- Email: [email protected]
Bilateral movement training and stroke rehabilitation: A systematic review and meta-analysis
Journal of the Neurological Sciences 244 (2006) 89 – 95 www.elsevier.com/locate/jns
Bilateral movement training and stroke rehabilitation: A systematic review and meta-analysis Kim C. Stewart a, James H. Cauraugh a,*, Jeffery J. Summers b b
a University of Florida Gainesville, Florida, USA University of Tasmania Hobart, Tasmania, Australia
Received 14 October 2005; received in revised form 30 December 2005; accepted 4 January 2006 Available online 14 February 2006
Abstract Objective and design: Bilateral movement training is being increasingly used as a post-stroke motor rehabilitation protocol. The contemporary emphasis on evidence-based medicine warrants a prospective meta-analysis to determine the overall effectiveness of rehabilitating with bilateral movements. Methods: After searching reference lists of bilateral motor recovery articles as well as PubMed and Cochrane databases, 11 stroke rehabilitation studies qualified for this systematic review. An essential requirement for inclusion was that the bilateral training protocols involved either functional tasks or repetitive arm movements. Each study had one of three common arm and hand functional outcome measures: Fugl-Meyer, Box and Block, and kinematic performance. Results: The fixed effects model primary meta-analysis revealed an overall effect size (ES = 0.732, S.D. = 0.13). These findings indicate that bilateral movement training was beneficial for improving motor recovery post-stroke. Moreover, a fail-safe analysis indicated that 48 null effects would be necessary to lower the mean effect size to an insignificant level. Conclusion: These meta-analysis findings indicate that bilateral movements alone or in combination with auxiliary sensory feedback are effective stroke rehabilitation protocols during the sub-acute and chronic phases of recovery. D 2006 Elsevier B.V. All rights reserved. Keywords: Bilateral; Meta-analysis; Fugl-Meyer test; Fail-safe analysis; Box; Block test
1. Introduction Over 725,000 people experience a stroke each year in the USA, and over 40,000 in Australia. As stroke survival rates increase, the number of survivors dealing with motor impairments after 12 months has reached an astounding 70% [1]. The disabilities in the upper extremities severely limit functional motor capabilities; therefore, stroke researchers and therapists are searching for more effective upper extremity rehabilitation techniques for regaining voluntary motor control. * Corresponding author. Motor Behavior Laboratory, P.O. Box 118206, University of Florida, Gainesville, Florida 32611, USA. Tel.: +1 352 392 0584x1273; fax: +1 352 392 0316. E-mail address: [email protected] (J.H. Cauraugh). 0022-510X/$ - see front matter D 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2006.01.005
A current prominent rehabilitation technique is bilateral movement training. This protocol applies sound neurological interlimb coordination postulates in activating motor synergies between limbs [2,3]. Specifically, voluntary movements of the intact limb may facilitate voluntary movements in the paretic limb. Activating the primary motor cortex and supplementary motor area for the intact limb increases the likelihood of voluntary muscle contractions (i.e., motor synergies) in the impaired limb when symmetrical movements are executed [4– 9]. However, relatively small sample sizes as well as inconsistent findings across bilateral movement studies have limited the overall conclusions that can be made about the efficacy of bilateral training [10 – 26]. Conducting a systematic review and quantitative meta-analysis across bilateral movement training studies will help
90
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
resolve the conflicting findings and determine the overall effectiveness of this upper extremity stroke rehabilitation protocol [27]. Moreover, given the contemporary emphasis on evidence-based medicine, all reliable information on stroke recovery attributed to bilateral movement training should be included in a systematic review and prospective meta-analysis. Recently, prospective metaanalyses have been computed as quantitative systematic reviews rather than narrative reviews as soon as preliminary findings begin accumulating on two sides of an issue. Further, as meta-analytic techniques allow specific research hypotheses to be tested for moderator effects, the current analysis included studies that combined at least one other protocol (i.e., active neuromuscular stimulation or auditory cuing) with bilateral movement training.
Table 1 Meta-analysis flowchart guide Bilateral movement training effect
Conduct literature search
Assign inclusion/exclusion criteria
Select appropriate outcome measures
Calculate effect size (using cohen's d),variance, and SD
Apply correction factor to effect size, variance, and SD
Calculate overall weighted mean effect size, variance, and SD
2. Method
Calculate confidence intervals and construct relevant plots
2.1. Subjects: study selection and inclusion/exclusion criteria Collecting all pertinent studies is crucial when conducting a meta-analysis. The current methods included an exhaustive search for references from stroke and bilateral movement studies, review articles, book chapters, as well as two computerized databases: (1) PubMed (1966 to 2005) and (2) Cochrane Controlled Trials Register (1st and 2nd Quarter 2005). Key search words included stroke/ hemiparesis/cerebrovascular accident, bilateral/bimanual coordination/training, motor recovery/rehabilitation, motor control/coordination, and interlimb coordination. The exhaustive literature search identified 17 relevant studies including one unpublished study for further analysis to determine if inclusion in this meta-analysis was appropriate [10 –26]. Criteria for study inclusion were: (1) upper extremity stroke hemiparesis, with enough residual motor control in the impaired arm to be able to perform the motor capabilities tests, (2) bilateral movement training as a treatment group, and (3) functional analysis of the arm/hand. Six studies were excluded from the meta-analysis because five did not involve bilateral movements as a treatment, and one did not report a functional outcome measure for the impaired arm. The 11 remaining studies used bilateral movement training alone as a rehabilitation technique or combined bilateral movements with another treatment protocol, such as auditory cuing or active neuromuscular stimulation on the impaired arm while testing subjects in sub-acute and chronic stages of recovery. The necessary data from the 11 bilateral training studies were extracted by two authors and separately confirmed. The procedures involved in a meta-analysis vary considerably in the literature. To ensure clarity in our conservative approach, Table 1 displays the steps involved
Test for homogeneity
Analyze moderating variables
Calculate fail-safe N
in the current meta-analysis. Further, Table 2 summarizes seven distinct characteristics for each study. 2.2. Establishing outcome measures To accurately compare the studies and determine the overall effects of bilateral movement training, common outcome measures were selected, and the results of each measure were standardized. Three common arm/hand functional tests were reported as outcome measures: (1) Fugl-Meyer upper extremity motor test, (2) Box and Block test, and (3) kinematic performance rating. 2.3. Data synthesis Given the reported statistical information, individual effect sizes for the 11 bilateral movement training studies were calculated using Cohen’s d formula. This effect size formula is readily accepted in the meta-analytic literature and is calculated as the difference between pretest and posttest scores divided by the pooled standard deviation [27 – 29]. Because three of the 11 bilateral studies [10,11,17] tested different subjects in multiple experiments the total number of individual effect sizes increased to 18. Moreover, consistent with suggestions for heterogeneity across studies, a fixed effects model was used [30,31].
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
91
Table 2 Characteristics of each study used in the meta-analysis Study
Total N
Mean age: years
Lesion location
Mean time post-stroke (months)
Training duration
Length of study
Treatment protocol
Mudie and Matyas [10] Mudie and Matyas [11] Whitall et al. [12]
8 4 14
69.4 N/A 63.8
Right = 6 Left = 2 N/A Right = 7 Left = 7
4.3 N/A 66.9
Time å N/A Time å N/A Time å 50 min
8 weeks (40 sessions) 6 weeks (30 sessions) 6 weeks (18 sessions)
Cauraugh and Kim [13]
25
63.7
Right = 12 Left = 13
39.1
Time å 90 min
4 days over 2 weeks
Cauraugh and Kim [14]
26
66.4
Right = 15 Left = 11
33.6
Time å 90 min
4 days over 2 weeks
Lewis and Byblow [15] McCombe-Waller et al. [16]
6 20
58.7 N/A
Right = 5 Left = 1 N/A
16.2 >12
33 trials Time å 50 min
4 weeks (20 sessions) 6 weeks (18 sessions)
Stinear and Byblow [17] Luft et al. [18]
9 21
62 61.5
Right = 3 Left = 6 Right = 14 Left = 7
Time = 60 min Time å 50 min
4 weeks (20 sessions) 6 weeks (18 sessions)
Cauraugh et al. [19]
26
64.2
Right = 15 Left = 6
16.22 50.3 (median) 50.1
Time å 90 min
4 days over 2 weeks
Summers et al. [20]
12
61.7
Right = 4 Left = 8
62.2
50 trials
6 days
Single: bilateral tasks Single: bilateral tasks Coupled: AUD + bilateral movements Coupled: ANS + bilateral movements Coupled: ANS + bilateral movements Single: bilateral tasks Coupled: AUD + bilateral movements Single: bilateral training Coupled: AUD + bilateral movements Coupled: ANS + bilateral movements Single: bilateral training
List is in chronological order. Note. Single = only bilateral training; Coupled = two protocols simultaneously presented; AUD = auditory rhythmic cuing; ANS = active neuromuscular stimulation.
