Abstract
Keywords
People who have sustained a stroke are often left with residual motor impairments that limit their ability to engage in meaningful occupations such as self-care, work, and leisure activities. Sensorimotor deficits in the upper limb, such as weakness (Bourbonnais & Vanden Noven, 1989; Dewald & Beer, 2001; Tyson, Chillala, Hanley, Selley, & Tallis, 2006), decreased speed of movement (Trombly, 1992), decreased angular excursion and impaired temporal coordination of the joints (Cirstea & Levin, 2000; Levin, 1996), impaired upper-limb and trunk coordination (Cirstea & Levin, 2000; Michaelsen, Luta, Roby-Brami, & Levin, 2001), and impaired hand function (Lang, Wagner, Edwards, Sahrmann, & Dromerick, 2006; Welmer, Holmqvist, & Sommerfeld, 2008), may limit the ability to use the affected upper extremity to perform various tasks.
Consequently, occupational therapists working with this population use approaches that aim at optimizing motor behavior to restore occupational performance. Treatment interventions such as materials-based occupations (Lang, Nelson, & Bush, 1992), constraint-induced movement therapy (Wolf et al., 2006), modified constraint-induced movement therapy (Lin, Wu, Wei, Lee, & Liu, 2007), and task-related (Thielman, Dean, & Gentile, 2004; Thielman, Kaminski, & Gentile, 2008) or task-specific training (Ada, Canning, Carr, Kilbreath, & Shepherd, 1994; Michaelsen, Dannenbaum, & Levin, 2006) are common training methods for remediating impairments and restoring function in the upper limb. These training methods stress the person’s active participation, manipulation of goal-oriented tasks or environmental characteristics to drive motor behavior, and practice of the whole task or components of the task under varying conditions.
Although after stroke, people appear to benefit from substantial time spent in practice (Kwakkel, 2006; Wolf et al., 2006), it appears they may not be getting enough of it. When compared with other patient populations, patients who have sustained a stroke spend more time alone and inactive on rehabilitation units (Bear-Lehman, Bassile, & Gillen, 2001; Bernhardt, Dewey, Thrift, & Donnan, 2004). Thus, there appears to be a “practice gap” between the amount of training these patients need and the amount they receive. What accounts for this practice gap is unclear. Many factors may contribute, but it is likely that a reduced sensorimotor capacity plays a substantial role in this inactivity (Bear-Lehman et al., 2001). Therefore, it is of benefit to investigate adjuncts to physical practice that may assist in filling the gap. Recently, the literature has proposed the use of mental practice as an addition to physical practice or when physical practice is not possible (Jackson, Lafleur, Malouin, Richards, & Doyon, 2001).
Mental practice is a training method during which a person cognitively rehearses a physical skill using motor imagery in the absence of overt, physical movements for the purpose of enhancing motor skill performance (Annett, 1995; Driskell, Copper, & Moran, 1994; Feltz & Landers, 1983; Jackson et al., 2001). The use of mental practice to benefit motor skill performance in healthy people has been well established in the literature (Driskell et al., 1994; Feltz & Landers, 1983).
Why does mental practice work? Simulation Theory, elaborated by Jeannerod (2001), suggests that actions have a covert stage. This covert stage is a representation of the action, which includes the purpose of the action, the information needed to produce it, and the possible outcomes. The theory suggests that similarities exist in terms of neural activity between the state in which an action is simulated and the state of execution and that this similarity results in the benefits seen in mental practice (Jeannerod, 2001).
