Leg Muscle Activity During Whole-Body Vibration in Stroke
This was an experimental study, with subjects undergoing three different WBV conditions of no WBV, low-WBV-intensity protocol, and high-WBV-intensity protocol. In each condition, the subjects were asked to perform eight different exercises while leg muscle activity on both sides was measured using surface EMG. The sequence of WBV intensities used and exercises performed was randomized by drawing ballots using an opaque envelope to avoid order effect. For each subject, all measurement procedures were performed on the same day.
As no study has examined the EMG response during WBV in people with stroke, previous research investigating the EMG response during WBV in healthy adults was used to estimate the sample size needed for this study. In a study involving 15 healthy men, Roelants et al. obtained a large effect size (Cohen's d = 5–8) for various muscle groups when WBV (35 Hz) was applied. On the basis of ANOVA (three WBV conditions), assuming an effect size f = 0.6 (large), with an alpha of 0.05 and power of 0.8, a minimum of 30 subjects would be required.
Subjects were recruited from stroke self-help groups in the community via convenience sampling. The inclusion criteria were as follows: a diagnosis of a hemispheric stroke with onset ≥6 months (i.e., chronic stroke), community dwelling (i.e., noninstitutionalized), abbreviated Mental Test score ≥6, and having hemiparesis in the lower extremity, as indicated by a composite leg and foot motor score of 13 or lower according to the Chedoke–McMaster Stroke Assessment. The exclusion criteria were as follows: neurological conditions in addition to stroke, brainstem or cerebellar stroke, significant musculoskeletal conditions (e.g., recent fractures and amputations), substantial vestibular dysfunctions (e.g., vertigo), peripheral vascular disease, unable to maintain standing for 1 min with standby guarding assistance of one person, severe cardiovascular conditions (e.g., unstable angina, uncontrolled hypertension, and uncontrolled cardiac dysrhythmia), and pain conditions that affected performance in standing, walking, or other daily functional activities.
The study was approved by the Research Ethics Committee of the administrating institute before commencement. The experimental procedures were first fully explained to each subject before written informed consent was obtained. The study was conducted in accordance with the Declaration of Helsinki.
The basic demographic information (e.g., age, medical history, medications, etc.) was obtained from interviewing the subjects. To test spasticity on the paretic side, subjects were placed in a supine position and asked to relax. The researcher then moved the knee on the paretic side into flexion and extension alternately, and the resistance to passive motion was noted. The same test was done on the ankle joint on the paretic side. The Modified Ashworth Scale was used to indicate the severity of spasticity in each joint tested. A higher score is indicative of more severe spasticity (0 = normal muscle tone, 4 = tested part rigid).
The Jet-Vibe System (Danil SMC Co. Ltd., Seoul, Korea) was used to deliver the WBV stimulation. This device generates vertical vibrations and has an adjustable frequency range between 20 and 55 Hz with corresponding preset amplitudes.
The intensity of WBV, represented by the peak acceleration (apeak), was calculated by the following formula: apeak = (2πf)A, where A is the amplitude and f is the frequency. The apeak is usually represented as a unit of the gravitational constant (g = 9.81 m·s). The peak acceleration values generated by the device were validated by a triaxial accelerometer (Model 7523A5; Dytran Instruments Inc., Chatsworth, CA).
Each participant was subject to three different WBV conditions: (a) no WBV, (b) low-intensity WBV protocol (peak acceleration = 0.96g, frequency = 20 Hz, amplitude = 0.60 mm), and (c) high-intensity WBV protocol (peak acceleration = 1.61g, frequency = 30 Hz, amplitude = 0.44 mm) while performing different exercises. We chose these frequencies because WBV frequencies lower than 20 Hz may cause destructive resonance effects to the body. On the other hand, our pilot experiments showed that frequencies higher than 30 Hz caused discomfort and fatigue in some individuals. The higher peak acceleration values associated with higher frequencies may also be a potential hazard for people with compromised bone mass, such as chronic stroke survivors.
The subjects were required to perform eight different exercises while being exposed to the three WBV conditions as described in Table 1. These exercises were commonly used in previous WBV trials in different populations. Practice trials were given to ensure that the subjects were able to perform the exercises properly before actual data collection. The knee angle was measured by a manual goniometer (Baseline® HiRes™ plastic 360° ISOM Goniometer, Fabrication Enterprises, White Plains, NY) to indicate the desired knee flexion angle in standing (10°), semisquat (30°), and deep-squat (90°) exercises. All experimental procedures were monitored closely by the researcher throughout to ensure that the subjects were performing the exercises properly and consistently. For standardization, all subjects were encouraged to gently hold on to the handrail of the WBV device for balance only. To ensure safety, the researcher provided standby guarding assistance while the patient was standing on the vibration platform. The researcher was standing by the patient in a guarding position, using his hands to be ready to guard or guide the patient.