To avoid an overestimation of effect size for small samples in any of the qualified studies a correction procedure was applied [32]. The correction involved weighting an effect size by the reciprocal of its variance to determine the overall weighted mean effect size. This procedure allows a more precise effect size with smaller variances to receive a larger weight in the overall group mean [27,29,31,32]. A third accepted and applied meta-analytic procedure is to evaluate the contribution of moderating variables [31,32]. Two moderating variable analyses were proposed. First, the motor improvement contributions attributed to a single treatment protocol in comparison to the additive effect of simultaneously presented multiple treatment protocols is an intriguing moderating variable analysis. A second moderating variable analysis examining lesion location (i.e., cortical vs. subcortical) was proposed, however, the information provided in these studies was not sufficient to undertake such an analysis. 2.4. Fail-safe analysis To account for the possibility of publication bias because of the unpublished articles, a fail-safe analysis was conducted to determine the number studies with null effects that would be necessary to lower the calculated effect size to an insignificant level. Quantifying a potential confounding bias from unpublished studies has become accepted as a standard procedure in meta-analyses [29,31,32]. 2.5. Quality assessment Further analysis involved quality assessment for each of the studies. Consistent with recommendations by Jadad et al.
and Moher et al., the quality of studies was assessed via three criteria by the first author and checked by the second author: (1) randomization, (2) double blinding, and (3) withdrawals or drop-outs [33,34]. Randomization was recorded if the subjects were either randomly placed into a treatment or control group or if the treatment was randomly assigned to the subjects. As shown in Table 2, randomization was completed across 10 of the studies. Concerning the double blind criteria, the rating scale varied from 0– 2: (1) not described or inappropriate = 0, (2) single blind = 1, and (3) double blind = 2. A second quality assessment involved dropout guidelines. Accordingly, three studies had subject(s) drop-out, whereas only one study excluded subjects from their results because they were unable to perform the task or their outcome scores represented outliers. Thus, the quality assessment data indicated a favorable group of studies and all 11 bilateral movement training studies were included in the present meta-analysis. See Table 3 for a summary of the quality assessments of the studies that used bilateral movement training as a treatment.
3. Results 3.1. Primary meta-analysis: mean effect size The initial meta-analysis revealed a significant overall mean effect size of 0.732 (S.D. = 0.13) with a 95% confidence interval ranging from 0.66 to 0.80. Consistent with Rosenthal’s rule for inspecting confidence intervals in a meta-analysis, if the range does not pass through zero, then the cumulative effect size is significant [27,29]. Moreover, an effect size equal to 0.732 represents a
92
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
Table 3 Quality assessments for each study included in the meta-analysis Study
Random assignment
Not described
Single blind
Double blind
Drop-outs
Treatment group N / Total N
Mudie and Matyas [10] Mudie and Matyas [11] Whitall et al. [12] Cauraugh and Kim [13] Cauraugh and Kim [14] Lewis and Byblow [15] McCombe-Waller et al. [16] Stinear and Byblow [17] Luft et al. [18] Cauraugh et al. [19] Summers et al. [20]
1 1 0 1 1 1 0 1 1 1 1
0 0 1 0 0 0 1 0 0 0 0
1 1 0 1 1 1 0 1 1 1 1
0 0 0 0 0 0 0 0 0 0 0
0 0 2 0 0 0 1 0 2 0 0
8/8 4 / 12 14 / 14 10 / 25 26 / 26 6/6 8 / 20 9/9 9 / 21 11 / 26 6 / 12
Note. Consistent with quality assessment suggestions, 1 indicates ‘‘Yes’’ and 0 indicates ‘‘No’’. For the last two columns, the values represent the exact number of subjects.
relatively large effect (e.g., 0.2 = small; 0.5 = medium; 0.8 = large) according to Cohen [28]. This large mean effect was based on 11 studies and 18 individual effect sizes with 111 bilateral movement treatment subjects out of 171 total subjects. Table 4 shows the outcome measures, bilateral movement training protocol, and individual weighted effect sizes with confidence intervals for the studies used in the meta-analysis. 3.2. Fail-safe analysis A fail-safe analysis determines the number findings with null effects necessary to lower the calculated effect size to an insignificant level. For the combination of these bilateral movement findings, 48 null effect results are
necessary to reduce the overall effect size to an insignificant value. 3.3. Homogeneity test Consistent with meta-analytic recommendations, we examined the variability of the weighted effect sizes [27]. Testing for homogeneity across the studies determines whether the results reflect a single underlying effect or a distribution of effects [27,29,31,32]. For the current metaanalysis, the distribution of the effect sizes was heterogeneous as determined by the Q statistic, v 2 (17, n = 18) = 30.65, p < 0.02. Even though the overall effect size was significant, the Q statistic finding warrants further analysis of potential moderating variables [29,32].
Table 4 Summary statistics for total sample in the meta-analysis Study
Outcome measure
Bilateral training
Weighted effect size
Mudie and Matyas [10]
Study 1: kinematic performance; percent deficit
Block placement Simulated drinking Peg targeting Block placement Simulated drinking Peg targeting Block placement Simulated drinking BATRAC Wrist/finger extension Wrist/finger Extension Upper extremity tasks BATRAC Synchronized: active – passive bilateral training Asynchronized: active – passive bilateral training BATRAC Wrist/finger extension Upper extremity reaching: Dowel placement task
1.203 1.883 1.938 2.692 2.609 2.155 3.533 2.738 0.341 0.457 0.242 0.630 0.603 0.368
0.019 – 2.425 0.532 – 3.235 0.574 – 3.302 1.139 – 4.246 1.079 – 4.120 0.741 – 3.570 1.733 – 5.333 1.172 – 4.304 0.122 – 0.561 0.008 – 0.652 0.130 – 0.355 0.009 – 1.250 0.176 – 1.031 0.424 – 1.159
0.690
0.468 – 1.849
0.783 0.730 0.410
0.394 – 1.172 0.304 – 0.890 0.203 – 1.034
Study 2: kinematic performance; percent deficit Mudie and Matyas [11] Whitall et al. [12] Cauraugh and Kim [13] Cauraugh and Kim [14] Lewis and Byblow [15] McCombe-Waller et al. [16] Stinear and Byblow [17]
Luft et al. [18] Cauraugh et al. [19] Summers et al. [20]
Study 3: kinematic performance; percent deficit Fugl-Meyer test: upper extremity Box and Block Test Box and Block Test Fugl-Meyer test: upper extremity Fugl-Meyer test: upper extremity Fugl-Meyer test: upper extremity
Fugl-Meyer test: upper extremity Movement time Motor assessment scale
Confidence intervals
Note. BATRAC = bilateral arm training with rhythmic auditory cuing. Confidence intervals were calculated with standard error values.
