Abstract
Background and Purpose. The effects of different durations of rehabilitation sessions for the upper extremities (UEs) and lower extremities (LEs) on the recovery of interlimb coordination in hemiplegic gait in patients who have had a stroke were investigated. Subjects and Methods. Fifty-three subjects who had strokes involving their middle cerebral arteries were assigned to rehabilitation programs with (1) an emphasis on the LEs, (2) an emphasis on the paretic UE, or (3) a condition in which the paretic arm (UE) and leg (LE) were immobilized with an inflatable pressure splint (control treatment). The 3 treatment regimens were applied for 30 minutes, 5 days a week, during the first 20 weeks after onset of stroke. All subjects also participated in a rehabilitation program 5 days a week that consisted of 15 minutes of UE exercises and 15 minutes of LE exercises in addition to a weekly 1½-hour session of training in activities of daily living. A repeated-measures design was used. Differences among the 3 treatment regimens were evaluated in terms of comfortable and maximal walking speeds. In addition, mean continuous relative phase (CRP) between paretic arm and leg (PAL) movements and nonparetic arm and leg (NAL) movements and standard deviations of CRP of both limb pairs as a measurement of stability (variability) were evaluated. Results. Comfortable walking speed improved in the group that received interventions involving the LEs compared with the group that received interventions involving the UEs and the group that received the control treatment. No differences among the 3 treatment conditions were found for the mean CRP of NAL and PAL as well as the standard deviation of CRP of both limb pairs. Discussion and Conclusion. With the exception of an improved comfortable walking speed as a result of a longer duration of rehabilitation sessions, no differential effects of duration of rehabilitation sessions for the LEs and UEs on the variable we measured related to hemiplegic gait were found. Increasing walking speed, however, resulted in a larger mean CRP for both limb pairs, with increased stability and asymmetry of walking, indicating that walking speed influences interlimb coordination in hemiplegic gait.
- Cerebrovascular disorders
- Dynamic Pattern Theory
- Gait
- Hemiplegia
- Physical therapy
- Walking speed
Recently, a randomized controlled trial on a relatively homogeneous group of patients with stroke with respect to neurological diagnosis and disability at onset showed that a greater duration of rehabilitation for the lower extremities (LEs) during the first 20 weeks poststroke led to improved recovery in terms of activities of daily living (ADL), walking ability, and postural control.1 This treatment regimen was compared with a control condition in which the paretic arm (UE) and paretic leg (LE) were immobilized by means of an inflatable pressure splint.1 In addition, the researchers found that greater durations of rehabilitation for the LEs increased the patients' comfortable and maximal walking speeds. Whether these changes in walking speed coincide with improvements in the interlimb coordination of walking is not known.
The responsiveness of walking speed to change in functional outcome in the assessment of hemiplegic gait has been illustrated in several other intervention studies showing favorable effects of gait training on walking speed, even when other measures failed to detect these improvements in hemiplegic gait.2–5 Recently, Goldie et al6 suggested that the ability to detect differential effects in stroke rehabilitation may be due to the responsiveness of walking speed to change in functional outcome. Several researchers7,8 have recommended the use of walking speed as an independent variable in the evaluation of healthy and pathological gait.