Surface EMG was used to measure activity of the VL and GS muscles in all test conditions. After proper skin preparation, the bipolar bar electrodes (Bagnoli EMG system; Delsys, Inc., Boston, MA) were placed on the muscle belly of GS and distal one third of VL muscles, according to the specifications of the Surface EMG for a Non-invasive Assessment of Muscles (SENIAM) project. A reference electrode was placed at the head of fibula. Insulated EMG cables were fastened to avoid movement artifacts.
For each WBV condition, subjects were asked to assume each of the eight postures (Table 1) for 10 s while VL and GS EMG activity was being recorded. A total of three trials were performed for each of the eight exercises in a given WBV condition, with a 1-min rest period in between trials. After all eight exercises were completed in the first WBV condition, the subjects were then asked to do the same eight exercises in the second and third WBV conditions. A 10-min rest period was given between each WBV condition. Only the EMG data obtained during the middle 6 s of each trial was extracted to obtain the EMG root mean squares (EMGrms), and the mean value of the three trials was used for subsequent analysis.
All EMG data collected were preamplified (×1000) and sampled at 1 kHz (Bagnoli-8; DelSys, Inc.) using a personal computer with LabView version 7 software (National Instruments Corp., Austin, TX). Data processing was performed using MyoResearch XP, Master Package version 1.06 (Noraxon USA, Inc., Scottsdale, AZ). The EMG data were filtered with 20- to 500-Hz band-pass Butterworth filter, and the Infinite Impulse Response (IIR) rejector was implemented to eliminate the associated harmonics at the frequencies of 20, 30, and 60 Hz. After filtering, bias was calculated and removed from each EMG signal, and then the data were rectified and the EMGrms calculated in 100-ms windows around every data point.
At the beginning of the session, the EMG activity of VL and GS during maximal voluntary isometric contraction (MVC) was first recorded. For measuring the EMG amplitude of VL during MVC of knee extension, each subject was comfortably seated, and the tested leg was fixed horizontally on a dynamometer (Cybex Norm Testing & Rehabilitation System, Stoughton, MA) with hip and knee stabilized at 90°. Subjects were then asked to perform isometric knee extension for 10 s. The same device was used to stabilize the hip and knee when measuring the EMG amplitude of GS during MVC of ankle plantarflexion. The foot was placed at 90° on a wedged platform, and the subjects were instructed to isometrically plantarflex the ankle against the wedge with maximal effort and sustain for 10 s. Subjects were provided with verbal encouragement to ensure a maximal effort during testing.
EMG root mean square values (EMGrms) were calculated during intervals of 0.5 s. For each muscle, the maximum EMGrms values from the three MVC trials were averaged to obtain the mean value, which was then used for normalization of the EMGrms value obtained in each WBV condition. Therefore, the EMG amplitude of each muscle obtained in all WBV conditions was expressed as a percentage of the EMG amplitude obtained during the MVC (%MVC). The reliability of the EMGrms data obtained from the three MVC trials was excellent, as demonstrated by the intraclass correlation coefficients (ICC3,1) (paretic VL = 0.99, paretic GS = 0.94, nonparetic VL = 0.99, nonparetic GS = 0.99).
Analysis was performed with IBM SPSS Statistics software (version 20.0; IBM, Armonk, NY). The level of significance was set at P ≤ 0.05. Two-way repeated-measures ANOVA (within-subject factors: intensity [no WBV vs low-intensity WBV vs high-intensity WBV] and exercises) was used to compare the normalized EMGrms data across the different conditions. When sphericity assumption was violated, the Greenhouse–Geisser epsilon adjustment was used. Contrast analysis using paired t-test with Bonferroni adjustment was performed if any overall significant results were obtained for the EMG data. To compare the influence of WBV on the paretic side versus the nonparetic side, the ratio of normalized EMGrms (%MVC) of the VL and GS on the paretic side to the corresponding muscles on the nonparetic side was computed. A ratio greater than 1 indicated that the paretic side achieved a higher %MVC than the nonparetic side. A second two-way repeated-measures ANOVA model (within-subject factors: WBV intensity and exercises) was then constructed, using the EMGrms ratio as the dependent variable. Effect size was denoted by partial eta-squared (partial η). Large, medium, and small effect sizes were represented by partial η values of 0.14, 0.06, and 0.01, respectively. To examine the potential effect of spasticity on the EMG data, Spearman's rho was used to examine the relationship between (a) paretic knee spasticity score and normalized EMGrms of paretic VL and (b) paretic ankle spasticity score and normalized EMGrms of paretic GS in each testing condition.