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
3.4. Moderating variables As shown in Table 2, five studies used one protocol and six studies used coupled protocols (i.e., two protocols simultaneously). However, with multiple experiments reported in the studies, the total number of individual weighted effect sizes for the single treatment protocols equaled 12 whereas the coupled protocols remained at six. A one-way ANOVA indicated a larger effect size for the various single protocols (M = 1.78, S.D. = 1.01) than the various coupled protocols (M = 0.41, S.D. = 0.23), F(1, 16) = 10.46, p < 0.005. 3.5. Second meta-analysis: mean effect size A second meta-analysis was conducted to avoid any potential bias in the overall weighted effect size. Specifically, the three independent studies of Mudie and Matyas had multiple bilateral training sessions (see Table 4). Within the guidelines of the meta-analytic technique, including each of the training sessions as separate studies was acceptable [27,31,32]. However, to ensure that an inordinate influence to the overall effect size was not coming from one laboratory, this second meta-analysis used only one bilateral training session, block placement data. Thus, removing the simulated drinking and peg targeting data reduced the number of effect sizes from 18 to 13. The 13 individual effect sizes were calculated on data from 11 separate articles and two additional studies. Mudie and Matyas’ [10] studies 1 and 2 were included as well as Stinear and Byblow’s [17] two different experiments with different subjects (i.e., synchronous and asynchronous active – passive bilateral training). Even with the reduced numbers of studies, the overall weighted effect still reached a medium size, ES = 0.582 (S.D. = 0.14). As revealed by the lower and upper confidence intervals (.49 to .67), this second meta-analysis was significant, and a new fail-safe test indicated that 25 null effect studies would have to be reported to reduce the effect size to 0.2. Moreover, the Q statistic [v 2 (12, n = 13) = 13.00, p > 0.05] indicated a homogenous group of studies in this second meta-analysis. Thus, overall these bilateral movement training meta-analysis results can be interpreted with confidence. Further, the moderating variables comparison of single bilateral treatment tasks versus coupled simultaneous treatment protocols was not significant. In the second meta-analysis, bilateral movement training alone improved motor capabilities as well as bilateral movements coupled with auditory rhythmic cuing and active neuromuscular stimulation.
4. Discussion Although bilateral movement training has been argued to be a viable stroke rehabilitation protocol [4,10 – 26], some
93
studies have reported inconsistent motor recovery improvements from the intervention. The current systematic review and prospective meta-analysis evaluated the findings in stroke rehabilitation studies using bilateral movements. An initial meta-analysis of 18 weighted effect sizes and 111 total subjects revealed an overall effect size of 0.732. A more conservative second meta-analysis with five less studies/experiments indicated a 0.582 effect size. Most importantly, the magnitude of both cumulative effect sizes indicate that bilateral movements, either as a single protocol or as a dual protocol with rhythmic auditory cuing or active stimulation, is an effective rehabilitation protocol for improving motor capabilities post-stroke. Indeed, both overall effect sizes support the argument that significant upper extremity motor recovery gains evolve from activity-dependent interventions based on theoretical motor control constructs founded on the behavioral and neurophysiological mechanisms involved in interlimb coordination [4]. Specifically, the efficacy of bilateral movements alone or in combination with other protocols provide further evidence that the apparent default organization of the motor system towards the coupling of muscles directly improves motor performance in paretic limbs post-stroke [4]. Possible neural mechanisms underlying bilateral movements abound. A basic assumption of the use of bilateral movement therapy is that symmetrical (in-phase) bilateral movements activate similar neural networks in both hemispheres when homologous muscle groups are simultaneously activated [9,33 – 39]. Bilateral symmetrical movements, therefore, may allow the activation of the undamaged hemisphere to increase activation of the damaged hemisphere and facilitate movement control of the impaired limb promoting neural plasticity [4,37]. Given that both hemispheres are activated during symmetrical bilateral tasks, researchers have proposed a central regulatory mechanism controlling both limbs [5]. Projections of the supplementary motor area (SMA) make this mechanism a viable bilateral movement executive function structure. The SMA in each hemisphere projects to the ipsilateral primary cortex and to a lesser extent to homologous muscles in the contralateral primary motor cortex [4,7]. Indeed, recent imaging studies have confirmed the higher-order control of bilateral arm movements by the SMA, as well as identifying four additional areas involved during interlimb coordination: (1) cingulate motor cortex, (2) lateral premotor cortex, (3) superior parietal cortex, and (4) cerebellum [4,6,34,37]. However, it is not clear whether symmetrical (in-phase) movements are necessary for bilateral movement training to be effective. For example, improved motor recovery has been shown from training programs involving both in-phase and anti-phase bilateral pushing/pulling movements combined with rhythmic auditory cuing [12,16,18]. Moreover, both synchronous and asynchronous active– passive bilateral movement therapy may differentiate stroke recovery
94
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
[17]. Thus, symmetrical and asymmetrical bilateral movement training should be further investigated to clarify the contribution of each coordination mode. Additional considerations for future stroke rehabilitation research include determining the effect of the (1) level of intensity of bilateral movement therapy on specific motor outcome measures, (2) duration of bilateral movement training on both immediate and long-term motor improvements, and (3) most effective combination of bilateral movement training and supplementary assistive protocols (e.g., auditory cues, active neuromuscular stimulation, or rhythmic cadence). Investigating these specific bilateral rehabilitation concerns for each of the three stages of recovery would contribute immensely to our understanding of stroke induced motor capabilities. As stated by Winstein et al., post-stroke motor learning (relearning) presents multiple avenues for future research [40]. Moreover, investigating the long-term cumulative effect of bilateral movement protocols with a robust sample size of chronic stroke patients would differentiate temporary learning from permanent learning [41]. The favorable overall effect of bilateral movement training in assisting stroke motor recovery is compelling even with the inherent limitations of the meta-analysis technique. First, criteria for admission into the metaanalysis can directly influence the overall mean effect size. However, rather than accepting all bilateral movement studies, the present analysis followed a conservative approach and clustered studies on outcome measures that focused on arm/hand functions. Second, homogeneity tests on the distributions of the effect sizes for the functional outcome measures revealed heterogeneity in the primary meta-analysis and homogeneity in the additional metaanalysis. Thus, potential moderating variables were investigated to further elaborate the present findings. For the initial meta-analysis, a comparison of only bilateral movement tasks versus coupled (two simultaneously combined) protocols revealed a motor advantage to the various single treatment protocols. However, the second meta-analysis indicated that bilateral movement training alone as well as bilateral movements coupled with either rhythmic auditory cues or active neuromuscular stimulation increased motor improvements. In addition, a fail-safe analysis was conducted to counter the argument that this meta-analytic procedure reports publication bias because we only included one article that has not been published in refereed journal. Based on this conservative analysis, 48 studies with null effects are necessary to lower the large calculated effect size (0.732) to a small, insignificant level. Moreover, the second meta-analysis revealed a medium effect size (0.582), 25 null effect studies for the fail-safe analysis, and homogeneous variability in the reduced number of bilateral studies. Thus, based on the two present metaanalyses four lines of evidence indicate a relatively robust bilateral movement training effect: (1) overall weighted
effect sizes, (2) fail-safe analyses, (3) Q statistics, and (4) quality of study ratings. In conclusion, the magnitude of the overall mean effects of these meta-analyses taken together with the conservative approach, the additional statistical tests conducted, as well as the quality of study ratings demonstrate that bilateral movements are effective in improving motor functions for the sub-acute and chronic phases of stroke recovery. These bilateral movement protocol (e.g., alone or in conjunction with auxiliary sensory feedback) findings extend Whitall’s persuasive argument that rehabilitation research should focus on maximizing benefits from practicing motor actions [42]. Rehabilitation specialists are encouraged to incorporate the findings from this systematic review and two metaanalyses into their informed decisions on protocols for improving motor capabilities and making progress toward stroke motor recovery [42 – 45].
Acknowledgements James Cauraugh was supported by an award from the American Heart Association, Florida/Puerto Rico Affiliate and as an affiliated investigator of the Brain Rehabilitation Research Center, Research Service, North Florida/South Georgia Veteran’s Health System, Gainesville, FL. Jeffery Summers was supported by awards from the Australian Research Council and the Australian Health Management Group - Medical Research Fund.
