Unilateral and Bilateral Upper-Limb Training Interventions After Stroke Have Similar Effects on Bimanual Coupling Strength

Abstract

Background. Bilateral training in poststroke upper-limb rehabilitation is based on the premise that simultaneous movements of the nonparetic upper limb facilitate performance and recovery of paretic upper-limb function through neural coupling effects. Objective. To determine whether the degree of coupling between both hands is higher after bilateral than after unilateral training and control treatment. Methods. In a single-blinded randomized controlled trial, we investigated rhythmic interlimb coordination after unilateral (mCIMT) and bilateral (mBATRAC) upper-limb training and a dose-matched control treatment (DMCT) in 60 patients suffering from stroke. To this end, we used a series of tasks to discern intended and unintended coupling effects between the hands. In addition, we investigated the control over the paretic hand as reflected by movement harmonicity and amplitude. All tasks were performed before and after a 6-week intervention period and at follow-up 6 weeks later. Results. There were no significant between-group differences in change scores from baseline to postintervention and from postintervention to follow-up with regard to interlimb coupling. However, the mBATRAC group showed greater movement harmonicity and larger amplitudes with the paretic hand after training than the mCIMT and DMCT groups. Conclusions. The degree of coupling between both hands was not significantly higher after bilateral than after unilateral training and control treatment. Although improvements in movement harmonicity and amplitude following mBATRAC may indicate a beneficial influence of the interlimb coupling, those effects were more likely due to the particular type of limb movements employed during this training protocol.

Keywords

stroke upper limb rehabilitation kinematics bilateral arm training constraint-induced movement therapy

Introduction

Many, if not most, of our daily activities involve the simultaneous use of both hands.¹ One of the main arguments to provide bilateral training to patients suffering from a unilateral upper-limb paresis following stroke capitalizes on this basic observation.² Bilateral arm training protocols, such as Bilateral Arm Training with Rhythmic Auditory Cueing (BATRAC),^3-5 are based on the premise that, through neural coupling effects, regaining paretic upper-limb functionality is facilitated by simultaneous movements of the nonparetic upper limb. These coupling effects are engendered through intact neural pathways connecting both sides of the central nervous system, such as the corpus callosum.^6-8

In contrast to bilateral arm training, there are also protocols in which training is restricted to the most affected arm, such as (modified versions of) constraint-induced movement therapy (CIMT).⁹ A meta-analysis and recent experimental studies directly comparing the effects of equally intensive bilateral and unilateral upper-limb training have shown that both types of training are similarly effective in terms of upper-limb motor impairment, (perceived) upper-limb functioning, and dexterity.^10-12 These findings suggest that the improved functionality of the paretic upper limb after bilateral training does not necessarily result from exploitation of the neurally driven coupling effects between the upper limbs but simply from active training of the paretic upper limb. Nevertheless, bilateral training may be expected to have the additional advantage of improving bimanual coordination, thanks to reliance on this coupling during the training sessions.¹³

Coupling effects have been investigated extensively in studies of rhythmic interlimb coordination involving healthy subjects^14-20 and may be altered due to stroke.²¹ In the single-blinded randomized controlled Upper Limb TRaining After stroke (ULTRA-stroke) trial, we investigated rhythmic interlimb coordination together with the clinical merits of unilateral and bilateral upper-limb training and a dose-matched control treatment in patients suffering from stroke.²²

Based on the premise that in bilateral training neurally engendered coupling influences between both upper limbs are exploited, and in line with empirically observed changes in intended interlimb coupling due to practice,¹³ we hypothesized that the degree of intended coupling between the hands would be stronger after bilateral than after unilateral training and control treatment. The degree of coupling was assessed using rhythmic bimanual coordination tasks in which patients were instructed to simultaneously flex and extend the wrists in specific coordination patterns. We anticipated that a change in coupling would benefit the control of both hands, thanks to a higher degree of control over the paretic hand (examined using active movements) and improved processing of its sensory signals (examined using passive movements), which might be affected after stroke.²³ Because unintended coupling was not targeted in any of the interventions, unintended entrainment effects between the hands could be expected to be similar in all training groups. Alternatively, it was also deemed possible that bilateral training would help patients rely on this source of coupling, or to become more competent in modulating its unintended influences. The unintended entrainment was examined using tasks in which patients were instructed to ignore passive movements of one of their hands while rhythmically moving their other hand at a tempo specified by an auditory pacing signal.

In addition to changes in the degree of coupling between the hands, we also investigated changes in amplitude and movement harmonicity^18,24 of the paretic hand. These measures served as indicators of control over the paretic hand. Amplitude and movement harmonicity were expected to increase similarly in the 3 intervention groups.^11,12

Methods

The ULTRA-Stroke trial was registered (URL: http://www.trialregister.nl; Unique identifier: NTR1665) and approved by the Medical Ethical Reviewing Committee of VU University Medical Center (protocol number 2008/296, Dutch Central Committee on Research Involving Human Subjects, CCMO, protocol number NL20456.029.08). A detailed description of the ULTRA-stroke trial has been presented elsewhere.²² Relevant details of the trial are presented below.

Participants and Trial

Sixty patients admitted to rehabilitation center Reade in Amsterdam with a first-ever stroke and an upper-limb paresis with at least minimal distal control (ie, able to execute at least 10° of active wrist extension, 10° of active thumb abduction/extension, and 10° active extension in at least 2 additional digits) were recruited. After obtaining informed consent, a baseline assessment of outcome variables was performed. Next, patients were randomized in permuted blocks and allocated to 1 of 3 intervention groups: modified CIMT (mCIMT; N = 22), modified BATRAC (mBATRAC; N = 19), or dose-matched control treatment (DMCT; N = 19). Concealed allocation was effectuated online, using the minimization method. Patient characteristics are presented in Table 1. The postintervention tests were conducted in the week after a 6-week intervention period and followed by follow-up tests 6 weeks later.

Table 1.