Methods
Study Design
This was an experimental study, with subjects undergoing three different WBV conditions of no WBV, low-WBV-intensity protocol, and high-WBV-intensity protocol. In each condition, the subjects were asked to perform eight different exercises while leg muscle activity on both sides was measured using surface EMG. The sequence of WBV intensities used and exercises performed was randomized by drawing ballots using an opaque envelope to avoid order effect. For each subject, all measurement procedures were performed on the same day.
Subjects and Sample Size Estimation
As no study has examined the EMG response during WBV in people with stroke, previous research investigating the EMG response during WBV in healthy adults was used to estimate the sample size needed for this study. In a study involving 15 healthy men, Roelants et al. obtained a large effect size (Cohen's d = 5–8) for various muscle groups when WBV (35 Hz) was applied. On the basis of ANOVA (three WBV conditions), assuming an effect size f = 0.6 (large), with an alpha of 0.05 and power of 0.8, a minimum of 30 subjects would be required.
Subjects were recruited from stroke self-help groups in the community via convenience sampling. The inclusion criteria were as follows: a diagnosis of a hemispheric stroke with onset ≥6 months (i.e., chronic stroke), community dwelling (i.e., noninstitutionalized), abbreviated Mental Test score ≥6, and having hemiparesis in the lower extremity, as indicated by a composite leg and foot motor score of 13 or lower according to the Chedoke–McMaster Stroke Assessment. The exclusion criteria were as follows: neurological conditions in addition to stroke, brainstem or cerebellar stroke, significant musculoskeletal conditions (e.g., recent fractures and amputations), substantial vestibular dysfunctions (e.g., vertigo), peripheral vascular disease, unable to maintain standing for 1 min with standby guarding assistance of one person, severe cardiovascular conditions (e.g., unstable angina, uncontrolled hypertension, and uncontrolled cardiac dysrhythmia), and pain conditions that affected performance in standing, walking, or other daily functional activities.
The study was approved by the Research Ethics Committee of the administrating institute before commencement. The experimental procedures were first fully explained to each subject before written informed consent was obtained. The study was conducted in accordance with the Declaration of Helsinki.
Basic Demographics and Spasticity
The basic demographic information (e.g., age, medical history, medications, etc.) was obtained from interviewing the subjects. To test spasticity on the paretic side, subjects were placed in a supine position and asked to relax. The researcher then moved the knee on the paretic side into flexion and extension alternately, and the resistance to passive motion was noted. The same test was done on the ankle joint on the paretic side. The Modified Ashworth Scale was used to indicate the severity of spasticity in each joint tested. A higher score is indicative of more severe spasticity (0 = normal muscle tone, 4 = tested part rigid).
WBV Protocol
The Jet-Vibe System (Danil SMC Co. Ltd., Seoul, Korea) was used to deliver the WBV stimulation. This device generates vertical vibrations and has an adjustable frequency range between 20 and 55 Hz with corresponding preset amplitudes.
The intensity of WBV, represented by the peak acceleration (apeak), was calculated by the following formula: apeak = (2πf)A, where A is the amplitude and f is the frequency. The apeak is usually represented as a unit of the gravitational constant (g = 9.81 m·s). The peak acceleration values generated by the device were validated by a triaxial accelerometer (Model 7523A5; Dytran Instruments Inc., Chatsworth, CA).
Each participant was subject to three different WBV conditions: (a) no WBV, (b) low-intensity WBV protocol (peak acceleration = 0.96g, frequency = 20 Hz, amplitude = 0.60 mm), and (c) high-intensity WBV protocol (peak acceleration = 1.61g, frequency = 30 Hz, amplitude = 0.44 mm) while performing different exercises. We chose these frequencies because WBV frequencies lower than 20 Hz may cause destructive resonance effects to the body. On the other hand, our pilot experiments showed that frequencies higher than 30 Hz caused discomfort and fatigue in some individuals. The higher peak acceleration values associated with higher frequencies may also be a potential hazard for people with compromised bone mass, such as chronic stroke survivors.
Exercise Protocol
The subjects were required to perform eight different exercises while being exposed to the three WBV conditions as described in Table 1. These exercises were commonly used in previous WBV trials in different populations. Practice trials were given to ensure that the subjects were able to perform the exercises properly before actual data collection. The knee angle was measured by a manual goniometer (Baseline® HiRes™ plastic 360° ISOM Goniometer, Fabrication Enterprises, White Plains, NY) to indicate the desired knee flexion angle in standing (10°), semisquat (30°), and deep-squat (90°) exercises. All experimental procedures were monitored closely by the researcher throughout to ensure that the subjects were performing the exercises properly and consistently. For standardization, all subjects were encouraged to gently hold on to the handrail of the WBV device for balance only. To ensure safety, the researcher provided standby guarding assistance while the patient was standing on the vibration platform. The researcher was standing by the patient in a guarding position, using his hands to be ready to guard or guide the patient.