References [1] Hendricks HT, van Limbeck J, Geurts AC, Zwarts MJ. Motor recovery after stroke: a systematic review of the literature. Arch Phys Med Rehabil 2002;83:1629 – 37. [2] Bernstein NA. The co-ordination and regulation of movements. Oxford, UK’ Pergamon Press; 1967. [3] Swinnen SP. Intermanual coordination: from behavioural principles to neural-network interactions. Nat Rev Neurosci 2002;3:350 – 61. [4] Cauraugh JH, Summers JJ. Neural plasticity and bilateral movements: a rehabilitation approach for chronic stroke. Progress Neurobiol 2005;75:309 – 20. [5] Cohen L. Interaction between limbs during bimanual voluntary activity. Brain 1970;93:259 – 72. [6] Debaere F, Wenderoth N, Sunaert S, Van Hecke P, Swinnen SP. Changes in brain activation during the acquisition of a new bimanual coordination task. Neuropsychologia 2004;42:855 – 67. [7] Goldberg G. Supplementary motor area structure and function: review and hypotheses. Behav Brain Sci 1985;8:567 – 616. [8] Swinnen SP, Wenderoth N. Two hands, one brain: cognitive neuroscience of bimanual skill. Trends Cog Neurosci 2004;8:18 – 25. [9] Wenderoth N, Debaere F, Suraert, van Hecke P, Swinnen SP. Parietopremotor areas mediate directional interference during bimanual movements. Cereb Cortex 2004;14:1153 – 63. [10] Mudie MH, Matyas TA. Upper extremity retraining following stroke: effects of bilateral practice. J Neural Rehabil 1996;10: 167 – 84. [11] Mudie MH, Matyas TA. Can simultaneous bilateral movement involve the undamaged hemisphere in reconstruction of neural networks damaged by stroke? J Disabil Rehabil 2000;22:23 – 37.
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95 [12] Whitall J, Waller S, Silver K, Macko R. Repetitive bilateral arm training with rhythmic auditory cueing improves motor function in chronic hemiparetic stroke. Stroke 2000;31:2390 – 5. [13] Cauraugh JH, Kim SB. Two coupled motor recovery protocols are better than one: electromyogram-triggered neuromuscular stimulation and bilateral movements. Stroke 2002;33:1589 – 94. [14] Cauraugh JH, Kim SB. Chronic stroke motor recovery: duration of active neuromuscular stimulation. J Neurol Sci 2003;215: 13 – 9. [15] Lewis GN, Byblow WD. Neurophysiological and behavioral adaptations to a bilateral training intervention in individuals following stroke. Clin Rehabil 2004;18:48 – 59. [16] McCombe-Waller S, Whitall J. Fine motor control in adults with and without chronic hemiparesis: baseline comparison to nondisabled adults and effects of bilateral arm training. Arch Phys Med Rehabil 2004;85:1076 – 83. [17] Stinear JW, Byblow WD. Rhythmic bilateral movement training modulates corticomotor excitability and enhances upper limb motoricity poststroke: a pilot study. J Clin Neurophys 2004; 21:124 – 31. [18] Luft AR, McCombe-Waller S, Whitall J, Forrester LW, Macko R, Sorkin JD, et al. Repetitive bilateral arm training and motor cortex activation in chronic stroke: a randomized controlled trial. JAMA 2004;292:1853 – 61. [19] Cauraugh JH, Kim SB, Duley A. Coupled bilateral movements and active neuromuscular stimulation: intralimb transfer evidence during bimanual aiming. Neurosci Lett 2005;382:39 – 44. [20] Summers JJ, Garry MI, Kagerer FA, Hiraga CY, Loftus A. Bilateral training and recovery of upper arm function after stroke. Paper presented at the Xth International Symposium on Motor Control, Sofia, Bulgaria; 2004. [21] Dickstein R, Hocherman S, Amdor G, Pillar T. Reaction and movement times in patients with hemiparesis for unilateral and bilateral elbow flexion. Phys Ther 1993;73:374 – 80. [22] Roby-Brami A, Fuchs S, Mokhtari M, Bussel B. Reaching and grasping strategies in hemiparetic patients. Motor Control 1997;1:72 – 91. [23] Mudie MH, Matyas TA. Responses of the densely hemiplegic upper extremity to bilateral training. Neurorehabil Neural Repair 2001;15:129 – 40. [24] Cunningham CL, Phillips Stoykov ME, Walter CB. Bilateral facilitation of motor control in chronic hemiplegia. Acta Psychol 2002;110:321 – 37. [25] Rice MS, Newell KM. Interlimb coupling in left hemiplegia because of right cerebral vascular accident. Occup Ther J Res 2001;21:12 – 28. [26] Rice MS, Newell KM. Interlimb coupling in left hemiplegia because of right cerebral vascular accident. Arch Phys Med Rehabil 2005; 85:629 – 34. [27] Rosenthal R, DeMatteo MR. Meta-analysis: recent developments in quantitative methods for literature reviews. Annu Rev Psychol 2001; 52:59 – 82.
95
[28] Cohen J. Statistical power analysis for the behavioral sciences, 2nd ed. Hillsdale, NJ’ Erlbaum; 1988. [29] Rosenthal R. Writing meta-analytic reviews. Psychol Bull 1995;118:1173 – 81. [30] Thompson SG. Meta-analysis of clinical trials. In: Armitage P, Colton T. editors. Encyclopedia of biostatistics, vol. 4. New York’ Wiley; 1998. p. 2570 – 9. [31] Sutton AJ, Abrams KR, Jones DR, Sheldon TA, Song F. Methods for meta-analysis in medical research. New York’ Wiley; 2000. [32] Hedges LV, Olkin I. Statistical methods for meta-analysis. Orlando’ Academic Press; 1985. [33] Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJM, Gavaghan DJ, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996;17: 21 – 12. [34] Moher D, Schulz KF, Altman D. The CONSORT statement: revised recommendations for improving the quality of reports of parallelgroup randomized trials. JAMA 2001;285:1987 – 91. [35] Hallett M. Plasticity of the human motor cortex and recovery from stroke. Brain Res Rev 2001;36:169 – 74. [36] Lacroix S, Havton LA, McKay H, Yang H, Brant A, Roberts J, et al. Bilateral corticospinal projections arise from each motor cortex in the macaque monkey: a quantitative study. J Comp Neurol 2004;473: 147 – 61. [37] Carson RG. Neural pathways mediating bilateral interactions between the upper limbs. Brain Res Rev 2005;49:641 – 62. [38] Spijkers W, Heuer H. Behavioral principles of interlimb coordination. In: Swinnen SP, Duysens J, editors. Neuro-behavioral determinants of interlimb coordination: a multidisciplinary approach. Boston’ Kluwer; 2004. p. 223 – 58. [39] Jancke L, Peters M, Himmelbach M, Nosselt T, Shah J, Steinmetz H. fMRI study of bimanual coordination. Neuropsychologia 2000; 38:164 – 74. [40] Winstein CJ, Wing AM, Whitall J. Motor control and learning principles for rehabilitation of upper limb movements after brain injury. In: Boller F, Grafman J, Robertson IH, editors. Handbook of neuropsychology. Amsterdam’ North Holland; 2003. p. 77 – 137. [41] Cauraugh JH, Kim SB, Summers JJ. Cumulative neural plasticity: longitudinal motor improvements. 2005. Currently submitted for publication. [42] Whitall J. Stroke rehabilitation research: time to answer more specific questions? Neurorehabil Neural Repair 2004;18:3 – 8. [43] Cauraugh JH. Coupled rehabilitation protocols and neural plasticity: upper extremity improvements in chronic hemiparesis. Restor Neurol Neurosci 2004;22:337 – 47. [44] Bolton DAE, Cauraugh JH, Hausenblas HA. Electromyogramtriggered neuromuscular stimulation and stroke motor recovery of the upper limb: a meta-analysis. J Neurol Sci 2004;223:121 – 7. [45] Cohen J. Things I have learned (so far). Am Psychol 1990;45: 1304 – 12.