Baseline Characteristics.

			mCIMT	mBATRAC	DMCT	χ² (df)	Between-Subject P^a
Female/male			8/14	8/11	3/16	3.39 (2)	.183
Arm affected, left/right			12/10	11/8	11/8	0.06 (2)	.969
Affected dominant side, yes/no			11/11	9/10	7/12	0.78 (2)	.678
			Mean (SD; N)	Mean (SD; N)	Mean (SD; N)	F (df)
Age, years			59.8 (13.8; 22)	62.6 (9.8; 19)	56.9 (12.7; 19)	1.02 (2, 59)	.367
Time since stroke, weeks			9.2 (6.8; 22)	7.8 (4.9; 19)	11.1 (6.8; 19)	1.33 (2, 59)	.273
						H (df)
ARAT			23.8 (16.8; 22)	25.8 (16.1; 19)	23.3 (15.4; 19)	0.14 (2)	.932
Task	(hand)	Measure	Mean (SD; range; N)	Mean (SD; range; N)	Mean (SD; range; N)	F (df)
Bimanual Coord. Antiphase	(both)	CC	−0.73 (0.22; −0.95 to −0.22; 18)	−0.81 (0.16; −0.95 to −0.47; 11)	−0.81 (0.21; −0.96 to −0.26; 12)	0.79 (2, 40)	.461
	(paretic)	Harm	0.79 (0.14; 0.42 to 0.93; 18)	0.80 (0.12; 0.58 to 0.93; 11)	0.82 (0.12; 0.57 to 0.94; 12)	0.26 (2, 40)	.771
	(paretic)	Ampl	13.2 (8.7; 0.2 to 30.2; 18)	13.4 (14.0; 0.4 to 37.0; 11)	10.4 (7.4; 1.8 to 22.5; 12)	0.35 (2, 40)	.705
Bimanual Coord. In-phase	(both)	CC	0.85 (0.17; 0.27 to 0.97; 22)	0.89 (0.10; 0.61 to 0.98; 16)	0.89 (0.10; 0.58 to 0.98; 16)	0.69 (2, 53)	.532
	(paretic)	Harm	0.86 (0.11; 0.57 to 0.96; 22)	0.85 (0.08; 0.57 to 0.96; 16)	0.84 (0.12; 0.50 to 0.95; 16)	0.16 (2, 53)	.853
	(paretic)	Ampl	15.1 (12.0; 0.1 to 44.8; 22)	16.8 (14.1; 1.2 to 40.7; 16)	15.1 (10.5; 2.4 to 37.7; 16)	0.11 (2, 53)	.893
Kinest. Tracking Antiphase	(paretic)	CC	−0.82 (0.15; −0.47 to −0.95; 15)	−0.80 (0.15; −0.53 to −0.97; 12)	−0.67 (0.23; −0.26 to −0.96; 12)	2.75 (2, 38)	.077
Kinest. Tracking Antiphase	(nonparetic)	CC	−0.85 (0.17; −0.29 to −0.99; 15)	−0.75 (0.25; −0.19 to −0.96; 13)	−0.81 (0.18; −0.26 to −0.97; 14)	0.91 (2, 41)	.411
Kinest. Tracking In-phase	(paretic)	CC	0.88 (0.11; 0.49 to 0.96; 20)	0.86 (0.10; 0.63 to 0.96; 15)	0.83 (0.18; 0.39 to 0.96; 17)	0.72 (2, 51)	.491
Kinest. Tracking In-phase	(nonparetic)	CC	0.88 (0.17; 0.31 to 0.99; 17)	0.91 (0.07; 0.69 to 0.97; 13)	0.84 (0.24; 0.17 to 0.98; 17)	0.55 (2, 46)	.582
Unimanual Motor	(paretic)	RP Shift	−17.9 (29.8; −95.0 to 28.1; 18)	−26.7 (30.4; −110.9 to 11.5; 13)	−8.8 (19.0; −29.5 to 25.9; 13)	1.39 (2, 43)	.261
Unimanual Motor	(nonparetic)	RP Shift	−14.0 (16.9; −63.4 to 10.1; 19)	−20.0 (17.6; −63.4 to 4.3; 13)	−3.4 (25.0; −37.0 to 63.8; 15)	2.50 (2, 46)	.094
Unimanual Ref.	(paretic)	Harm	0.90 (0.06; 0.73 to 0.96; 19)	0.86 (0.05; 0.77 to 0.93; 16)	0.84 (0.07; 0.71 to 0.93; 16)	4.39 (2, 50)	.018
	(paretic)	Ampl	19.7 (12.1; 3.2 to 44.7; 19)	17.4 (15.2; 1.4 to 40.3; 16)	14.8 (8.8; 3.1 to 37.2; 16)	0.69 (2, 50)	.508

Abbreviations: mCIMT, modified constraint-induced movement therapy; mBATRAC, modified bilateral arm training with rhythmic auditory cueing; DMCT, dose-matched control treatment; SD, standard deviation of the mean; N, number of participants; df, degrees of freedom; ARAT, Action Research Arm Test; CC, cross-correlation coefficient; Harm, movement harmonicity; Ampl, amplitude; RP, relative phase.

Significant between-subject P values are in boldface.

The mCIMT therapy involved unilateral repetitive task practices and shaping of the desired movements.^9,25 The exercises followed a quasihierarchical bottom-up approach from more relatively simple gross motor functions of the upper limb to more complex in-hand manipulations and combinations of movements in activities of daily life. Patients were encouraged to wear a mitt on the nonparetic hand for 6 hours each weekday. The mBATRAC group followed a modification of the original BATRAC protocol^3,4 that entailed rhythmic flexion and extension movements about the wrist rather than movements of proximal parts of the upper limb. The apparatus used for mBATRAC was mounted on a chair with arm rests. At the distal end of each arm rest a manipulandum with a handgrip was fitted. Rhythmic wrist rotations in the horizontal plane were paced by an auditory metronome at an individually selected tempo between 0.8 Hz and 1.8 Hz. Over the course of training the tempo was adjusted in response to improvement in task performance. In the mCIMT and mBATRAC training protocols, emphasis was placed on the increase of control of wrist and finger extensors, given its importance for functional recovery.^26-30 The DMCT was the usually applied exercise therapy based on existing guidelines for upper-limb rehabilitation after stroke.^31,32 All patients received 60-minute therapy sessions, 3 days a week for 6 consecutive weeks. They were instructed to also practice outside therapy hours and encouraged to perform activities of daily living according to the concept of their allocated treatment. As self-practice was inconsistently documented in the patients’ diaries, the assumption of equivalent self-practice between groups relied on verbal staff reports.