In addition, walking speed in patients with hemiplegia has been found to be related to some aspects of motor function such as stage of synergistic patterned movements in the paretic leg9,10; muscle force of hip extensors, knee flexors, ankle dorsiflexors, and ankle plantar flexors of the paretic leg11; maximal ankle power12; standing balance13; use of walking aids14; number of falls15; and dependency in ADL.10,16 For example, a study of ambulatory patients with stroke demonstrated that “limited household walkers” (ie, those who needed assistance for some walking activities at home such as stair walking) had an average walking speed of 0.23 m/s (SD=0.17), “least-limited community walkers” (ie, those who were independent in stair walking and at least 2 moderate community activities) had an average walking speed of 0.58 m/s (SD=0.18), and “community walkers” (ie, those who were independent in all home and community walking activities) walked at speeds of 0.80 m/s (SD=0.18) or higher.16 Although improvement in walking speed seems to reflect a improvement in mobility, evaluating walking speed, in some people's view, does not reflect the quality of the gait,3 whereas achieving symmetry in coordination of the limb pairs of both sides of the body may be a better indicator.17–21
Correlations have been reported in the literature between walking speed and stride time, stance time, swing time, stride length, and step length.9,13,14,22–28 Recently, Olney et al29 showed, on the basis of principal component analysis, the importance of speed on hemiparetic gait by relating it to clusters of related kinematic and kinetic variables. They were able to account for 63.8% of the variance of the kinematic and kinetic variables of gait on the basis of speed (40.8%), asymmetry (12.8%), and postural flexion bias (10.2%). Some authors30,31 have concluded, therefore, that walking speed is an important factor in the coordination of walking. In Dynamic Patterns Theory it has been argued that systematically changing these nonspecific factors takes a system through its repertoire of states (ie, state of being or condition).30,31 If that is the case, such a variable is referred to as a “control parameter.”30(p832)
Wagenaar and Beek8 have shown that systematically varying walking speed as an independent parameter can induce a change in the coordination of pelvic and thoracic rotation (ie, trunk rotation) in people with and without strokes. In the subjects without strokes in their study, maximal pelvic rotation decreased when increasing walking speed from 0.25 to 1.0 m/s, whereas an increase in walking speed from 1.0 to 1.5 m/s resulted in an exponential increase of maximal pelvic rotation. The maximal thoracic rotation showed a small decrement with increasing walking speed. Almost linear relationships were observed between walking speed and the maximal amplitude trunk rotation as well as the phase difference (ie, timing) between pelvic and thoracic rotations. The phase difference changed from a 20-degree in-phase relationship to a 120-degree out-of-phase relationship when increasing walking speed.32,33 Patients who have had a stroke demonstrated similar relationships between walking speed and the above-mentioned dependent variables as compared with subjects without strokes. However, manipulating walking speed in the patients resulted in smaller pelvic and thoracic rotations and a larger trunk rotation. The possible relationship between the observed disorders in trunk rotations during hemiplegic gait and neurological symptoms were not investigated.
Wagenaar and Beek8 conducted their gait study from the perspective of nonlinear dynamics or Dynamic Pattern Theory. Following the latter approach, it is possible to investigate qualitative changes in the spatiotemporal organization of movement coordination by the interplay between control and order parameters of the system. Order parameters or collective variables represent the cooperativeness among body components on a more macroscopic level and can be expressed as the phase and frequency relationships among body segments.33(p57) The changes (or transitions) in order or pattern formation (ie, “flexibility”) can occur spontaneously as a result of unspecific, continuous changes of the control parameter (eg, walking speed). These changes from one stable coordination pattern to the other can be more or less abrupt and are often accompanied by an increase in the standard deviation of the relative phase (“critical fluctuations”) just before the so-called “transition” occurs. In previous studies on trunk and interlimb coordination, Wagenaar and colleagues found that the variability of the relative phase increased in the intermediate speed range (ie, 0.6–0.9 m/s) when systematically varying walking speed, suggesting the occurrence of these fluctuations between 2 patterns.34 The occurrence of abrupt or gradual transitions between coordination patterns is dependent on the stability of the patterns involved.28,30,31 Van Emmerik et al,35 for example, have demonstrated that patients with Parkinson disease not only have a reduced ability to make transitions (“flexibility”), but also have more hyperstable coordination as a result of rigidity.
In this study, the following 3 research questions were addressed:
Does a long duration of rehabilitation for the LEs and UEs influence the recovery of walking speed as well as the flexibility and stability of coordination patterns between arm and leg movements of both the hemiplegic and nonhemiplegic sides of the body during walking?
Does instruction to walk faster influence the flexibility and stability of coordination patterns between arm and leg movements of both the hemiplegic and nonhemiplegic sides of the body during functional recovery?
Is the severity of the disorder in the coordination of walking related to the severity of the paresis of the LE and UE?
Method
Patients
In 32 months, between September 1, 1994, and May 1, 1997, we recruited 101 patients with stroke from 7 hospitals.1 Stroke diagnosis was based on the World Health Organization's definition of stroke.36 To ensure weekly follow-up assessments, 3 rehabilitation centers and 15 nursing homes were selected in Amsterdam and Haarlem, the Netherlands, to participate in the present study. The study was coordinated by the Department of Physical Therapy of the University Hospital Vrije Universiteit, and the research protocol was approved by the ethics committee of each participating hospital.