Bilateral movement training and stroke rehabilitation: A systematic review and meta-analysis Kim C. Stewart a, James H. Cauraugh a,*, Jeffery J. Summers b b
a University of Florida Gainesville, Florida, USA University of Tasmania Hobart, Tasmania, Australia
Received 14 October 2005; received in revised form 30 December 2005; accepted 4 January 2006 Available online 14 February 2006
Abstract Objective and design: Bilateral movement training is being increasingly used as a post-stroke motor rehabilitation protocol. The contemporary emphasis on evidence-based medicine warrants a prospective meta-analysis to determine the overall effectiveness of rehabilitating with bilateral movements. Methods: After searching reference lists of bilateral motor recovery articles as well as PubMed and Cochrane databases, 11 stroke rehabilitation studies qualified for this systematic review. An essential requirement for inclusion was that the bilateral training protocols involved either functional tasks or repetitive arm movements. Each study had one of three common arm and hand functional outcome measures: Fugl-Meyer, Box and Block, and kinematic performance. Results: The fixed effects model primary meta-analysis revealed an overall effect size (ES = 0.732, S.D. = 0.13). These findings indicate that bilateral movement training was beneficial for improving motor recovery post-stroke. Moreover, a fail-safe analysis indicated that 48 null effects would be necessary to lower the mean effect size to an insignificant level. Conclusion: These meta-analysis findings indicate that bilateral movements alone or in combination with auxiliary sensory feedback are effective stroke rehabilitation protocols during the sub-acute and chronic phases of recovery. D 2006 Elsevier B.V. All rights reserved. Keywords: Bilateral; Meta-analysis; Fugl-Meyer test; Fail-safe analysis; Box; Block test
1. Introduction Over 725,000 people experience a stroke each year in the USA, and over 40,000 in Australia. As stroke survival rates increase, the number of survivors dealing with motor impairments after 12 months has reached an astounding 70% [1]. The disabilities in the upper extremities severely limit functional motor capabilities; therefore, stroke researchers and therapists are searching for more effective upper extremity rehabilitation techniques for regaining voluntary motor control. * Corresponding author. Motor Behavior Laboratory, P.O. Box 118206, University of Florida, Gainesville, Florida 32611, USA. Tel.: +1 352 392 0584x1273; fax: +1 352 392 0316. E-mail address: [email protected] (J.H. Cauraugh). 0022-510X/$ - see front matter D 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2006.01.005
A current prominent rehabilitation technique is bilateral movement training. This protocol applies sound neurological interlimb coordination postulates in activating motor synergies between limbs [2,3]. Specifically, voluntary movements of the intact limb may facilitate voluntary movements in the paretic limb. Activating the primary motor cortex and supplementary motor area for the intact limb increases the likelihood of voluntary muscle contractions (i.e., motor synergies) in the impaired limb when symmetrical movements are executed [4– 9]. However, relatively small sample sizes as well as inconsistent findings across bilateral movement studies have limited the overall conclusions that can be made about the efficacy of bilateral training [10 – 26]. Conducting a systematic review and quantitative meta-analysis across bilateral movement training studies will help
90
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
resolve the conflicting findings and determine the overall effectiveness of this upper extremity stroke rehabilitation protocol [27]. Moreover, given the contemporary emphasis on evidence-based medicine, all reliable information on stroke recovery attributed to bilateral movement training should be included in a systematic review and prospective meta-analysis. Recently, prospective metaanalyses have been computed as quantitative systematic reviews rather than narrative reviews as soon as preliminary findings begin accumulating on two sides of an issue. Further, as meta-analytic techniques allow specific research hypotheses to be tested for moderator effects, the current analysis included studies that combined at least one other protocol (i.e., active neuromuscular stimulation or auditory cuing) with bilateral movement training.
Table 1 Meta-analysis flowchart guide Bilateral movement training effect
Conduct literature search
Assign inclusion/exclusion criteria
Select appropriate outcome measures
Calculate effect size (using cohen's d),variance, and SD
Apply correction factor to effect size, variance, and SD
Calculate overall weighted mean effect size, variance, and SD
2. Method
Calculate confidence intervals and construct relevant plots
2.1. Subjects: study selection and inclusion/exclusion criteria Collecting all pertinent studies is crucial when conducting a meta-analysis. The current methods included an exhaustive search for references from stroke and bilateral movement studies, review articles, book chapters, as well as two computerized databases: (1) PubMed (1966 to 2005) and (2) Cochrane Controlled Trials Register (1st and 2nd Quarter 2005). Key search words included stroke/ hemiparesis/cerebrovascular accident, bilateral/bimanual coordination/training, motor recovery/rehabilitation, motor control/coordination, and interlimb coordination. The exhaustive literature search identified 17 relevant studies including one unpublished study for further analysis to determine if inclusion in this meta-analysis was appropriate [10 –26]. Criteria for study inclusion were: (1) upper extremity stroke hemiparesis, with enough residual motor control in the impaired arm to be able to perform the motor capabilities tests, (2) bilateral movement training as a treatment group, and (3) functional analysis of the arm/hand. Six studies were excluded from the meta-analysis because five did not involve bilateral movements as a treatment, and one did not report a functional outcome measure for the impaired arm. The 11 remaining studies used bilateral movement training alone as a rehabilitation technique or combined bilateral movements with another treatment protocol, such as auditory cuing or active neuromuscular stimulation on the impaired arm while testing subjects in sub-acute and chronic stages of recovery. The necessary data from the 11 bilateral training studies were extracted by two authors and separately confirmed. The procedures involved in a meta-analysis vary considerably in the literature. To ensure clarity in our conservative approach, Table 1 displays the steps involved
Test for homogeneity
Analyze moderating variables
Calculate fail-safe N
in the current meta-analysis. Further, Table 2 summarizes seven distinct characteristics for each study. 2.2. Establishing outcome measures To accurately compare the studies and determine the overall effects of bilateral movement training, common outcome measures were selected, and the results of each measure were standardized. Three common arm/hand functional tests were reported as outcome measures: (1) Fugl-Meyer upper extremity motor test, (2) Box and Block test, and (3) kinematic performance rating. 2.3. Data synthesis Given the reported statistical information, individual effect sizes for the 11 bilateral movement training studies were calculated using Cohen’s d formula. This effect size formula is readily accepted in the meta-analytic literature and is calculated as the difference between pretest and posttest scores divided by the pooled standard deviation [27 – 29]. Because three of the 11 bilateral studies [10,11,17] tested different subjects in multiple experiments the total number of individual effect sizes increased to 18. Moreover, consistent with suggestions for heterogeneity across studies, a fixed effects model was used [30,31].
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
91
Table 2 Characteristics of each study used in the meta-analysis Study
Total N
Mean age: years
Lesion location
Mean time post-stroke (months)
Training duration
Length of study
Treatment protocol
Mudie and Matyas [10] Mudie and Matyas [11] Whitall et al. [12]
8 4 14
69.4 N/A 63.8
Right = 6 Left = 2 N/A Right = 7 Left = 7
4.3 N/A 66.9
Time å N/A Time å N/A Time å 50 min
8 weeks (40 sessions) 6 weeks (30 sessions) 6 weeks (18 sessions)
Cauraugh and Kim [13]
25
63.7
Right = 12 Left = 13
39.1
Time å 90 min
4 days over 2 weeks
Cauraugh and Kim [14]
26
66.4
Right = 15 Left = 11
33.6
Time å 90 min
4 days over 2 weeks
Lewis and Byblow [15] McCombe-Waller et al. [16]
6 20
58.7 N/A
Right = 5 Left = 1 N/A
16.2 >12
33 trials Time å 50 min
4 weeks (20 sessions) 6 weeks (18 sessions)
Stinear and Byblow [17] Luft et al. [18]
9 21
62 61.5
Right = 3 Left = 6 Right = 14 Left = 7
Time = 60 min Time å 50 min
4 weeks (20 sessions) 6 weeks (18 sessions)
Cauraugh et al. [19]
26
64.2
Right = 15 Left = 6
16.22 50.3 (median) 50.1
Time å 90 min
4 days over 2 weeks
Summers et al. [20]
12
61.7
Right = 4 Left = 8
62.2
50 trials
6 days
Single: bilateral tasks Single: bilateral tasks Coupled: AUD + bilateral movements Coupled: ANS + bilateral movements Coupled: ANS + bilateral movements Single: bilateral tasks Coupled: AUD + bilateral movements Single: bilateral training Coupled: AUD + bilateral movements Coupled: ANS + bilateral movements Single: bilateral training
List is in chronological order. Note. Single = only bilateral training; Coupled = two protocols simultaneously presented; AUD = auditory rhythmic cuing; ANS = active neuromuscular stimulation.