The clinical results of the ULTRA-stroke trial are presented in detail elsewhere.¹² For the present study, it is noteworthy that, although all groups showed significant improvement, the clinical outcomes did not differ significantly between intervention groups in the ULTRA-stroke trial.¹²

Apparatus

Participants sat on a chair with their elbows slightly flexed. Their forearms were placed and fixated on armrests in a neutral position. Both hands were positioned in 2 manipulanda with the palms facing inward. The manipulanda, which allowed only flexion and extension movements about the wrist, could either register wrist movements by means of a potentiometer (FCP40A, tolerance ±0.1%, Sakae Tsushin Kogyo Co, Ltd, Nakahara-ku, Kawasaki-city, Japan; sampling rate: 200 Hz) or control wrist movements by means of a servo-motor (Parvex RS440GR1031, SSD Parvex SAS, Dijon Cedex, France) and a precision gearbox (alpha TP010S-MF1-7-0C0, backlash ±0.02°, Wittenstein, Inc, Bartlett, IL), for active and passive movements, respectively. A gray opaque screen prevented vision of the hand movements. Auditory pacing stimuli (pitch: 440 Hz; duration: 50 ms; one beep per cycle) were presented through headphones.

Tasks

The current study exploited 4 experimental tasks that are known to differ with regard to the types of interlimb coupling they entail (based on Ridderikhoff et al¹⁹). Participants executed these tasks at baseline, postintervention, and follow-up. Instructions were provided before each task at each test. Participants executed 2 trials per task. If the participant failed to perform a trial (as judged by the experimenter), the trial was repeated up to a maximum of 4 attempts per task.

For each participant the highest possible movement frequency, to be used for all tasks in a given test, was assessed prior to each test based on the Bimanual Coordination task (see below). Starting with a pacing frequency of 0.6 Hz, the frequency was increased in steps of 0.2 Hz until the participant indicated that the frequency was the highest possible frequency that could be maintained during testing. This frequency was maintained throughout the test.

In the Bimanual Coordination task participants executed 50 cyclic wrist movements with both hands at a tempo specified by a rhythmic auditory pacing signal. In healthy individuals this type of coordination is characterized by bilateral coupling between the hands, which stabilizes the bimanual coordination pattern.^33,34 This intended coupling involves both afferent and efferent signals.¹⁹ Participants were instructed to let peak flexion of both hands coincide with the beeps in the in-phase pattern. In the antiphase pattern, peak flexion of the right hand and peak extension of the left hand had to coincide with the beeps.

In the Kinesthetic Tracking task 30 cycles were executed. One hand was moved passively (by means of the motor) and participants were instructed to track these movements with their contralateral, actively moving hand, either in a mirror symmetrical fashion (in-phase pattern) or an alternating fashion (antiphase pattern). As no auditory pacing signals were provided (and vision of the hands was occluded), these active movements (and, hence, the intended coupling between the hands) had to rely on afferent signals stemming from the contralateral (passively moving) hand. The Kinesthetic Tracking task was performed with both the paretic and nonparetic hands as actively moving hand. Ideally, the passive movement trajectories for the Kinesthetic Tracking task are identical to the kinematics of the recorded (active) movements during bimanual coordination,^18,34 such that the variability in amplitude and cycle duration is identical to that of the active movements during the Bimanual Coordination task. However, for about 40% of the participants we could not use the movement trajectories obtained for the Bimanual Coordination task as signal for the passive (motor) movements, because they contained too many irregularities (eg, pattern transitions). We anticipated this problem and created passive movement trajectories for each participant by programming a sinusoidal pattern with added random noise to the cycle duration (2%) and amplitude (5%), using a maximum amplitude of 95% of the maximum amplitude as recorded (for each individual) in the Bimanual Coordination task.

The Unimanual Motor task (unimanual coordination, with simultaneous motor-driven movements of the contralateral hand) also involved cyclic passive wrist movements of one hand, but in this case the participants had to ignore these movements, while rhythmically moving their other hand at a tempo specified by the auditory pacing signal. For each individual, these passive movements were identical to the adapted sinusoids used in the Kinesthetic Tracking task. In this task 65 cycles were executed. Participants were instructed to let peak flexion of the active hand coincide with the metronome beeps and to ignore the movements of the passively moving hand. For the first 30 cycles, peak flexion of the passive hand coincided with the metronome beeps. In the next 5 cycles, the passive hand movements were gradually shifted to induce a phase difference of −30° between the passive hand and the metronome (ie, peak flexion of the passive hand was leading the metronome in time). During the last 30 cycles, the phase relation between the passive hand and the metronome was maintained at −30°. The Unimanual Motor task was performed with both the paretic and nonparetic hands as actively moving hand. Because the passive movements were phase-shifted with respect to the metronome beeps, unintended coupling influences stemming from the passively moving hand could be discerned. Such influences result from afferent signals stemming from the passive hand and are characterized by attraction (entrainment) of the active hand’s movements toward those of the passive hand.^19,35,36

In the Unimanual Reference task participants executed 30 cycles with their paretic hand, while synchronizing peak flexion to the metronome beeps. This task served as a reference condition, in which no coupling between the hands was involved.

Data Analysis

For all tasks, the potentiometer signals were converted from Volts to degrees, and then band-pass filtered (second-order bidirectional Butterworth filter, cut-off frequencies 0.2 Hz and 18 Hz). To avoid transient effects at the start of the trial, the first 4 cycles were removed.