The patients participating in the main intervention of the present study: (1) had a primary, first-ever stroke in the territory of the middle cerebral artery as revealed by computed axial tomography or magnetic resonance imaging scanning, (2) were 30 to 80 years of age, (3) had impaired LE and UE motor function as assessed with the Motricity Index (MI) (ie, scores of less than 100 points for each paretic limb),37 (4) were unable to walk without assistance on admission, (5) had no complicating medical history on the basis of review of medical records such as cardiac, pulmonary, or neurological disorders, (6) had no severe deficits in communication, memory, or understanding, and (7) gave written or verbal informed consent and were sufficiently motivated to participate.1
A speech therapist assessed the subjects' ability to communicate and accepted a cutoff point of 50th percentile corrected for age on the Dutch Foundation Aphasia Test.38 The Mini-Mental State Examination (MMSE) was used to assess orientation in time and place, and only subjects with a score of 24 points or more were included in the trial.39
Only patients who were able to walk 10 m without physical assistance within 10 weeks poststroke (N=53) were included in the study. The patients were not allowed to use any walking device, with the exception of an ankle-foot orthosis (AFO).
Within 24 hours after onset of stroke, the subjects were examined by a neurologist in order to confirm the diagnosis of stroke and to record clinical symptoms such as level of consciousness (assessed with the Glasgow Coma Scale).40 In addition, subjects were classified according to the Oxford Community Stroke Project classification41 as (1) those with total anterior circulation infarcts, (2) those with partial anterior circulation infarcts, or (3) those with lacunar anterior circulation infarcts. The Oxford Community Stroke Project classification not only has reliability between observers,41,42 but also has a high predictive validity with the side and size of the cerebral infarct when compared with computed axial tomography scanning.43,44 In order to control for heterogeneity of the sample, muscle force in the arm, balance, proprioception, and cognitive function were assessed according to the Orpinton Prognostic Scale.45
Design
Within the first 14 days poststroke, patients were randomly assigned to one of the 3 treatment conditions: (1) immobilization of the paretic LE and UE by means of an inflatable pressure splint,* which was applied for 30 minutes in a lying position 5 days a week (control treatment)46,47 (Fig. 1), (2) 30 minutes of LE training, or (3) 30 minutes of UE training.1 Randon assignment to groups took place within each hospital separately. Rehabilitation was individually applied by physical therapists and occupational therapists working at different institutions involved in the study, 5 days a week, for a period of 20 weeks poststroke. In addition, all 3 groups participated daily in a basic treatment program of 15 minutes of LE exercises and 15 minutes of UE exercises as well as a weekly 1½-hour session of ADL training administered by an occupational therapist.
Application of an inflatable pressure splint around upper and lower extremities.
Before random assignment to groups, the subjects and their families were informed that every additional type of intervention may improve outcome, and they were kept naive with respect to the type of intervention given. Randomization (permuted blocks of 9), with random number tables for every participating hospital, was applied. Concealed allocation was done by use of sealed envelopes. All measurements were carried out by an independent observer who had more than 15 years of experience in the use of these measurement instruments. Patients were assessed during the first 10 weeks on a weekly basis and biweekly from week 10 to week 20. The first kinematic assessment was carried out as soon as the subject walked 10 m without assistance. Subsequently, the first 6 consecutive kinematic measurements were used for analysis to look for differences in interlimb coordination among the 3 treatment groups. The final kinematic measurements took place at week 20 after onset of stroke. Wearing an AFO was allowed. The number of walking devices applied at the start of the study was not different among the 3 treatment groups (χ2=0.01, P=.94), indicating comparable walking ability among the 3 groups.
Nurses, speech therapists, and social workers provided customary care, depending on patients' needs, without having knowledge about treatment group assignment. With the exception of preventive medications such as antithrombotic and antihypertensive medications, no other medical interventions or therapies to improve skills were allowed during the first 20 weeks poststroke. From week 20 onward, type of treatment and its duration was determined by the physical therapists and occupational therapists involved, on average 3 times half an hour a week. With exception of kinematic measurements, final assessment took place at 26 weeks poststroke.