Multiple lines of evidence in the literature converge to suggest, as proposed by Simulation Theory, that imagined and executed actions share similar processes. For instance, at the behavioral level, several commonalities between the two have been observed. The contextual interference effect, which states that blocked practice facilitates skill acquisition and random practice benefits retention, and Fitts’ Law (Fitts, 1954, as cited in Magill, 2001), which states that movement times increase as target distance increases or target width decreases (i.e., speed–accuracy trade-off), are two robust findings in the motor learning and control literature (Magill, 2001). Interestingly, both findings appear to hold true when movements are imagined (Decety & Jeannerod, 1996; Gabriele, Hall, & Lee, 1989; Sirigu et al., 1996). The notion that healthy people’s physical and mental actions share similar durations is another behavioral observation suggesting that the two may be functionally equivalent (Decety & Jeannerod, 1996; Sirigu et al., 1996). For example, a review of the duration of mental movements found temporal equivalence for reaching; grasping; writing; and cyclical activities, such as walking and running (Guillot & Collet, 2005). Additional evidence for a similarity between the two comes from literature that has shown that physiological correlates for imagined and executed actions are comparable. When people engage in mental practice, changes in vegetative functions (Decety, Jeannerod, Germain, & Pastene, 1991; Livesay, & Samaras, 1998; Roure et al., 1998), muscle activity (Guillot et al., 2007; Livesay & Samaras, 1998), and muscle strength (Fontani et al., 2007; Sidaway & Trzaska, 2005; Yue & Cole, 1992) occur in a fashion comparable to that seen during physical activity. Perhaps the most convincing line of evidence comes from studies investigating the neural correlates of imagined and executed actions in healthy people. Studies using positron emission tomography (Lafleur et al., 2002; Stephan et al., 1995), functional magnetic resonance imaging (Ehrsson, Geyer, & Naito, 2003; Gerardin et al., 2000; Johnson, Rotte, et al., 2002; Lotze et al., 1999; Michelon, Vettel, & Zacks, 2006; Oullier, Jantzen, Steinberg, & Kelso, 2005; Porro et al., 1996), and transcranial magnetic stimulation (Pascual-Leone et al., 1995) have all revealed commonalities in the neural substrate governing imagined and executed actions.
Because overt and covert actions are similar at the behavioral, physiological, and neural levels and because mental practice has been shown to enhance motor performance in healthy people, researchers have speculated about its utility in neurorehabilitation (de Vries & Mulder, 2007; Jackson et al., 2001; Sharma, Pomeroy, & Baron, 2006). In fact, several review articles examining the impact of mental practice have been published. Two reviews (Braun, Beurskens, Borm, Schack, & Wade, 2006; Zimmermann-Schlatter, Schuster, Puhan, Siekierka, & Steurer, 2008) examined stroke outcomes in general and did not limit their review to upper-extremity–focused outcomes. Both articles included studies that were published in 2005 or earlier. Two additional reviews focused on upper-extremity outcomes from studies published in 2003 or earlier (Bell & Murray, 2004) or 2005 or earlier (Sharma et al., 2006); those reviews, however, did not attempt to rate the studies reviewed in terms of the level of evidence they provided. Thus, in this review, we determined whether mental practice is an effective intervention strategy to remediate impairments and improve upper-limb function after stroke by examining and rating the current evidence.
Method
Sources
The first two authors of this paper (Nilsen and Gillen) performed independent searches of electronic databases including Medline, PubMed, the Cochrane Database, PsycINFO, and CINAHL from 1985 to February 2009. In addition, they hand-searched reference lists of obtained articles. Search terms included mental practice, mental imagery, motor imagery, stroke, cerebrovascular accident, limb, arm, and upper extremity. The search was limited to journals published in English.
Study Selection
We included published and completed studies in this review if they met the following criteria: The participant’s primary diagnosis was a stroke, mental practice was used as part of the intervention plan either in isolation or in conjunction with other therapies, and mental practice was used to reduce upper-extremity impairment or improve upper-extremity function. All levels of evidence were considered for this review, from Level I (randomized controlled trials, or RCTs) to Level V (case reports).
Data Extraction
Nilsen and Gillen reviewed the articles that met the inclusion criteria and extracted the following data: study objectives, level of evidence and design, description of participants (e.g., age, time since stroke, lesion sites), intervention description, outcome measures, and dimension of the outcome measures based on the International Classification of Functioning, Disability and Health (World Health Organization, 2001). For this review, we used the following levels of evidence: Level I (systematic reviews, meta-analyses, RCTs), Level II (nonrandomized controlled trials, case control trials), Level III (pretest–posttest designs, cross-sectional designs), Level IV (single-subject designs, case series), and Level V (case reports, narrative literature reviews). Additionally, Nilsen and Gillen assessed the quality of the included Level I and II studies using the PEDro scale (partitioned; 2003). The PEDro (partitioned) rates a study’s internal validity and statistical reporting using 8-point and 2-point scales, respectively, with higher scores indicating higher quality. This scale has been found to be a reliable instrument for assessing the internal validity and statistical reporting of RCTs (Tooth et al., 2005). Consensus was used if there was disagreement on whether a study met the inclusion criteria, level of evidence, or quality ratings.
Results
Nilsen and Gillen originally located 20 articles using the previously mentioned key words and study selection criteria. Of those articles, 2 were excluded because they did not address remediation of upper-limb function directly (Liu, Chan, Lee, & Hui-Chan, 2004a; Liu, Chan, Lee, & Hui-Chan, 2004b), and 3 were excluded because they were not completed studies (Braun et al., 2007; Ietswaart et al., 2006; Verbunt et al., 2008). Therefore, a total of 15 articles were reviewed and classified according to the criteria described earlier. The results are summarized in Table 1.