Trials with a mean frequency deviation >0.1 Hz from the prescribed frequency were discarded, as well as trials with an incorrect pattern (in-phase/antiphase). On average, about 12% of the trials were discarded, ranging from 6% for the Unimanual Reference task at the posttest to 28% for the antiphase Bimanual Coordination task at baseline. The number and distribution of discarded trials did not differ significantly over the groups (χ² test). A positive cross-correlation coefficient between both hand movements around lag 0 (±90°) indicated an in-phase coordination pattern, whereas a negative value indicated an antiphase coordination pattern.

The highest absolute value of the cross-correlation coefficient (CC; normalized for movement amplitude) was used as measure of intended coupling strength in the Bimanual Coordination and Kinesthetic Tracking tasks. If the highest absolute value of the CC was found for a positive lag, the nonparetic hand was leading the paretic hand in time (relative to the identified coordination pattern), whereas the reverse was true for a negative lag. The shift in relative phase (RP shift) between the active hand and metronome in the Unimanual Motor task was used as measure of unintended coupling influence, that is, entrainment to the phase-shifted passive hand movements.^19,35,36 This shift was calculated as the mean relative phase in the 25 cycles before the last cycle (ie, when the passive movements were −30° phase shifted with respect to the metronome) minus the mean relative phase in the 25 cycles following the removed first 4 cycles (ie, when the passive movements were synchronized with the metronome).

The mean amplitudes of the paretic hand were calculated for the Bimanual Coordination and Unimanual Reference tasks only, to assess the range of motion about the paretic wrist. The Unimanual Reference task served as an uncoupled reference for the coupled situation in the Bimanual Coordination task. For each movement cycle, the amplitude was defined as the distance between peak extension and peak flexion divided by 2. The obtained amplitudes were averaged for each individual trial. Segments in the signal without movement (ie, when for more than one cycle duration the absolute difference between consecutive samples was less than 20% of the mean absolute difference between consecutive samples of the complete signal) were discarded. These were found in less than 4% of all trials and ranged from 2 to 5 cycles.

Movement smoothness has been used as a measure of motor performance in patients with stroke.^37-39 In a cyclic task with a fixed frequency, additional accelerations and decelerations (relative to sinusoidal movements) render the movements less smooth. One way to quantify smoothness is by measuring movement harmonicity. Movement harmonicity was defined as the power of the peak movement frequency ±0.1 Hz (ie, bandwidth of 0.2 Hz) relative to the total power of the frequency spectrum of the complete signal (values between 0 and 1).^18,24 A higher relative power indicates larger movement harmonicity, reflecting smoother movement trajectories. Movement harmonicity was calculated for the paretic hand in the Bimanual Coordination and Unimanual Reference tasks.

Statistical Analyses

We anticipated that not all patients would be able to perform all tasks at every test. Therefore, we used χ² tests to check for group-size differences per task and per test.

We then checked for potential baseline differences between groups using analyses of variance (ANOVAs). To this end, we examined CC (Bimanual Coordination and Kinesthetic Tracking), RP shift (Unimanual Motor), amplitude (Bimanual Coordination and Unimanual Reference), and movement harmonicity (Bimanual Coordination and Unimanual Reference). Given the randomized allocation of participants, no differences between groups were expected at baseline.

We conducted a mixed-model ANOVA on the movement frequencies with Test (baseline, postintervention, follow-up) as within-subject factor and Intervention (mCIMT, mBATRAC, DMCT) as between-subject factor, expecting a similar increase in movement frequency for all intervention groups.

Next, the change scores from baseline to postintervention and from postintervention to follow-up were calculated for CC (Bimanual Coordination and Kinesthetic Tracking), RP shift (Unimanual Motor), amplitude (Bimanual Coordination and Unimanual Reference), and movement harmonicity (Bimanual Coordination and Unimanual Reference) and tested with 2-tailed, 1-sample t tests to determine whether these change scores differed significantly from zero.

Next, these change scores were submitted to ANOVAs with Intervention as between-subject factor. Comparisons were performed separately for each task. From baseline to postintervention, we expected a larger increase in coupling (ie, higher absolute CC) in the Bimanual Coordination and Kinesthetic Tracking tasks (for both hands) after mBATRAC than after mCIMT and DMCT. From postintervention to follow-up, we expected these changes to sustain. There were no specific anticipations related to the changes in RP shift. Amplitude and movement harmonicity were expected to increase from baseline to postintervention in the Bimanual Coordination and Unimanual Reference tasks, similarly for all intervention groups.^11,12

For the patients that were able to execute the Bimanual Coordination and Kinesthetic Tracking tasks in both patterns, we also conducted a repeated-measures ANOVA on CC, with Pattern as within-subject factor and Intervention as between-subject factor for each test and task separately. We expected a main effect of Pattern, with stronger coupling (ie, higher CC) for in-phase than for antiphase coordination for the Bimanual Coordination task^19,40,41 and perhaps also for the Kinesthetic Tracking task.^42-44

In the Results section significant effects (P < .05) are reported. Effect sizes (f) were calculated in terms of η² for the one-way ANOVAs and partial η² for the repeated-measures ANOVAs with small, medium, and large effect sizes being indicated by 0.1, 0.25, and 0.4.⁴⁵ For post hoc analyses we used t tests with a Holm–Bonferroni correction for multiple comparisons.

Results

During the ULTRA-stroke trial 8 patients dropped out (see flowchart in Figure 1).¹² In addition, some patients who were unable to execute one or more tasks at baseline were able to perform these in later test sessions. Therefore, the number of patients that were included for a given task varied over the tests. On average, 14.5 (SD = 2.6) patients per intervention group were included for each test (range = 9-22; for DMCT: antiphase Kinesthetic Tracking task at follow-up and mCIMT: in-phase Bimanual Coordination task at baseline, respectively). For none of the tests significant differences were observed in the number of patients per group that were included for a given task; all χ² tests were nonsignificant.

Figure 1.

Flow chart.

Baseline

Figure 2 depicts movement segments from the 4 tasks performed at baseline by the same patient. Table 1 shows the baseline data. There were no baseline differences between intervention groups, except for movement harmonicity in the Unimanual Reference task. The mCIMT group showed larger movement harmonicity than the DMCT group (P = .009). However, the effect size was small (f = 0.15).