Measurement of Leg Muscle Activity Responses
Surface EMG was used to measure activity of the VL and GS muscles in all test conditions. After proper skin preparation, the bipolar bar electrodes (Bagnoli EMG system; Delsys, Inc., Boston, MA) were placed on the muscle belly of GS and distal one third of VL muscles, according to the specifications of the Surface EMG for a Non-invasive Assessment of Muscles (SENIAM) project. A reference electrode was placed at the head of fibula. Insulated EMG cables were fastened to avoid movement artifacts.
For each WBV condition, subjects were asked to assume each of the eight postures (Table 1) for 10 s while VL and GS EMG activity was being recorded. A total of three trials were performed for each of the eight exercises in a given WBV condition, with a 1-min rest period in between trials. After all eight exercises were completed in the first WBV condition, the subjects were then asked to do the same eight exercises in the second and third WBV conditions. A 10-min rest period was given between each WBV condition. Only the EMG data obtained during the middle 6 s of each trial was extracted to obtain the EMG root mean squares (EMGrms), and the mean value of the three trials was used for subsequent analysis.
All EMG data collected were preamplified (×1000) and sampled at 1 kHz (Bagnoli-8; DelSys, Inc.) using a personal computer with LabView version 7 software (National Instruments Corp., Austin, TX). Data processing was performed using MyoResearch XP, Master Package version 1.06 (Noraxon USA, Inc., Scottsdale, AZ). The EMG data were filtered with 20- to 500-Hz band-pass Butterworth filter, and the Infinite Impulse Response (IIR) rejector was implemented to eliminate the associated harmonics at the frequencies of 20, 30, and 60 Hz. After filtering, bias was calculated and removed from each EMG signal, and then the data were rectified and the EMGrms calculated in 100-ms windows around every data point.
At the beginning of the session, the EMG activity of VL and GS during maximal voluntary isometric contraction (MVC) was first recorded. For measuring the EMG amplitude of VL during MVC of knee extension, each subject was comfortably seated, and the tested leg was fixed horizontally on a dynamometer (Cybex Norm Testing & Rehabilitation System, Stoughton, MA) with hip and knee stabilized at 90°. Subjects were then asked to perform isometric knee extension for 10 s. The same device was used to stabilize the hip and knee when measuring the EMG amplitude of GS during MVC of ankle plantarflexion. The foot was placed at 90° on a wedged platform, and the subjects were instructed to isometrically plantarflex the ankle against the wedge with maximal effort and sustain for 10 s. Subjects were provided with verbal encouragement to ensure a maximal effort during testing.
EMG root mean square values (EMGrms) were calculated during intervals of 0.5 s. For each muscle, the maximum EMGrms values from the three MVC trials were averaged to obtain the mean value, which was then used for normalization of the EMGrms value obtained in each WBV condition. Therefore, the EMG amplitude of each muscle obtained in all WBV conditions was expressed as a percentage of the EMG amplitude obtained during the MVC (%MVC). The reliability of the EMGrms data obtained from the three MVC trials was excellent, as demonstrated by the intraclass correlation coefficients (ICC3,1) (paretic VL = 0.99, paretic GS = 0.94, nonparetic VL = 0.99, nonparetic GS = 0.99).
Statistical Analysis
Analysis was performed with IBM SPSS Statistics software (version 20.0; IBM, Armonk, NY). The level of significance was set at P ≤ 0.05. Two-way repeated-measures ANOVA (within-subject factors: intensity [no WBV vs low-intensity WBV vs high-intensity WBV] and exercises) was used to compare the normalized EMGrms data across the different conditions. When sphericity assumption was violated, the Greenhouse–Geisser epsilon adjustment was used. Contrast analysis using paired t-test with Bonferroni adjustment was performed if any overall significant results were obtained for the EMG data. To compare the influence of WBV on the paretic side versus the nonparetic side, the ratio of normalized EMGrms (%MVC) of the VL and GS on the paretic side to the corresponding muscles on the nonparetic side was computed. A ratio greater than 1 indicated that the paretic side achieved a higher %MVC than the nonparetic side. A second two-way repeated-measures ANOVA model (within-subject factors: WBV intensity and exercises) was then constructed, using the EMGrms ratio as the dependent variable. Effect size was denoted by partial eta-squared (partial η). Large, medium, and small effect sizes were represented by partial η values of 0.14, 0.06, and 0.01, respectively. To examine the potential effect of spasticity on the EMG data, Spearman's rho was used to examine the relationship between (a) paretic knee spasticity score and normalized EMGrms of paretic VL and (b) paretic ankle spasticity score and normalized EMGrms of paretic GS in each testing condition.