Summary of Evidence on Use of Mental Practice to Improve Upper-Limb Recovery After Stroke
Note. Impairments are problems in body function (physiological functions of body systems) or structure (anatomical parts of the body such as organs, limbs, and their components), such as a significant deviation or loss. Activity limitations are difficulties a person may have in executing activities. Participation restrictions are problems a person may experience in involvement in life situations. Level I = systematic reviews, meta-analyses, randomized controlled trials; Level II = nonrandomized controlled trials, case control trials; Level III = pretest–posttest designs, cross-sectional designs; Level IV = single-subject designs, case series; Level V = case reports, narrative literature reviews. ADLs = activities of daily living; ARAT = Action Research Arm Test; CCT = clinical control trial; CIMT = constraint-induced movement therapy; CVA = cerebrovascular accident; FMA = Fugl-Meyer Motor Assessment; MAL = Motor Activity Log; mCIMT = Modified Constraint Induced Movement Therapy; MP = mental practice; OT = occupational therapy; PT = physical therapy; RCT = randomized controlled trial; STREAM = Stroke Rehabilitation Assessment of Movement; UE = upper extremity.
Information in parentheses is the disability dimension according to the World Health Organization’s (2001) International Classification of Functioning and Disability (ICF).
As can be seen in Table 1, the research designs differed substantially among studies. A critical appraisal of the study designs resulted in 4 studies being categorized as Level I evidence (Page, 2000; Page, Levine, & Leonard, 2005, 2007; Page, Levine, Sisto, & Johnston, 2001b), 2 as Level II (Dijkerman, Ietswaart, Johnston, & MacWalter, 2004; Müller, Bütefisch, Seitz, & Hömberg, 2007), 1 as Level III (Simmons, Sharma, Baron, & Pomeroy, 2008), 6 as Level IV (Butler & Page, 2006; Crosbie, McDonough, Gilmore, & Wiggam, 2004; Hewett, Ford, Levine, & Page, 2007; Page, Levine, & Hill, 2007; Stevens & Phillips Stoykov, 2003; Yoo, Park, & Chung, 2001), and 2 as Level V (Gaggioli, Meneghini, Morganti, Alcaniz, & Riva, 2006; Page, Levine, Sisto, & Johnston, 2001a).
The quality assessment of the Level I and II studies are summarized in Table 2. In terms of internal validity, the overall scores ranged from 2 to 5. Four of the 6 studies were rated positively for random allocation of groups (Page, 2000; Page et al., 2001b, 2005; Page, Levine, & Leonard, 2007) and similar groups at baseline (Dijkerman et al., 2004; Müller et al., 2007; Page et al., 2005; Page, Levine, & Leonard, 2007). Three of 6 studies scored positively for blinding of the assessors (Page et al., 2001b, 2005; Page, Levine, & Leonard, 2007), whereas only 2 studies included blinding of the therapists who provided the treatment (Page et al., 2001b; Page, Levine, & Leonard, 2007). All the studies scored positively on obtaining key outcome measures from ≥85% of participants initially allocated to groups. Conversely, all the studies scored negatively on concealment of allocation, blinding of participants, and intention-to-treat analysis. In regard to statistical reporting, the overall scores ranged from 1 to 2; all but 1 study (Page et al., 2001b) scored positively on both items.
Quality Assessment of Internal Validity of the Clinical and Randomized Clinical Trials With the PEDro Scale (Partitioned)
Note. + = 1 point; – = 0 points.
In addition to differences in study design and methodologic quality, the studies differed with respect to patient characteristics, intervention protocols, and outcome measures. In terms of patient characteristics, all the studies included participants who had sustained a unilateral right or left cerebrovascular accident. However, stroke onset varied considerably within and across studies from the acute phase (i.e., several days after stroke; e.g., Müller et al., 2007) to the chronic phase (i.e., several years after stroke; e.g., Page, Levine, & Leonard, 2007). Additionally, inclusion and exclusion criteria varied from clearly stated inclusion–exclusion criteria (e.g., Müller et al., 2007; Page, Levine, & Leonard, 2007; Page et al., 2001b, 2005) to practically none stated at all (e.g., Dijkerman et al., 2004). Most studies mentioned did not assess individual participants’ ability to engage in imagery. This issue is important because although it has been suggested that after stroke, many patients retain the ability to mentally represent movements of the impaired limb (Johnson, Sprehn, & Saykin, 2002), it appears that certain brain lesions (e.g., parietal lobe) result in an inability to perform motor imagery (Sirigu & Duhamel, 2001; Sirigu et al., 1996). Additionally, data for healthy participants indicate that imagery ability may affect performance gains (Goss, Hall, Buckolz, & Fishburne, 1986).