To avoid an overestimation of effect size for small samples in any of the qualified studies a correction procedure was applied [32]. The correction involved weighting an effect size by the reciprocal of its variance to determine the overall weighted mean effect size. This procedure allows a more precise effect size with smaller variances to receive a larger weight in the overall group mean [27,29,31,32]. A third accepted and applied meta-analytic procedure is to evaluate the contribution of moderating variables [31,32]. Two moderating variable analyses were proposed. First, the motor improvement contributions attributed to a single treatment protocol in comparison to the additive effect of simultaneously presented multiple treatment protocols is an intriguing moderating variable analysis. A second moderating variable analysis examining lesion location (i.e., cortical vs. subcortical) was proposed, however, the information provided in these studies was not sufficient to undertake such an analysis. 2.4. Fail-safe analysis To account for the possibility of publication bias because of the unpublished articles, a fail-safe analysis was conducted to determine the number studies with null effects that would be necessary to lower the calculated effect size to an insignificant level. Quantifying a potential confounding bias from unpublished studies has become accepted as a standard procedure in meta-analyses [29,31,32]. 2.5. Quality assessment Further analysis involved quality assessment for each of the studies. Consistent with recommendations by Jadad et al.
and Moher et al., the quality of studies was assessed via three criteria by the first author and checked by the second author: (1) randomization, (2) double blinding, and (3) withdrawals or drop-outs [33,34]. Randomization was recorded if the subjects were either randomly placed into a treatment or control group or if the treatment was randomly assigned to the subjects. As shown in Table 2, randomization was completed across 10 of the studies. Concerning the double blind criteria, the rating scale varied from 0– 2: (1) not described or inappropriate = 0, (2) single blind = 1, and (3) double blind = 2. A second quality assessment involved dropout guidelines. Accordingly, three studies had subject(s) drop-out, whereas only one study excluded subjects from their results because they were unable to perform the task or their outcome scores represented outliers. Thus, the quality assessment data indicated a favorable group of studies and all 11 bilateral movement training studies were included in the present meta-analysis. See Table 3 for a summary of the quality assessments of the studies that used bilateral movement training as a treatment.
3. Results 3.1. Primary meta-analysis: mean effect size The initial meta-analysis revealed a significant overall mean effect size of 0.732 (S.D. = 0.13) with a 95% confidence interval ranging from 0.66 to 0.80. Consistent with Rosenthal’s rule for inspecting confidence intervals in a meta-analysis, if the range does not pass through zero, then the cumulative effect size is significant [27,29]. Moreover, an effect size equal to 0.732 represents a
92
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
Table 3 Quality assessments for each study included in the meta-analysis Study
Random assignment
Not described
Single blind
Double blind
Drop-outs
Treatment group N / Total N
Mudie and Matyas [10] Mudie and Matyas [11] Whitall et al. [12] Cauraugh and Kim [13] Cauraugh and Kim [14] Lewis and Byblow [15] McCombe-Waller et al. [16] Stinear and Byblow [17] Luft et al. [18] Cauraugh et al. [19] Summers et al. [20]
1 1 0 1 1 1 0 1 1 1 1
0 0 1 0 0 0 1 0 0 0 0
1 1 0 1 1 1 0 1 1 1 1
0 0 0 0 0 0 0 0 0 0 0
0 0 2 0 0 0 1 0 2 0 0
8/8 4 / 12 14 / 14 10 / 25 26 / 26 6/6 8 / 20 9/9 9 / 21 11 / 26 6 / 12
Note. Consistent with quality assessment suggestions, 1 indicates ‘‘Yes’’ and 0 indicates ‘‘No’’. For the last two columns, the values represent the exact number of subjects.
relatively large effect (e.g., 0.2 = small; 0.5 = medium; 0.8 = large) according to Cohen [28]. This large mean effect was based on 11 studies and 18 individual effect sizes with 111 bilateral movement treatment subjects out of 171 total subjects. Table 4 shows the outcome measures, bilateral movement training protocol, and individual weighted effect sizes with confidence intervals for the studies used in the meta-analysis. 3.2. Fail-safe analysis A fail-safe analysis determines the number findings with null effects necessary to lower the calculated effect size to an insignificant level. For the combination of these bilateral movement findings, 48 null effect results are
necessary to reduce the overall effect size to an insignificant value. 3.3. Homogeneity test Consistent with meta-analytic recommendations, we examined the variability of the weighted effect sizes [27]. Testing for homogeneity across the studies determines whether the results reflect a single underlying effect or a distribution of effects [27,29,31,32]. For the current metaanalysis, the distribution of the effect sizes was heterogeneous as determined by the Q statistic, v 2 (17, n = 18) = 30.65, p < 0.02. Even though the overall effect size was significant, the Q statistic finding warrants further analysis of potential moderating variables [29,32].
Table 4 Summary statistics for total sample in the meta-analysis Study
Outcome measure
Bilateral training
Weighted effect size
Mudie and Matyas [10]
Study 1: kinematic performance; percent deficit
Block placement Simulated drinking Peg targeting Block placement Simulated drinking Peg targeting Block placement Simulated drinking BATRAC Wrist/finger extension Wrist/finger Extension Upper extremity tasks BATRAC Synchronized: active – passive bilateral training Asynchronized: active – passive bilateral training BATRAC Wrist/finger extension Upper extremity reaching: Dowel placement task
1.203 1.883 1.938 2.692 2.609 2.155 3.533 2.738 0.341 0.457 0.242 0.630 0.603 0.368
0.019 – 2.425 0.532 – 3.235 0.574 – 3.302 1.139 – 4.246 1.079 – 4.120 0.741 – 3.570 1.733 – 5.333 1.172 – 4.304 0.122 – 0.561 0.008 – 0.652 0.130 – 0.355 0.009 – 1.250 0.176 – 1.031 0.424 – 1.159
0.690
0.468 – 1.849
0.783 0.730 0.410
0.394 – 1.172 0.304 – 0.890 0.203 – 1.034
Study 2: kinematic performance; percent deficit Mudie and Matyas [11] Whitall et al. [12] Cauraugh and Kim [13] Cauraugh and Kim [14] Lewis and Byblow [15] McCombe-Waller et al. [16] Stinear and Byblow [17]
Luft et al. [18] Cauraugh et al. [19] Summers et al. [20]
Study 3: kinematic performance; percent deficit Fugl-Meyer test: upper extremity Box and Block Test Box and Block Test Fugl-Meyer test: upper extremity Fugl-Meyer test: upper extremity Fugl-Meyer test: upper extremity
Fugl-Meyer test: upper extremity Movement time Motor assessment scale
Confidence intervals
Note. BATRAC = bilateral arm training with rhythmic auditory cuing. Confidence intervals were calculated with standard error values.