Treatment Conditions
The UE intervention was focused on the improvement of grasping, reaching, leaning, and dressing and hair combing, whereas the LE intervention was focused on the recovery of tasks such as turning over and maintaining sitting and standing balance. In addition, the LE intervention was designed to improve the symmetry in interlimb coordination during walking. The guidelines were based on evidence-based practice patterns derived from findings reported in 165 intervention studies in the field of stroke rehabilitation.48,49 We used what we believe is an eclectic approach based on research indicating that subjects' practice of motor skills needs to be both task and context specific.50 All participating therapists were instructed by the primary investigator (GK) during a specific course lasting 4 evenings on rehabilitation and recovery of LE and UE function.
In order to document duration of rehabilitation, the amount of therapy, as measured in 15-minute increments of face-to-face contact between subject and therapist, was recorded in a diary after each treatment session. In addition, content of therapy was reported daily, using 25 different codes representing task-specific goals for rehabilitation of the paretic LE and UE. In this way, not only the subjects' adherence to therapy but also the amount of therapy applied within one group were assessed.1 The organization of patient care was coordinated by 2 physical therapists.
Measurements
The first measurement of interlimb coordination was carried out as soon as the subject was able to walk 10 m independently without any support by a therapist (ie, level 3 of the Functional Ambulation Categories [FAC]).14,24 The FAC is an assessment comprising 6 categories designed to give detail on the physical support needed by patients and is reliable and valid.14,24 During measurements, the subjects were instructed not to use any walking device, with the exception of an AFO.
The coordination of LE and UE movements was studied while subjects walked at comfortable and maximal walking speeds.14,25 During every trial, the subjects were instructed to walk 10 m at comfortable and maximal walking speeds, and walking time was recorded from the “go” instruction to the moment the subjects crossed the 10-m line using a stopwatch. Between laps, subjects were was allowed to rest for about 1 minute. The mean of 3 repeated measurements was calculated in order to reduce measurement error. Speed was calculated for each trial by dividing the distance walked by walking time.1 High test-retest reliability coefficients were found for both comfortable and maximal walking speeds over 10 m (intraclass correlation coefficient=.97, P<.001, for both).
When the study began and following the first kinematic measurement, ADL were assessed with the Barthel Index (BI). The Dutch version of the BI yields reliable and valid measurements that represent a person's ability to perform 10 ADL tasks (ie, bladder control, bowel control, toilet use, dressing, feeding, ambulation, personal toilet, transfer activities, bathing, and stair climbing).51 Recovery of strength and synergism in the LEs and UEs was assessed by means of the MI37 and the Fugl-Meyer Sensorimotor Assessment (FM) (motor part).52,53 Both instruments are used to assess paresis in the LEs and UEs of people with stroke. In the present study, test-retest Spearman rank correlation coefficients for intraobserver reliability for the arm and leg components of the MI and the arm (hand and wrist included) and leg components of the FM (motor part) were .96 or higher (P<.001).
Material
Arm and leg swing in the sagittal plane during walking at comfortable and maximal walking speeds was recorded with 4 uniaxial accelerometers (Coulbourn type T-45†). The accelerometers were attached at the ventral part of the skin overlying the distal tibia of both legs and to the lateral part of the wrist of both arms. The accelerometer signals were amplified through a transducer-coupler (Coulbourn A-s72–25†). The acceleration time series were acquired with a computer (DAC-PAC‡), using a sampling frequency of 100 Hz. The software program “Poly”§ was used for data acquisition. Possible systematic differences in walking speed between walking with and without accelerometers were studied in 10 patients. No differences were found between the 2 conditions at both comfortable walking speed (F=0.07; df=1,9; P=.79) and maximal walking speed (F=0.80; df=1,9; P=.39). We do not know, however, whether similar values would be obtained with and without accelerometers.