Concerning the interventions used, most of the studies combined mental and physical practice, although the exact intervention protocols differed substantially among studies. For example, how the mental practice was facilitated (e.g., audiotape, written instruction, pictures), the imagery perspective used (i.e., internal vs. external), the tasks that were practiced, and the duration and intensity of the practice sessions varied among studies. With regard to how the mental practice was facilitated, studies used audiotaped instructions (Hewett et al., 2007; Page, 2000; Page et al., 2001b, 2005; Page, Levine, & Hill, 2007; Page, Levine, & Leonard, 2007), written instructions (Dijkerman et al., 2004), videotaped instructions (Crosbie et al., 2004; Müller et al., 2007), or visual displays of the less affected limb via a mirror box (Stevens & Phillips Stoykov, 2003) or computer-generated image (Gaggioli et al., 2006). Moreover, it appeared some studies had participants use an internal (i.e., first-person) view (e.g., Müller et al., 2007; Page et al., 2005; Page, Levine, & Leonard, 2007; Simmons et al., 2008), and others had participants use an external (i.e., third-person) view (e.g., Page, 2000; Page et al., 2001a, 2001b). In several studies, the perspective used was not clearly articulated (e.g., Crosbie et al., 2004; Dijkerman et al., 2004; Yoo et al., 2001). Additionally, the studies used various tasks during the mental practice sessions. These tasks included selective finger opposition sequences (Müller et al., 2007), simple wrist and forearm movements (Stevens & Phillips Stoykov, 2003), line tracing (Yoo et al., 2001), and functional reaching and grasping tasks (Crosbie et al., 2004; Dijkerman et al., 2004; Page, 2000; Page et al., 2005; Page, Levine, & Hill, 2007; Page, Levine, & Leonard, 2007). With reference to intensity and duration, RCT studies varied considerably regarding the number of trials or minutes engaged in practice, with participants receiving approximately 4–5 hr of mental practice over a 4-wk (Page, 2000) or 6-wk period (Page et al., 2001b), respectively, at a minimum to a maximum of 21 hr over a 6-wk period (Page et al., 2005).
Most studies used the Fugl-Meyer Motor Assessment (i.e., Gaggioli et al., 2006; Page, 2000; Page et al., 2001a, 2001b; Page, Levine, & Hill, 2007; Page, Levine, & Leonard, 2007; Stevens & Phillips Stoykov, 2003), the Motricity Index (i.e., Crosbie et al., 2004; Simmons et al., 2008), the Jebsen Test of Hand Function (i.e., Müller et al., 2007; Stevens & Phillips Stoykov, 2003), or the Action Research Arm Test (i.e., Gaggioli et al., 2006; Page et al., 2001b, 2005; Page, Levine, & Hill, 2007; Page, Levine, & Leonard, 2007) as primary outcome measures. One study (Hewitt et al., 2007) used a kinematic analysis of functional reaching tasks, and another used outcome measures aimed at complex task function (Dijkerman et al., 2004). The results of most of these primary outcome measures suggest that mental practice has a positive effect on upper-limb recovery at both the impairment (as supported by 3 Level I studies, 2 Level II studies, 1 Level III study, 6 Level IV studies, and 2 Level V studies) and activity limitations levels (as supported by 3 Level I studies, 1 Level II study, 1 Level III study, 3 Level IV studies, and 2 Level V studies). In fact, 7 of the studies showed statistically significant differences favoring mental practice on at least one outcome measure (Crosbie et al., 2004; Dijkerman et al., 2004; Hewett et al., 2007; Müller et al., 2007; Page, 2000; Page et al., 2005; Page, Levine, & Leonard, 2007). Moreover, as can be seen by an examination of the Level I and II studies, it appears that mental practice combined with physical practice yields better outcomes than physical practice alone or conventional therapy (Dijkerman et al., 2004; Müller et al., 2007; Page, 2000; Page et al., 2005; Page, Levine, & Leonard, 2007).
