminal phase of disease. Cochrane Database Syst Rev. 2012;(5):CD007544.
24. van Tulder M, Furlan A, Bombardier C, Bouter L. Updated method guidelines for systematic reviews in the Cochrane Collaboration Back Review Group. Spine (Phila Pa 1976). 2003;28:1290–1299.
25. Sedgwick P, Marston L. Meta-analyses: standardised mean differences. BMJ. 2013;347:f7257.
26. Hemmingsen B, Krogh J, Metzendorf MI, Richter B. Sodium-glucose cotransporter (SGLT) 2 inhibitors for prevention or delay of type 2 diabetes mellitus and its associated complications in people at risk for development of type 2 diabetes mellitus. Cochrane Database Syst Rev. 2016;4:CD012106.
27. Ferreira HP, Leite Lopes MA, Luiz RR, Cardoso L, Andre C. Is visual scanning better than mental practice
in hemispatial neglect? Results from a pilot study. Top Stroke Rehabil. 2011;18:155–161.
28. Grabherr L, Jola C, Berra G, Theiler R, Mast FW. Motor imagery training improves precision of an upper limb movement in patients with hemiparesis. NeuroRehabilitation. 2015;36:157–166.
29. Hwang S, Jeon HS, Yi CH, Kwon OY, Cho SH, You SH. Locomotor imagery training improves gait performance in people with chronic hemiparetic stroke: a controlled clinical trial. Clin Rehabil. 2010;24:514–522.
30. Letswaart M, Johnston M, Dijkerman HC, et al Mental practice
with motor imagery in stroke recovery: randomized controlled trial of efficacy. Brain. 2011;134:1373–1386.
31. Kim SS, Lee BH. Motor imagery training improves upper extremity performance in stroke patients. J Phys Ther Sci. 2015;27:2289–2291.
32. Kim SS, Lee HJ, You YY. Effects of ankle strengthening exercises combined with motor imagery training on the Timed Up and Go test score and weight bearing ratio in stroke patients. J Phys Ther Sci. 2015;27(7):2303–2305.
33. Lee GC, Song CH, Lee YW, Cho HY, Lee SW. Effects of motor imagery training on gait ability of patients with chronic stroke. J Phys Ther Sci. 2011;23(2):197–200.
34. Lee H, Kim H, Ahn M, You Y. Effects of proprioception training with exercise imagery on balance ability of stroke patients. J Phys Ther Sci. 2015;27:1–4.
35. Liu H, Song LP, Zhang T. Mental practice
combined with physical practice to enhance hand recovery in stroke patients. Behav Neurol. 2014;2014:876416.
36. Liu KP, Chan CC, Lee TM, Hui-Chan CW. Mental imagery for promoting relearning for people after stroke: a randomized controlled trial. Arch Phys Med Rehabil. 2004;85:1403–1408.
37. Liu KP, Chan CC, Wong RS, et al A randomized controlled trial of mental imagery augment generalization of learning in acute poststroke patients. Stroke. 2009;40:2222–2225.
38. Malouin F, Richards CL, Durand A, Doyon J. Added value of mental practice
combined with a small amount of physical practice on the relearning of rising and sitting post-stroke: a pilot study. J Neurol Phys Ther. 2009;33:195–202.
39. Muller K, Butefisch CM, Seitz RJ, Homberg V. Mental practice
improves hand function after hemiparetic stroke. Restor Neurol Neurosci. 2007;25:501–511.
40. Nilsen DM, Gillen G, di Russo T, Gordon AM. Effect of imagery perspective on occupational performance after stroke: a randomized controlled trial [with consumer summary]. Am J Occup Ther. 2012;66(3):320–329.
41. O'Brien J, Martyn Bracewell R, Castillo M, Juan JA. The role of kinaesthetic motor imagery in promoting activation of the lumbrical muscles in the hemiparetic hand: a randomised controlled study. Rev Cienc Salud. 2011;9:5–16.
42. Oostra KM, Oomen A, Vanderstraeten G, Vingerhoets G. Influence of motor imagery training on gait rehabilitation in sub-acute stroke: a randomized controlled trial. J Rehabil Med. 2015;47:204–209.
43. Page SJ. Imagery improves upper extremity motor function in chronic stroke patients: a pilot study. Occup Ther J Res. 2000;20:200–215.
44. Page SJ, Dunning K, Hermann V, Leonard A, Levine P. Longer versus shorter mental practice
sessions for affected upper extremity movement after stroke: a randomized controlled trial. Clin Rehabil. 2011;25:627–637.
45. Page SJ, Levine P, Leonard A. Mental practice
in chronic stroke: results of a randomized, placebo-controlled trial. Stroke. 2007;38:1293–1297.
46. Park JH, Lee JH. The effects of mental practice
on unilateral neglect in patients with chronic stroke: a randomized controlled trial. J Phys Ther Sci. 2015;27:3803–3805.
47. Riccio I, Iolascon G, Barillari MR, Gimigliano R, Gimigliano F. Mental practice
is effective in upper limb recovery after stroke: a randomized single-blind cross-over study. Eur J Phys Rehabil Med. 2010;46:19–25.
48. Schuster C, Butler J, Andrews B, Kischka U, Ettlin T. Comparison of embedded and added motor imagery training in patients after stroke: results of a randomised controlled pilot trial. Trials. 2012;13:11.
49. Timmermans AA, Verbunt JA, van Woerden R, Moennekens M, Pernot DH, Seelen HA. Effect of mental practice
on the improvement of function and daily activity performance of the upper extremity in patients with subacute stroke: a randomized clinical trial. J Am Med Dir Assoc. 2013;14:204–212.
50. Welfringer A, Leifert-Fiebach G, Babinsky R, Brandt T. Visuomotor imagery as a new tool in the rehabilitation of neglect: a randomised controlled study of feasibility and efficacy. Disabil Rehabil. 2011;33:2033–2043.
51. Thill EE, Bryche D, Poumarat G, Rigoulet N. Task-involvement and ego-involvement goals during actual and imagined movements: their effects on cognitions and vegetative responses. Behav Brain Res. 1997;82:159–167.
52. Hall C, Pongrac J, Buckholz E. The measurement of imagery ability. Hum Mov Sci. 1985;4:107–118.
53. Sharma N, Pomeroy VM, Baron J-C. Motor imagery a backdoor to the motor system after stroke? Stroke. 2006;37:1941–1952.
54. Bhogal SK, Teasell RW, Foley NC, Speechley MR. The PEDro scale provides a more comprehensive measure of methodological quality than the Jadad scale in stroke rehabilitation
literature. J Clin Epidemiol. 2005;58:668–673.
*References 10, 15, 27, 30, 36, 37, 40, 41, 44, 47–50.
†References 8, 14, 28, 29, 31, 33, 37, 39, 42, 45, 50.
‡References 9, 12, 14, 30, 31, 38, 40, 43, 45.
§References 8, 9, 12, 29–32, 35, 38, 39, 43, 46.
¶References 10, 11, 14, 15, 33, 40, 42, 45, 49, 50.
∥References 8, 14, 28, 30, 32, 35, 38, 39, 45, 46, 48–50.
#References 10–12, 27, 29, 31, 33, 34, 40, 42, 44.
**References 9–12, 15, 29, 32–34, 37, 38, 42.
††References 9–12, 15, 29, 32–34, 37, 38, 42.
‡‡References 8, 10, 14, 27, 28, 30, 31, 35–37, 39–41, 43–47, 49, 50.
§§References 8, 10, 12, 14, 28, 35, 36, 48, 49.
¶¶References 9, 11, 13, 15, 27, 29–34, 37–47, 50.