Data Analysis
Interlimb coordination was evaluated on the basis of the raw accelerometer signals of both LEs and UEs. The continuous relative phase (CRP)32,33 was calculated between the following 2 limb pairs: (1) paretic arm and leg (PAL) and (2) nonparetic arm and leg (NAL). The signals of the 4 body segments were filtered with a low-pass Butterworth second-order frequency using a cutoff frequency of 5 Hz. Subsequently, the first derivative of the 4 filtered signals was obtained and normalized to the shortest stride period. After normalizing the maxima and minima from the filtered acceleration signals and the derivative of the acceleration signals to 1 and −1 in order to eliminate effects of amplitude, the movements of each body segment could be determined by a pair of phase variables (as, ds):
where as represents the acceleration signal for body segment s, ds represents the derivative of acceleration, rs represents the amplitude, ss represents the original signal, and t represents time. On the basis of these 2 phase variables, the phase angle was determined for each body segment by means of the following equation:
where φ represents the phase angle. The phase angles were calculated in the range 0 to 180 degrees. All stride cycles were normalized to the shortest stride period (ie, swing of the nonparetic LE in the sagittal plane), which allowed for superpositioning of stride cycles within one walking speed condition. The CRP between 2 segments was calculated by subtracting the phase angle of one segment from the phase angle of another segment for each point in a stride cycle. Subsequently, the mean and standard deviation of the relative phase over all corresponding data points within the different stride cycles at one walking speed condition were calculated. The standard deviation of the relative phase at one walking speed condition was used as a measure for the stability of the phase relation between limb pairs.
Differences in initial values among the 3 groups for relevant independent variables were tested with the Fisher exact test (ie, sex, hemisphere of stroke, and social support) or the chi-square test (type of stroke) for nominal data. The Kruskall-Wallis test was used for ordinal data (ie, BI, FAC, MMSE, Orpinton Prognostic Scale, MI-total and FM-total), and a one-way analysis of variance (ANOVA) was used for interval data (ie, age, walking speed, and start of therapy). The distribution of interval-scaled measurements was first tested for normality with the Kolmogorov-Smirnov test. A one-way ANOVA was used to evaluate the changes from the start of the study to final assessment in comfortable and maximal walking speeds among the 3 groups. When differences were found, a post hoc analysis was performed to test which groups differed from each, using a Student t test.
An ANOVA for repeated measurements was applied to evaluate differences among the 3 groups as well as within-group effects of the factors limb pairs (2 levels: hemiplegic side versus nonhemiplegic side), speed (2 levels: comfortable walking speed versus maximal walking speed), and time (6 levels: consecutive kinematic measurements as soon as subjects were able to walk independently) in terms of the mean CRP and standard deviation of the CRP. Finally, an analysis of covariance was applied to study the relationship between the covariates (MI and FM scores) and the dependent variables (mean and standard deviation of CRP of limbs on the paretic side). For all tests, a two-tailed significance level of .05 was chosen.
Results
Fifty-three patients with an initial severe stroke in the region of the middle cerebral artery were able to walk independently within 10 weeks after onset of stroke and were included in our study. Table 1 presents the initial (prestudy) characteristics of the 18 subjects assigned to group that received the control treatment (control group), the 18 subjects assigned to the group that received the UE intervention (UE group), and the 17 subjects assigned to the group that received the LE intervention (LE group). The mean interval between stroke onset and start of therapy was 7.9 days (SD=2.7, range=2–14), and the first gait assessment was conducted, on average, at 6.0 weeks (SD=3.4, range=2–10). The diaries revealed that during the 6 consecutive measurements (X̄=8.0 weeks, SD=2, range=6–10), the LE group received about 24.4 hours of rehabilitation focusing on the LEs, whereas the UE and control groups received about 8.9 and 9.2 hours of rehabilitation focusing on the LEs, respectively.1 No differences in subject characteristics were found among the 3 treatment groups at either the time of onset or the first gait assessment (Tab. 1). In addition, at the first gait measurement, no differences were found among the 3 groups for comfortable walking speed (F=0.03, df=2, P=.864) or maximal walking speed (F=0.54, df=2, P=.468). Finally, no differences were found in the number of walking devices applied during functional recovery (ie, 4 for the control group, 2 for the UE group, and 4 for the LE group) (χ2=1.85, P=.40).