In addition, 4 studies contained outcome measures that assessed the participants’ perception of recovery (Butler & Page, 2006; Dijkerman et al., 2004; Page et al., 2005; Simmons et al., 2008). Although not explicitly stated, it appears that those measures were treated as secondary measures. One study found no effect on perceived recovery (Dijkerman et al., 2004), and the other 3 suggested that participants perceived they were using their arm more, had better quality of arm movement, or both (Butler & Page, 2006; Page et al., 2005; Simmons et al., 2008). Interestingly, only 2 studies included a measure of retention (Simmons et al., 2008; Stevens & Phillips Stoykov, 2003). Thus, it is unclear whether the improvements seen as a result of mental practice are retained over time.
Discussion and Implications for Practice, Education, and Research
Most studies have shown that mental practice reduces impairments and improves functional recovery of the upper limb. Thus, it appears to be an appropriate intervention strategy to be used during poststroke rehabilitation. As indicated in previous reviews (e.g., Braun et al., 2006; Sharma et al., 2006), however, given the studies’ heterogeneity, generalizations are difficult to make at this time. Further investigations are needed to determine which people are most likely to benefit from the intervention after stroke, whether the benefits of mental practice are retained over time, and how broad the effects are in terms of improving perceived occupational performance. In addition, it remains unclear what the optimal dosing is regarding the amount of mental practice or the ratio of mental to physical practice needed to obtain a positive effect.
Moreover, it would be of interest to determine whether one type of instruction is more beneficial to facilitating imagery than another and to what extent imagery training before engagement in mental practice is beneficial (Hall, Buckoltz, & Fishburne, 1992). Moreover, although not explicitly stated, all of the studies likely used a combination of visual and kinesthetic imagery modalities, but they may have done so from different imagery perspectives. Unfortunately, imagery modality and perspective are often considered one and the same in the literature, making conclusions regarding their benefits difficult to discern (Callow & Hardy, 2004; Glisky & Williams, 1996). For example, neurorehabilitation literature in the area of mental practice has proposed that the first-person perspective and kinesthetic imagery are synonymous, whereas the third-person perspective corresponds to visual imagery (de Vries & Mulder, 2007).
According to Stevens (2005), a motor representation contains both the kinesthetic and the biomechanical constraints associated with the action, as well as the spatial coordinates of the action. Therefore, the visual imagery modality (i.e., “seeing” your hand move) may be used to represent the spatial coordinates (i.e., distance and location), and the kinesthetic imagery modality (i.e., “feeling” your hand move) may be used to represent the biomechanical constraints (Stevens, 2005). Thus, the use of both visual and kinesthetic imagery during mental practice would appear to be valuable. However, one can engage in both imagery modalities from either perspective. For example, one may be able to “see” or “feel” one’s hand moving from either point of view. The assumption that only the internal perspective is suitable for the generation of kinesthetic imagery appears erroneous and has been questioned in the literature (Callow & Hardy, 2004). In addition, behavioral evidence in healthy people suggests that the imagery modality used (Féry, 2003) and the imagery perspective chosen (Hardy & Callow, 1999; Sirigu & Duhamel, 2001; White & Hardy, 1995) during mental practice may affect task performance differently. Thus, although we agree with the conclusions reached by de Vries and Mulder (2007) regarding the need to determine the benefits of internal versus external motor imagery, we suggest that perspective (i.e., internal vs. external), modality (i.e., visual vs. kinesthetic), and the relationship between the two are separate areas needing further investigation in patient populations.
In summary, although the benefits of mental practice in poststroke rehabilitation appear promising, general conclusions are difficult to make at this time. Further investigation is warranted in terms of appropriate dosing, mode of presentation, the effects of visual and kinesthetic motor imagery, and the effects of imagery perspective during mental practice. In addition, it remains to be determined whether the benefits of mental practice are retained over time and to what degree mental practice affects occupational performance. Although several studies have shown support for the notion that mental practice can reduce upper-limb impairments and increase functional use of the affected limb (e.g., Page et al., 2001b, 2005; Page, Levine, & Leonard, 2007; Müller et al., 2007), none of the studies investigated whether mental practice promotes return to occupational performance as perceived by the participants engaged in the treatment.
Limitations to this systematic review that may threaten the validity of findings include limiting the search to journals published in English, the possibility of missing some studies because of combinations of search terms and terminology used in published papers, and, in general, studies with statistically significant or positive findings being more likely to be published than trials with nonsignificant or negative findings. Finally, the levels of evidence that were used for this systematic review did not differentiate between large and small RCTs, granting Level I evidence to all RCTs.