Figure 2.

Movement segments from the 4 tasks.

The movement frequencies increased per test for all intervention groups (F_2,72 = 52.84; P < .001; f = 0.60). There were no significant group differences for any of the tests. The mean movement frequencies over all groups increased from 0.83 Hz (SD = 0.13 Hz) at baseline to 0.99 Hz (SD = 0.12 Hz) postintervention (P < .001), and further to 1.04 Hz (SD = 0.13 Hz) at follow-up (P = .001).

Baseline to Postintervention

Table 2 shows the change scores (significant change scores are indicated with asterisks) and the results of the ANOVAs on the change scores from baseline to postintervention and from postintervention to follow-up. From baseline to postintervention, there were no significant between-group differences in change scores related to intended (CC in the Bimanual Coordination and Kinesthetic Tracking tasks) and unintended (RP shift in the Unimanual Motor task) interlimb coupling. For the Bimanual Coordination and Unimanual Reference tasks significant between-group differences related to changes in movement harmonicity and amplitude were observed from baseline to postintervention. Post hoc analyses on movement harmonicity (see also Figure 3) revealed a larger increase after mBATRAC than after DMCT in the Bimanual Coordination task for the in-phase pattern (P = .015), whereas for the Unimanual Reference task a smaller increase was observed after mCIMT than after mBATRAC (P = .007) and after DMCT (P = .015). Post hoc analyses on amplitude (see also Figure 3) revealed a larger increase after mBATRAC than after DMCT in the Bimanual Coordination task for both antiphase (P = .012) and in-phase coordination (P = .012). Furthermore, there was a larger increase in amplitude after mBATRAC than after mCIMT in the Unimanual Reference task (P = .004).

Table 2.

Comparison of Change Scores for the 3 Intervention Groups^a.

			Change at Postintervention			Between-Subject Analysis			Change at Follow-Up			Between-Subject Analysis
Task	(hand)	Measure	mCIMT	mBATRAC	DMCT	F (df)	P	f	mCIMT	mBATRAC	DMCT	F (df)	P	f
Bimanual Coord. Antiphase	(both)	CC	0.06 (0.19)	0.09 (0.16)	0.01 (0.06)	0.85 (2, 36)	.435	0.05	0.04 (0.09)	−0.05 (0.14)	−0.02 (0.06)	2.64 (2, 34)	.087	0.14
	(paretic)	Harm	0.03 (0.12)	0.10 (0.13)*	0.01 (0.08)	1.82 (2, 36)	.177	0.10	0.06 (0.10)	−0.06 (0.12)	−0.04 (0.06)*	5.45 (2, 34)	.009	0.25
	(paretic)	Ampl	2.36 (5.53)	13.78 (12.19)**	1.45 (5.52)	8.31 (2, 36)	.001	0.33	0.58 (6.60)	−5.20 (8.33)*	1.98 (5.14)	3.48 (2, 34)	.043	0.18
Bimanual Coord. In-phase	(both)	CC	0.04 (0.08)	0.07 (0.09)*	0.02 (0.02)**	1.39 (2, 46)	.261	0.06	−0.01 (0.06)	0.00 (0.02)	−0.01 (0.04)	0.35 (2, 39)	.705	0.02
	(paretic)	Harm	0.03 (0.05)*	0.08 (0.09)**	0.00 (0.06)	4.31 (2, 46)	.020	0.16	0.00 (0.04)	0.00 (0.03)	−0.02 (0.05)	1.30 (2, 39)	.284	0.07
	(paretic)	Ampl	2.28 (7.76)	12.72 (13.94)**	1.58 (4.93)	6.37 (2, 46)	.004	0.22	−1.12 (7.32)	0.43 (7.24)	2.20 (6.07)	0.76 (2, 39)	.474	0.04
Kinest. Tracking Antiphase	(paretic)	CC	0.05 (0.18)	0.08 (0.16)	−0.04 (0.26)	0.96 (2, 32)	.394	0.06	0.00 (0.20)	0.04 (0.13)	0.01 (0.08)	0.18 (2, 31)	.840	0.01
Kinest. Tracking Antiphase	(nonparetic)	CC	−0.05 (0.19)	0.03 (0.32)	0.02 (0.09)	0.49 (2, 32)	.617	0.03	−0.01 (0.08)	0.14 (0.23)	−0.06 (0.10)	5.06 (2, 30)	.013	0.27
Kinest. Tracking In-phase	(paretic)	CC	0.02 (0.09)	0.00 (0.11)	0.06 (0.15)	0.66 (2, 43)	.523	0.03	−0.08 (0.23)	−0.04 (0.22)	0.05 (0.07)*	1.56 (2, 40)	.223	0.08
Kinest. Tracking In-phase	(nonparetic)	CC	−0.03 (0.06)	0.00 (0.12)	0.15 (0.56)	1.10 (2, 37)	.345	0.06	−0.01 (0.10)	0.01 (0.14)	−0.03 (0.06)	0.31 (2, 37)	.738	0.02
Unimanual Motor	(paretic)	RP Shift	13.77 (24.99)	14.09 (36.67)	9.23 (32.43)	0.09 (2, 37)	.911	0.01	−17.19 (37.74)	3.76 (28.80)	−4.96 (26.10)	1.26 (2, 33)	.298	0.08
Unimanual Motor	(nonparetic)	RP Shift	3.12 (26.25)	6.68 (21.56)	7.49 (27.14)	0.11 (2, 37)	.894	0.01	−10.61 (29.32)	−11.66 (29.80)	−9.55 (31.88)	0.02 (2, 39)	.984	0.00
Unimanual Ref.	(paretic)	Harm	−0.02 (0.08)	0.05 (0.06)**	0.06 (0.08)*	5.32 (2, 44)	.009	0.20	0.00 (0.04)	−0.01 (0.04)	−0.07 (0.11)	3.88 (2, 41)	.029	0.17
Unimanual Ref.	(paretic)	Ampl	−0.17 (7.33)	10.42 (11.87)**	4.03 (6.33)*	5.70 (2, 44)	.006	0.21	−0.52 (6.01)	0.21 (6.14)	1.38 (7.62)	0.28 (2, 41)	.755	0.01

Abbreviations: mCIMT, modified constraint-induced movement therapy; mBATRAC, modified bilateral arm training with rhythmic auditory cueing; DMCT, dose-matched control treatment; SD, standard deviation; df, degrees of freedom; f, effect size; CC, cross-correlation coefficient; Harm, movement harmonicity; Ampl, amplitude; RP, relative phase.