Characteristics of Subjects Who Received Control Treatment (Control Group), Upper-Extremity Intervention (UE Group), and Lower-Extremity Intervention (LE Group)
Walking Speed
Mean comfortable walking speed of the 3 groups improved from 0.39 m/s (SD=0.25, range=0.07–0.71) at the time of the first kinematic assessment to 0.73 m/s (SD=0.35, range=0.17–1.18) at the final assessment, and mean maximal walking speed increased from 0.53 m/s (SD=0.34, range=0.08–1.08) to 0.96 m/s (SD=0.49, range=0.18–1.82). A difference among the 3 groups was found for comfortable walking speed (F=3.52, df=2, P=.037), whereas the improvement in maximal walking speed approached the level of significance (F=2.90, df=2, P=.064) during the 6 consecutive measurements. The average gain in comfortable walking speed for the LE group was 0.18 m/s when compared with the control group and 0.21 m/s when compared with the UE group, whereas the average gains in favor of LE at maximal walking speed were 0.21 and 0.22 m/s, respectively. A post hoc analysis revealed larger improvements in comfortable walking speed for the LE group compared with the control group (t=−2.408, df=33, P=.022) and the UE group (t=−2.144, df=33, P=.039), whereas no difference was found between the UE and control groups (t=−0.540, df=34, P=.467).
Mean CRP for NAL and PAL
Changes in mean CRP of PAL and NAL for the 3 treatment conditions as function of time and walking speed are summarized in Table 2 and are depicted in Figure 2. No main group effect or interaction effects between group and time, group and speed, or group and limb pair were found, indicating there were no differences among the 3 treatment conditions (Tab. 2).
Group means and standard deviations of mean continuous relative phase (CRP) difference between nonparetic arm and leg (NAL) and paretic arm and leg (PAL) presented for comfortable walking speed (A and B) and maximal walking speed (C and D). Dashed lines represent walking speed.
Main and Interaction Effects at Different Durations of Rehabilitation Sessions for Lower Extremities and Upper Extremities on the Phase Relationship Between Arm and Leg Movements During Walking
Main effects, however, were found for time (F=7.95; df=5,250; P<.001), walking speed (F=7.49; df=1,50; P=.009), and limb pair (F=26.06; df=1,50; P<.001) (Tab. 2). These findings indicate that (1) both NAL and PAL showed an increase in mean CRP as a function of time, (2) the instruction to walk at a maximal speed resulted in a larger mean CRP for NAL and PAL than the instruction to walk at a comfortable speed, and (3) the mean CRP of NAL was larger than the mean CRP of PAL at both comfortable and maximal walking speeds.
The interaction between limb pair and speed (F=4.75; df=1,50; P=.034) indicated that the differences in mean CRP for NAL and PAL were larger when walking at maximal speed than when walking at a comfortable speed. In addition, the differences found in mean CRP between NAL and PAL due to walking speed were more pronounced at the end of the 6 consecutive kinematic measurements than at the start (F=2.54; df=5,250; P=.029). No interactions were found between time and limb pair (F=1.49; df=5,250; P=.193) or speed and time (F=0.97; df=5,250; P=.556) (Tab. 2).
Stability of Phase Relationships
Changes in standard deviation of NAL and PAL for the 3 treatment conditions as a function of time and walking speed are summarized in Table 3 and are depicted in Figure 3. No main group effect or interaction effect between group and time, group and speed, or group and limb pair were found, indicating there were no differences among the 3 treatment conditions for stability of walking.
Group means and standard deviations of stability of mean continuous relative phase (CRP) difference between nonparetic arm and leg (NAL) and paretic arm and leg (PAL) presented for comfortable walking speed (A and B) and maximal walking speed (C and D). Dashed lines represent walking speed.
Main and Interaction Effects at Different Durations of Rehabilitation Sessions for Lower Extremities and Upper Extremities on the Stability Between Arm and Leg Movements During Walking
Main effects, however, were found for time (F=42.20; df=5,250; P<.001), walking speed (F=4.48; df=1,50; P=.039), and limb pair (F=32.19; df=1,50; P<.001) (Tab. 3). These findings indicated that (1) both NAL and PAL showed an increase in stability of mean CRP as a function of time, (2) the instruction to walk at maximal walking speed resulted in higher stability of NAL and PAL than the instruction to walk at a comfortable speed, and (3) the stability of NAL was larger than the stability of PAL at both comfortable and maximal walking speeds.