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
3.4. Moderating variables As shown in Table 2, five studies used one protocol and six studies used coupled protocols (i.e., two protocols simultaneously). However, with multiple experiments reported in the studies, the total number of individual weighted effect sizes for the single treatment protocols equaled 12 whereas the coupled protocols remained at six. A one-way ANOVA indicated a larger effect size for the various single protocols (M = 1.78, S.D. = 1.01) than the various coupled protocols (M = 0.41, S.D. = 0.23), F(1, 16) = 10.46, p < 0.005. 3.5. Second meta-analysis: mean effect size A second meta-analysis was conducted to avoid any potential bias in the overall weighted effect size. Specifically, the three independent studies of Mudie and Matyas had multiple bilateral training sessions (see Table 4). Within the guidelines of the meta-analytic technique, including each of the training sessions as separate studies was acceptable [27,31,32]. However, to ensure that an inordinate influence to the overall effect size was not coming from one laboratory, this second meta-analysis used only one bilateral training session, block placement data. Thus, removing the simulated drinking and peg targeting data reduced the number of effect sizes from 18 to 13. The 13 individual effect sizes were calculated on data from 11 separate articles and two additional studies. Mudie and Matyas’ [10] studies 1 and 2 were included as well as Stinear and Byblow’s [17] two different experiments with different subjects (i.e., synchronous and asynchronous active – passive bilateral training). Even with the reduced numbers of studies, the overall weighted effect still reached a medium size, ES = 0.582 (S.D. = 0.14). As revealed by the lower and upper confidence intervals (.49 to .67), this second meta-analysis was significant, and a new fail-safe test indicated that 25 null effect studies would have to be reported to reduce the effect size to 0.2. Moreover, the Q statistic [v 2 (12, n = 13) = 13.00, p > 0.05] indicated a homogenous group of studies in this second meta-analysis. Thus, overall these bilateral movement training meta-analysis results can be interpreted with confidence. Further, the moderating variables comparison of single bilateral treatment tasks versus coupled simultaneous treatment protocols was not significant. In the second meta-analysis, bilateral movement training alone improved motor capabilities as well as bilateral movements coupled with auditory rhythmic cuing and active neuromuscular stimulation.
4. Discussion Although bilateral movement training has been argued to be a viable stroke rehabilitation protocol [4,10 – 26], some
93
studies have reported inconsistent motor recovery improvements from the intervention. The current systematic review and prospective meta-analysis evaluated the findings in stroke rehabilitation studies using bilateral movements. An initial meta-analysis of 18 weighted effect sizes and 111 total subjects revealed an overall effect size of 0.732. A more conservative second meta-analysis with five less studies/experiments indicated a 0.582 effect size. Most importantly, the magnitude of both cumulative effect sizes indicate that bilateral movements, either as a single protocol or as a dual protocol with rhythmic auditory cuing or active stimulation, is an effective rehabilitation protocol for improving motor capabilities post-stroke. Indeed, both overall effect sizes support the argument that significant upper extremity motor recovery gains evolve from activity-dependent interventions based on theoretical motor control constructs founded on the behavioral and neurophysiological mechanisms involved in interlimb coordination [4]. Specifically, the efficacy of bilateral movements alone or in combination with other protocols provide further evidence that the apparent default organization of the motor system towards the coupling of muscles directly improves motor performance in paretic limbs post-stroke [4]. Possible neural mechanisms underlying bilateral movements abound. A basic assumption of the use of bilateral movement therapy is that symmetrical (in-phase) bilateral movements activate similar neural networks in both hemispheres when homologous muscle groups are simultaneously activated [9,33 – 39]. Bilateral symmetrical movements, therefore, may allow the activation of the undamaged hemisphere to increase activation of the damaged hemisphere and facilitate movement control of the impaired limb promoting neural plasticity [4,37]. Given that both hemispheres are activated during symmetrical bilateral tasks, researchers have proposed a central regulatory mechanism controlling both limbs [5]. Projections of the supplementary motor area (SMA) make this mechanism a viable bilateral movement executive function structure. The SMA in each hemisphere projects to the ipsilateral primary cortex and to a lesser extent to homologous muscles in the contralateral primary motor cortex [4,7]. Indeed, recent imaging studies have confirmed the higher-order control of bilateral arm movements by the SMA, as well as identifying four additional areas involved during interlimb coordination: (1) cingulate motor cortex, (2) lateral premotor cortex, (3) superior parietal cortex, and (4) cerebellum [4,6,34,37]. However, it is not clear whether symmetrical (in-phase) movements are necessary for bilateral movement training to be effective. For example, improved motor recovery has been shown from training programs involving both in-phase and anti-phase bilateral pushing/pulling movements combined with rhythmic auditory cuing [12,16,18]. Moreover, both synchronous and asynchronous active– passive bilateral movement therapy may differentiate stroke recovery
94
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95
[17]. Thus, symmetrical and asymmetrical bilateral movement training should be further investigated to clarify the contribution of each coordination mode. Additional considerations for future stroke rehabilitation research include determining the effect of the (1) level of intensity of bilateral movement therapy on specific motor outcome measures, (2) duration of bilateral movement training on both immediate and long-term motor improvements, and (3) most effective combination of bilateral movement training and supplementary assistive protocols (e.g., auditory cues, active neuromuscular stimulation, or rhythmic cadence). Investigating these specific bilateral rehabilitation concerns for each of the three stages of recovery would contribute immensely to our understanding of stroke induced motor capabilities. As stated by Winstein et al., post-stroke motor learning (relearning) presents multiple avenues for future research [40]. Moreover, investigating the long-term cumulative effect of bilateral movement protocols with a robust sample size of chronic stroke patients would differentiate temporary learning from permanent learning [41]. The favorable overall effect of bilateral movement training in assisting stroke motor recovery is compelling even with the inherent limitations of the meta-analysis technique. First, criteria for admission into the metaanalysis can directly influence the overall mean effect size. However, rather than accepting all bilateral movement studies, the present analysis followed a conservative approach and clustered studies on outcome measures that focused on arm/hand functions. Second, homogeneity tests on the distributions of the effect sizes for the functional outcome measures revealed heterogeneity in the primary meta-analysis and homogeneity in the additional metaanalysis. Thus, potential moderating variables were investigated to further elaborate the present findings. For the initial meta-analysis, a comparison of only bilateral movement tasks versus coupled (two simultaneously combined) protocols revealed a motor advantage to the various single treatment protocols. However, the second meta-analysis indicated that bilateral movement training alone as well as bilateral movements coupled with either rhythmic auditory cues or active neuromuscular stimulation increased motor improvements. In addition, a fail-safe analysis was conducted to counter the argument that this meta-analytic procedure reports publication bias because we only included one article that has not been published in refereed journal. Based on this conservative analysis, 48 studies with null effects are necessary to lower the large calculated effect size (0.732) to a small, insignificant level. Moreover, the second meta-analysis revealed a medium effect size (0.582), 25 null effect studies for the fail-safe analysis, and homogeneous variability in the reduced number of bilateral studies. Thus, based on the two present metaanalyses four lines of evidence indicate a relatively robust bilateral movement training effect: (1) overall weighted
effect sizes, (2) fail-safe analyses, (3) Q statistics, and (4) quality of study ratings. In conclusion, the magnitude of the overall mean effects of these meta-analyses taken together with the conservative approach, the additional statistical tests conducted, as well as the quality of study ratings demonstrate that bilateral movements are effective in improving motor functions for the sub-acute and chronic phases of stroke recovery. These bilateral movement protocol (e.g., alone or in conjunction with auxiliary sensory feedback) findings extend Whitall’s persuasive argument that rehabilitation research should focus on maximizing benefits from practicing motor actions [42]. Rehabilitation specialists are encouraged to incorporate the findings from this systematic review and two metaanalyses into their informed decisions on protocols for improving motor capabilities and making progress toward stroke motor recovery [42 – 45].
Acknowledgements James Cauraugh was supported by an award from the American Heart Association, Florida/Puerto Rico Affiliate and as an affiliated investigator of the Brain Rehabilitation Research Center, Research Service, North Florida/South Georgia Veteran’s Health System, Gainesville, FL. Jeffery Summers was supported by awards from the Australian Research Council and the Australian Health Management Group - Medical Research Fund.