Data are mean change scores (standard deviation); Significant between-subject P values are in boldface.

P < .05 for within-subject changes (ie, change scores different from 0); **P < .01 for within-subject changes.

Figure 3.

Movement harmonicity and amplitude.

Postintervention to Follow-Up

From postintervention to follow-up there were significant differences between groups related to CC in the Kinesthetic Tracking task (antiphase pattern with an active nonparetic hand), amplitude and movement harmonicity in the Bimanual Coordination task (antiphase), and movement harmonicity in the Unimanual Reference task (see Table 2). However, post hoc analyses revealed significant differences only for movement harmonicity in the antiphase Bimanual Coordination task (see also Figure 3): an increase in movement harmonicity was observed for the mCIMT group, while a decrease was observed for the mBATRAC group (P = .008) and DMCT group (P = .014).

Coordination Patterns

Table 3 summarizes the results of the comparisons of bimanual coupling for the 2 coordination patterns (in-phase and antiphase) in the Bimanual Coordination and Kinesthetic Tracking tasks. Bimanual coupling was significantly stronger during in-phase coordination than during antiphase coordination for the Bimanual Coordination task in all tests. At baseline and postintervention similar results were found for the Kinesthetic Tracking task with the paretic hand (but not the nonparetic hand) as active hand. For this task there was a significant interaction between Pattern and Intervention group at follow-up. Post hoc analysis revealed a stronger coupling in the in-phase pattern than in the antiphase pattern at follow-up for the DMCT group only (P = .008). For the nonparetic hand as active hand in the Kinesthetic Tracking task, there was stronger coupling in the in-phase pattern than in the antiphase pattern postintervention only.

Table 3.

Bilateral Coupling in Antiphase and In-Phase Patterns for the 3 Intervention Groups^a.

			Correlation Coefficient (CC)
			Antiphase			In-Phase			Pattern			Group			Pattern * Group
Task	Test	Active Hand	mCIMT	mBATRAC	DMCT	mCIMT	mBATRAC	DMCT	F (df)	P	f	F (df)	P	f	F (df)	P	f
Bimanual Coord.	Baseline	Both	0.73 (0.22)	0.81 (0.16)	0.81 (0.21)	0.88 (0.11)	0.88 (0.12)	0.89 (0.11)	18.86 (1, 38)	.000	0.33	0.42 (2, 38)	.659	0.02	1.48 (2, 38)	.239	0.07
Bimanual Coord.	Post-Int.	Both	0.83 (0.18)	0.88 (0.19)	0.83 (0.19)	0.91 (0.08)	0.97 (0.02)	0.91 (0.10)	13.55 (1, 39)	.001	0.26	0.99 (2, 39)	.381	0.05	0.00 (2, 39)	.966	0.00
	Follow-Up	Both	0.89 (0.08)	0.89 (0.14)	0.78 (0.24)	0.90 (0.13)	0.97 (0.02)	0.89 (0.12)	8.14 (1, 32)	.008	0.20	1.71 (2, 32)	.197	0.10	1.48 (2, 32)	.243	0.09
Kinest. Tracking	Baseline	Paretic	0.82 (0.15)	0.80 (0.15)	0.67 (0.23)	0.91 (0.06)	0.87 (0.09)	0.84 (0.16)	18.76 (1, 36)	.000	0.34	2.76 (2, 36)	.077	0.13	1.40 (2, 36)	.260	0.07
	Post-Int.	Paretic	0.82 (0.14)	0.86 (0.12)	0.74 (0.23)	0.90 (0.08)	0.92 (0.08)	0.91 (0.03)	14.32 (1, 35)	.001	0.29	1.40 (2, 35)	.261	0.07	1.35 (2, 35)	.272	0.07
	Follow-Up	Paretic	0.82 (0.16)	0.92 (0.07)	0.72 (0.21)	0.82 (0.22)	0.90 (0.15)	0.95 (0.03)	3.17 (1, 30)	.085	0.10	1.48 (2, 30)	.245	0.09	3.78 (2, 30)	.034	0.20
	Baseline	Nonparetic	0.85 (0.17)	0.84 (0.12)	0.81 (0.21)	0.92 (0.09)	0.91 (0.08)	0.74 (0.52)	0.16 (1, 37)	.691	0.00	1.31 (2, 37)	.283	0.07	0.79 (2, 37)	.462	0.04
	Post-Int.	Nonparetic	0.80 (0.23)	0.80 (0.20)	0.92 (0.05)	0.89 (0.10)	0.87 (0.16)	0.94 (0.03)	4.49 (1, 34)	.042	0.12	1.52 (2, 34)	.232	0.08	0.50 (2, 34)	.614	0.03
	Follow-Up	Nonparetic	0.84 (0.19)	0.89 (0.14)	0.81 (0.18)	0.87 (0.12)	0.88 (0.20)	0.90 (0.06)	2.11 (1, 32)	.156	0.06	0.21 (2, 32)	.816	0.01	0.99 (2, 32)	.382	0.06

Data are absolute mean (standard deviation). Significant P-values are in boldface.

Discussion

The present study yielded some unexpected results. Based on the premise that in bilateral training neural coupling influences between both upper limbs are exploited, we expected the degree of intended coupling between both hands to be higher after bilateral than after unilateral training and control treatment. Our results did not provide support for this hypothesis. From baseline to postintervention, only 2 significant increases in intended coupling strength were observed: for the mBATRAC and DMCT groups in the in-phase Bimanual Coordination task (cf. asterisks in Table 2). However, the ANOVA on these change scores did not yield a significant difference between the 3 intervention groups. At follow-up the DMCT group showed significantly stronger intended coupling in the in-phase Kinesthetic Tracking task than at postintervention, but again, the change scores did not differ significantly between the 3 intervention groups. The results also showed no between-group differences related to unintended coupling (as obtained for the Unimanual Motor task), indicating that the degree of unintended entrainment was not differentially affected by the type of intervention.