An interaction between limb pair and walking speed (F=8.06; df=1,50; P=.007) was found, indicating an increase in asymmetry in stability between NAL and PAL when increasing walking speed. The interaction among limb pair, speed, and time (F=2.31; df=5,250; P=.045) indicated that the difference in stability between NAL and PAL became larger when walking at maximal speed. No interaction was found between time and limb pair (F=2.10; df=5,250; P=.066) or speed and time (F=1.10; df=5,250; P=.362) (Tab. 3).
Paresis Versus Stability and Flexibility
Table 4 presents the findings with respect to the influence of LE and UE motor function assessed with the MI, FM score, and FAC on mean CRP and standard deviation of CRP of PAL. The recovery of mean CRP on the hemiplegic side of the body was related to muscle force (MI) of the paretic arm (F=8.61; df=6,306; P<.001) and leg (F=9.24; df=6,306; P<.001) as well as to stage of synergism (FM scores) of the paretic arm (F=5.94; df=6,306; P<.001) and leg (F=10.19; df=6,306; P<.001), balance score on the FM (F=4.17; df=6,306; P<.001), and FAC score (F=4.99; df=6,306; P<.001). Even a stronger association was found for the standard deviation of CRP on the hemiplegic side of the body related to muscle force (MI) of the paretic arm (F=13.23; df=6,306; P<.001) and leg (F=23.83; df=6,306; P<.001) as well as to stage of synergism (FM scores) of the paretic arm (F=8.67; df=6,306; P<.001) and leg (F=27.82; df=6,306; P<.001), balance score on the FM (F=16.44; df=6,306; P<.001), and FAC score (F=9.73; df=6,306; P<.001).
Effects of Upper-Extremity and Lower-Extremity Motor Function on Mean Continuous Relative Phase and Stability on the Paretic Sidea
Discussion
The main purpose of our study was to investigate the effects of duration of rehabilitation sessions for the LEs and UEs on the recovery of walking speed as well as the flexibility and stability of coordination patterns between LE and UE movements of both the hemiplegic and nonhemiplegic sides of the body.
Our findings show that longer durations of rehabilitation sessions for the LEs resulted in a small increases in comfortable walking speed compared with longer durations of rehabilitation sessions for the paretic UE and the control treatment. Improvement in maximal walking speed approached the level of significance. These effects, however, are limited to the period of intervention.1 The average gain in comfortable walking speed between the LE intervention compared with the control treatment and the UE intervention during the 6 consecutive measurements (X̄=8 weeks, SD=3.4) varied from 0.18 m/s to 0.21 m/s. The average gain in maximal walking speed varied from 0.21 m/s to 0.22 m/s. Although gait speed is strongly related to the functional ambulation status of patients with stroke,16,25 we believe the clinical relevance of our findings is open to question. Our findings provide more evidence for the existence of a dose-response relationship between hours of individually applied physical therapy and occupational therapy and functional improvement.54 The findings also show how walking speed can reflect change in other variables.2,3,6,55
The difference in maximal walking speed approached the level of significance, whereas a significant difference in efficacy was obtained for comfortable walking speed. This finding may be related to a difference in smallest detectable difference, that is, 0.16 m/s for comfortable walking speed and 0.18 for maximal walking speed.
In our study, no differences in outcome among UE rehabilitation, LE rehabilitation, and the control condition in mean relative phase were found for both limb pairs. A similar finding was obtained for the stability of the phase relationships for both limb pairs. This finding suggests that the symmetry in interlimb coordination between hemiplegic and nonhemiplegic sides was not influenced by the amount of additional LE rehabilitation. This may be caused by the variability in interlimb coordination between subjects as a result of walking at different speeds, as well as the relatively small change in walking speed among groups as a result of duration of treatment.
Our results indicate that both limb pairs showed considerable improvement in CRP and stability during recovery. The recovery of mean CRP and standard deviation of CRP for the hemiplegic side of the body correlated with muscle force (MI) and stage of synergism of the paretic LE and UE as well as to the FM balance score and the FAC ambulation score. The effects of increasing walking speed as well as the instruction to walk as fast as possible on CRP and standard deviation of CRP were both larger for the nonhemiplegic side of the body compared with the hemiplegic side of the body. The interaction of walking speed and limb pair suggests that the contribution of the nonhemiplegic side of the body by alternating arm and leg swing to increased walking speed is relatively larger than the contribution of the nonhemiplegic side of the body.