References [1] Hendricks HT, van Limbeck J, Geurts AC, Zwarts MJ. Motor recovery after stroke: a systematic review of the literature. Arch Phys Med Rehabil 2002;83:1629 – 37. [2] Bernstein NA. The co-ordination and regulation of movements. Oxford, UK’ Pergamon Press; 1967. [3] Swinnen SP. Intermanual coordination: from behavioural principles to neural-network interactions. Nat Rev Neurosci 2002;3:350 – 61. [4] Cauraugh JH, Summers JJ. Neural plasticity and bilateral movements: a rehabilitation approach for chronic stroke. Progress Neurobiol 2005;75:309 – 20. [5] Cohen L. Interaction between limbs during bimanual voluntary activity. Brain 1970;93:259 – 72. [6] Debaere F, Wenderoth N, Sunaert S, Van Hecke P, Swinnen SP. Changes in brain activation during the acquisition of a new bimanual coordination task. Neuropsychologia 2004;42:855 – 67. [7] Goldberg G. Supplementary motor area structure and function: review and hypotheses. Behav Brain Sci 1985;8:567 – 616. [8] Swinnen SP, Wenderoth N. Two hands, one brain: cognitive neuroscience of bimanual skill. Trends Cog Neurosci 2004;8:18 – 25. [9] Wenderoth N, Debaere F, Suraert, van Hecke P, Swinnen SP. Parietopremotor areas mediate directional interference during bimanual movements. Cereb Cortex 2004;14:1153 – 63. [10] Mudie MH, Matyas TA. Upper extremity retraining following stroke: effects of bilateral practice. J Neural Rehabil 1996;10: 167 – 84. [11] Mudie MH, Matyas TA. Can simultaneous bilateral movement involve the undamaged hemisphere in reconstruction of neural networks damaged by stroke? J Disabil Rehabil 2000;22:23 – 37.
K.C. Stewart et al. / Journal of the Neurological Sciences 244 (2006) 89 – 95 [12] Whitall J, Waller S, Silver K, Macko R. Repetitive bilateral arm training with rhythmic auditory cueing improves motor function in chronic hemiparetic stroke. Stroke 2000;31:2390 – 5. [13] Cauraugh JH, Kim SB. Two coupled motor recovery protocols are better than one: electromyogram-triggered neuromuscular stimulation and bilateral movements. Stroke 2002;33:1589 – 94. [14] Cauraugh JH, Kim SB. Chronic stroke motor recovery: duration of active neuromuscular stimulation. J Neurol Sci 2003;215: 13 – 9. [15] Lewis GN, Byblow WD. Neurophysiological and behavioral adaptations to a bilateral training intervention in individuals following stroke. Clin Rehabil 2004;18:48 – 59. [16] McCombe-Waller S, Whitall J. Fine motor control in adults with and without chronic hemiparesis: baseline comparison to nondisabled adults and effects of bilateral arm training. Arch Phys Med Rehabil 2004;85:1076 – 83. [17] Stinear JW, Byblow WD. Rhythmic bilateral movement training modulates corticomotor excitability and enhances upper limb motoricity poststroke: a pilot study. J Clin Neurophys 2004; 21:124 – 31. [18] Luft AR, McCombe-Waller S, Whitall J, Forrester LW, Macko R, Sorkin JD, et al. Repetitive bilateral arm training and motor cortex activation in chronic stroke: a randomized controlled trial. JAMA 2004;292:1853 – 61. [19] Cauraugh JH, Kim SB, Duley A. Coupled bilateral movements and active neuromuscular stimulation: intralimb transfer evidence during bimanual aiming. Neurosci Lett 2005;382:39 – 44. [20] Summers JJ, Garry MI, Kagerer FA, Hiraga CY, Loftus A. Bilateral training and recovery of upper arm function after stroke. Paper presented at the Xth International Symposium on Motor Control, Sofia, Bulgaria; 2004. [21] Dickstein R, Hocherman S, Amdor G, Pillar T. Reaction and movement times in patients with hemiparesis for unilateral and bilateral elbow flexion. Phys Ther 1993;73:374 – 80. [22] Roby-Brami A, Fuchs S, Mokhtari M, Bussel B. Reaching and grasping strategies in hemiparetic patients. Motor Control 1997;1:72 – 91. [23] Mudie MH, Matyas TA. Responses of the densely hemiplegic upper extremity to bilateral training. Neurorehabil Neural Repair 2001;15:129 – 40. [24] Cunningham CL, Phillips Stoykov ME, Walter CB. Bilateral facilitation of motor control in chronic hemiplegia. Acta Psychol 2002;110:321 – 37. [25] Rice MS, Newell KM. Interlimb coupling in left hemiplegia because of right cerebral vascular accident. Occup Ther J Res 2001;21:12 – 28. [26] Rice MS, Newell KM. Interlimb coupling in left hemiplegia because of right cerebral vascular accident. Arch Phys Med Rehabil 2005; 85:629 – 34. [27] Rosenthal R, DeMatteo MR. Meta-analysis: recent developments in quantitative methods for literature reviews. Annu Rev Psychol 2001; 52:59 – 82.
95
[28] Cohen J. Statistical power analysis for the behavioral sciences, 2nd ed. Hillsdale, NJ’ Erlbaum; 1988. [29] Rosenthal R. Writing meta-analytic reviews. Psychol Bull 1995;118:1173 – 81. [30] Thompson SG. Meta-analysis of clinical trials. In: Armitage P, Colton T. editors. Encyclopedia of biostatistics, vol. 4. New York’ Wiley; 1998. p. 2570 – 9. [31] Sutton AJ, Abrams KR, Jones DR, Sheldon TA, Song F. Methods for meta-analysis in medical research. New York’ Wiley; 2000. [32] Hedges LV, Olkin I. Statistical methods for meta-analysis. Orlando’ Academic Press; 1985. [33] Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJM, Gavaghan DJ, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996;17: 21 – 12. [34] Moher D, Schulz KF, Altman D. The CONSORT statement: revised recommendations for improving the quality of reports of parallelgroup randomized trials. JAMA 2001;285:1987 – 91. [35] Hallett M. Plasticity of the human motor cortex and recovery from stroke. Brain Res Rev 2001;36:169 – 74. [36] Lacroix S, Havton LA, McKay H, Yang H, Brant A, Roberts J, et al. Bilateral corticospinal projections arise from each motor cortex in the macaque monkey: a quantitative study. J Comp Neurol 2004;473: 147 – 61. [37] Carson RG. Neural pathways mediating bilateral interactions between the upper limbs. Brain Res Rev 2005;49:641 – 62. [38] Spijkers W, Heuer H. Behavioral principles of interlimb coordination. In: Swinnen SP, Duysens J, editors. Neuro-behavioral determinants of interlimb coordination: a multidisciplinary approach. Boston’ Kluwer; 2004. p. 223 – 58. [39] Jancke L, Peters M, Himmelbach M, Nosselt T, Shah J, Steinmetz H. fMRI study of bimanual coordination. Neuropsychologia 2000; 38:164 – 74. [40] Winstein CJ, Wing AM, Whitall J. Motor control and learning principles for rehabilitation of upper limb movements after brain injury. In: Boller F, Grafman J, Robertson IH, editors. Handbook of neuropsychology. Amsterdam’ North Holland; 2003. p. 77 – 137. [41] Cauraugh JH, Kim SB, Summers JJ. Cumulative neural plasticity: longitudinal motor improvements. 2005. Currently submitted for publication. [42] Whitall J. Stroke rehabilitation research: time to answer more specific questions? Neurorehabil Neural Repair 2004;18:3 – 8. [43] Cauraugh JH. Coupled rehabilitation protocols and neural plasticity: upper extremity improvements in chronic hemiparesis. Restor Neurol Neurosci 2004;22:337 – 47. [44] Bolton DAE, Cauraugh JH, Hausenblas HA. Electromyogramtriggered neuromuscular stimulation and stroke motor recovery of the upper limb: a meta-analysis. J Neurol Sci 2004;223:121 – 7. [45] Cohen J. Things I have learned (so far). Am Psychol 1990;45: 1304 – 12.