Similar to the functional outcomes,¹² the present kinematic analysis of bimanual coupling did not yield systematic significant differences between groups. As stated in the Introduction, improved functionality of the paretic upper limb after bilateral training may not necessarily result from exploitation of the neurally driven coupling effects between the upper limbs, but simply from active training with the paretic upper limb.

The absence of significant differential effects on coupling strength between groups may be due to the large between-subject variability. In the present study, coupling strength varied widely across patients in all 3 groups. Similarly, in previous studies investigating interlimb coupling in stroke patients,^17,46 the variance in coupling strength across participants has been proposed as an important factor that impedes the detection of differential effects between groups. The currently applied measures of coupling strength may prove to be useful for tracking individual recovery of interlimb coordination. In this context, it is worthwhile to note that in the current examination of stroke patients, these measures were sufficiently sensitive to discern differences in coupling strength between the in-phase and antiphase patterns for every test in the Bimanual Coordination task and, to a smaller extent, in the Kinesthetic Tracking task, as typically observed in healthy subjects.^{19,35,40-44,47}

The significant differences at postintervention for changes in movement harmonicity and amplitude were in favor of the mBATRAC group. These results suggest a larger improvement of control over the paretic hand in both coupled and uncoupled situations after bilateral training. At first sight these improvements in the paretic hand’s movements after mBATRAC seem to indicate a beneficial influence of bimanual neural coupling. However, a more likely explanation may be found in the fact that mBATRAC involved rhythmic hand movements that were very similar to those required in the various tests, suggesting that the improvements were merely due to practicing rhythmic movements with the paretic hand. Note that these significant improvements in movement harmonicity and amplitude did not translate into better outcomes on clinical measures in favor of bilateral training (as presented in van Delden et al¹²). Hence, a better comprehension of the relevance of improvement in amplitude and movement harmonicity for functional recovery is desirable.

A limitation of the present study was the fact that not all patients could be included in every test, in one case amounting to only 9 patients in one of the intervention groups. Nevertheless, the average number of patients per group, per task, and per test was more than 14, sufficient for a meaningful analysis. Note that the number of patients that were able to perform a given task was comparable over groups, as there were no significant differences in group size at any test for any task.

In conclusion, contrary to our hypothesis, the degree of coupling between the hands was not significantly higher after bilateral than after unilateral training and control treatment. Although the observed improvements in the paretic hand’s movements after mBATRAC may suggest a beneficial influence of bimanual coupling, these improvements were more likely due to the type of movements employed during this training protocol. Our current analysis provided a first pass at determining potential training-induced changes in bilateral coupling, based solely on kinematic analyses. Future research on this topic should involve the study of brain dynamics in bilateral (and unilateral) tasks, using noninvasive neuroimaging techniques such as functional magnetic resonance imaging, electroencephalogram, or magneto-encephalography, to obtain more insight into bilateral coupling and its dependency on type of intervention.

Footnotes

Acknowledgements

The authors would like to thank the staff members of rehabilitation center Reade in Amsterdam for their cooperation.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the Dutch Scientific College of Physiotherapy (WCF) of the Royal Dutch Society for Physical Therapy (KNGF) and MOVE Research Institute Amsterdam.

References

Kilbreath

Heard

RC.

Frequency of hand use in healthy older persons. Aust J Physiother. 2005;51:119-122.

McCombe Waller

Whitall

. Bilateral arm training: why and who benefits? NeuroRehabilitation. 2008;23:29-41.

Whitall

McCombe Waller

Silver

Macko

RF.

Repetitive bilateral arm training with rhythmic auditory cueing improves motor function in chronic hemiparetic stroke. Stroke. 2000;31:2390-2395.

Whitall

Waller

Sorkin

. Bilateral and unilateral arm training improve motor function through differing neuroplastic mechanisms: a single blinded randomized controlled trial. Neurorehabil Neural Repair. 2011;25:118-129.

Luft

McCombe Waller

Whitall

. Repetitive bilateral arm training and motor cortex activation in chronic stroke. JAMA. 2004;292:1853-1861.

Carson

RG.

Neural pathways mediating bilateral interactions between the upper limbs. Brain Res Rev. 2005;49:641-662.

Cauraugh

Summers

JJ.

Neural plasticity and bilateral movements: a rehabilitation approach for chronic stroke. Prog Neurobiol. 2005;75:309-320.

Goble

DJ.

The potential for utilizing inter-limb coupling in the rehabilitation of upper limb motor disability due to unilateral brain injury. Disabil Rehabil. 2006;28:1103-1108.

Wolf

Winstein

Miller

. Effect of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: the EXCITE randomized clinical trial. JAMA. 2006;296:2095-2104.

10.

Brunner

Skouen

Strand

LI.

Is modified constraint-induced movement therapy more effective than bimanual training in improving arm motor function in the subacute phase post-stroke? A randomized controlled trial. Clin Rehabil. 2012;26:1078-1086.

11.

van Delden

AEQ

Peper

Beek

Kwakkel

Unilateral versus bilateral upper limb exercise therapy after stroke: a systematic review. J Rehabil Med. 2012;44:106-117.

12.

van Delden

AEQ

Peper

Nienhuys

Zijp

Beek

Kwakkel

Unilateral versus bilateral upper limb training after stroke: the ULTRA-Stroke clinical trial. Stroke. 2013;44:2613-2616.

13.

de Boer

Peper

Beek

PJ.

Learning a new bimanual coordination pattern: interlimb interactions, attentional focus, and transfer. J Mot Behav. 2013;45:65-77.

14.

Byblow

Lewis

Stinear

Austin

Lynch

The subdominant hand increases in the efficacy of voluntary alterations in bimanual coordination. Exp Brain Res. 2000;131:366-374.