If the coordination between LE and UE on the hemiplegic side of the body is the main limiting constraint for restoring a normal walking speed, perhaps physical therapy intervention should allow asymmetry in the coordination of limb pairs when the goal of gait training is to increase comfortable and maximal walking speeds. The objective of some treatment approaches is to improve the symmetry in coordination of the limb pairs of both sides of the body.18–21,56
The findings of our study are in agreement with those of Hesse and colleagues,57,58 who found that despite improvement in outcome variables such as muscle force, maximal walking speed, and stair climbing, there was no improvement in gait symmetry. They studied 40 patients with hemiplegia after a 4-week inpatient rehabilitation program based on what they referred to as the Bobath concept.57,58 Exercises such as systematically varying step frequency by means of an external (auditory) rhythm may be more successful in improvement of interlimb coordination.33,59
Our findings show that the gradual recovery in comfortable walking speed (about 0.34 m/s) and maximal walking speed (about 0.43 m/s) was correlated with a gradual increase in relative phase between LE and UE movements for both the hemiplegic and nonhemiplegic sides of the body. In addition, the interlimb coordination became more stable. Therefore, the assumption that improvement in walking speed is associated with improvement in coordination of walking and vice versa2,21,29,31 is confirmed by our findings. In some clinical trials,2–4 improvements were found in walking speed but not for other dependent variables such as walking ability as measured by the FAC or BI. Our findings support the idea that walking speed can be used as an independent variable or control variable in the evaluation and treatment of gait disorders.7,8,31–35
The findings of our study related to walking speed and interlimb coordination are in agreement with the findings of other studies. For example, Wagenaar and van Emmerik33 reported effects of systematically varying speed on the relative phase between LE and UE movements during treadmill walking of patients with stroke. The phase relationship between LE and UE movements on the paretic side changed from a relatively unstable alternating phase relationship (about 80°) at 0.2 m/s to a relatively stable more out-of-phase pattern (about 140°) at 1.2 m/s.
Finally, the effects of walking speed on the variability of interlimb coordination are in agreement with studies on the kinematics of walking by Borowski et al60 and Lehmann et al,61 who reported a higher variability of spatiotemporal characteristics (ie, step-length, stance, swing, and double-support duration) and limb kinematics at low walking speeds compared with higher walking speeds.
A possible limitation of our study is that the gait of people with hemiplegia was studied in the sagittal plane only. The assumed pendular activity of LE and UE movements is only a reflection of the richness of coordination patterns in LE and UE movements during walking.33 Our results are restricted to people with stroke who were able to walk without assistance (ie, FAC score of 3 or higher). In addition, the time intervals between measurements varied from 1 to 2 weeks, which limited our time series analysis. Finally, long-lasting treatment effects were not shown in our study.
Conclusion
We conclude that with the exception of an improved comfortable walking speed, no effects of differences in duration of rehabilitation sessions for the LEs and UEs in hemiplegic gait were found. Increasing walking speed during recovery after stroke resulted, however, in a larger mean CRP for both limb pairs, with increased stability and asymmetry of walking, indicating that walking speed influences the swing pattern of the limbs in hemiplegic gait.
Footnotes
Both authors provided concept/research design and project management. Dr Kwakkel provided writing, data collection and analysis, subjects, and facilities/equipment. Dr Wagenaar provided fund procurement, institutional liaisons, and consultation (including review of manuscript before submission).
↵* Svend Andersen Plastic Industrials, Haarlev, Denmark.
↵† Coulbourn Instruments LCC, 7462 Penn Dr, Allentown, PA 18106.
↵‡ Keithley Instruments Inc, 28775 Aurora Rd, Cleveland, OH 44139.
↵§ Poly Software International, PO Box 1457, Sandy, UT 84091.
- Received August 20, 1999.
- Accepted November 12, 2001.
- Physical Therapy
References
Issue highlights