15.

Carson

Riek

Smethurst

Párraga

Byblow

WD.

Neuromuscular-skeletal constraints upon the dynamics of unimanual and bimanual coordination. Exp Brain Res. 2000;131:196-214.

16.

Cohen

Synchronous bimanual movements performed by homologous and non-homologous muscles. Percept Motor Skills. 1971;32:639-644.

17.

Kelso

JAS

Southard

Goodman

. On the coordination of two-handed movements. J Exp Psychol Hum Percept Perform. 1979;5:229-238.

18.

Peper

Beek

van Wieringen

PC.

Coupling strength in tapping a 2/3 polyrhythm. Hum Mov Sci. 1995;14:217-245.

19.

Ridderikhoff

Peper

Beek

PJ.

Unraveling interlimb interactions underlying bimanual coordination. J Neurophysiol. 2005;94:311-325.

20.

Schmidt

Shaw

Turvey

MT.

Coupling dynamics in interlimb coordination. J Exp Psychol Hum Percept Perform. 1993;19:397-415.

21.

Lewis

Byblow

WD.

Bimanual coordination dynamics in poststroke hemiparetics. J Mot Behav. 2004;36:174-188.

22.

van Delden

AEQ

Peper

Harlaar

. Comparing unilateral and bilateral upper limb training: the ULTRA-stroke program design. BMC Neurol. 2009;9:57.

23.

Torre

Hammami

Metrot

. Somatosensory-related limitations of bimanual coordination after stroke. Neurorehabil Neural Repair. 2013;27:507-515.

24.

Post

Peper

Daffertshofer

Beek

PJ.

Relative phase dynamics in perturbed interlimb coordination: stability and stochasticity. Biol Cybern. 2000;83:443-459.

25.

Taub

Miller

Novack

. Technique to improve chronic motor deficit after stroke. Arch Phys Med Rehabil. 1993;74:347-354.

26.

Fritz

Light

Patterson

Behrman

Davis

SB.

Active finger extension predicts outcomes after constraint-induced movement therapy for individuals with hemiparesis after stroke. Stroke. 2005;36:1172-1177.

27.

Nijland

RHM

van Wegen

EEH

Harmeling-van der Wel

Kwakkel

Presence of finger extension and shoulder abduction within 72 hours after stroke predicts functional recovery: early prediction of functional outcome after stroke: the EPOS cohort study. Stroke. 2010;41:745-750.

28.

Smania

Paolucci

Tinazzi

. Active finger extension: a simple movement predicting recovery of arm function in patients with acute stroke. Stroke. 2007;38:1088-1090.

29.

Stinear

Prediction of recovery of motor function after stroke. Lancet Neurol. 2010;9:1228-1232.

30.

Stinear

Barber

Smale

Coxon

Fleming

Byblow

WD.

Functional potential in chronic stroke patients depends on corticospinal tract integrity. Brain. 2007;130:170-180.

31.

van Peppen

RPS

Kwakkel

Harmeling-van der Wel

. KNGF-richtlijn beroerte. Nederlands Tijschrift voor Fysiotherapie Supplement. 2004;114:3-78.

32.

Cup

EHC

Steultjens

EMJ.

Ergotherapierichtlijn beroerte. Utrecht, Netherlands: Nederlandse Vereniging voor Ergotherapie; 2005.

33.

de Poel

Peper

Beek

PJ.

Handedness-related asymmetry in coupling strength in bimanual coordination: furthering theory and evidence. Acta Psychol. 2007;124:209-237.

34.

Haken

Kelso

JAS

Bunz

A theoretical model of phase transition in human hand movements. Biol Cybern. 1985;51:347-356.

35.

de Boer

Peper

Beek

PJ.

Frequency-induced changes in interlimb interactions: increasing manifestations of closed-loop control. Behav Brain Res. 2011;220:202-214.

36.

Ridderikhoff

Peper

Beek

PJ.

Bilateral phase entrainment by movement-elicited afference contributes equally to the stability of in-phase and antiphase coordination. Neurosci Lett. 2006;399:71-75.

37.

van Dokkum

Hauret

Mottet

Froger

Métrot

Laffront

. The contribution of kinematics in the assessment of upper limb motor recovery after stroke. Neurorehabil Neural Repair. 2014;28:4-12.

38.

van Kordelaar

van Wegen

Kwakkel

. Impact of time on quality of motor control of the paretic upper limb after stroke. Arch Phys Med Rehabil. 2014;95:338-344.

39.

Chuang

Lin

Chen

Tsay

PK.

Randomized trial of distributed constraint-induced therapy versus bilateral arm training for the rehabilitation of upper-limb motor control and function after stroke. Neurorehabil Neural Repair. 2011;25:130-139.

40.

Kelso

JAS

. Phase transitions and critical behavior in human bimanual coordination. Am J Physiol. 1984;246:R1000-R1004.

41.

Rice

Newel

KM.

Upper-extremity interlimb coupling in persons with left hemiplegia due to stroke. Arch Phys Med Rehabil. 2004;85:629-634.

42.

Alaerts

Levin

Swinnen

SP.

Whether feeling or seeing is more accurate depends on tracking direction within the perception-action cycle. Behav Brain Res. 2007;18:29-34.

43.

Ridderikhoff

Peper

Beek

PJ.

Error correction in bimanual coordination benefits from bilateral muscle activity: evidence from kinesthetic tracking. Exp Brain Res. 2007;181:31-48.

44.

Stinear

Byblow

WD.

Phase transitions and postural deviations during bimanual kinesthetic tracking. Exp Brain Res. 2001;137:467-477.

45.

Cohen

Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Erlbaum; 1988.

46.

Rose

Winstein

CJ.

Temporal coupling is more robust than spatial coupling: an investigation of interlimb coordination after stroke. J Mot Behav. 2013;45:313-324.

47.

Sternad

Amazeen

Turvey

MT.

Diffusive, synaptic, and synergetic coupling: an evaluation through in-phase and antiphase rhythmic movements. J Mot Behav. 1996;28:255-269.