Effects of Single or Multiple Sessions of Whole Body Vibration in Stroke: Is There Any Evidence to Support the Clinical Use in Rehabilitation?

Background and Purpose Recently new technologies and new techniques, such as Whole Body Vibration (WBV), have been introduced by the health and fitness industry to pursue therapeutic or physical performance goals. The aim of this systematic review is to investigate the effectiveness of single or multiple WBV sessions alone or in association with traditional rehabilitation, compared to traditional rehabilitation therapy or with sham therapy in poststroke patients. Methods Randomized Control Trials and controlled clinical trials written in English between January 1st, 2003, and December 31st, 2017, were selected from PubMed, Cochrane-Central-Register-of-Controlled-Trials, and Physiotherapy-Evidence-Database (PEDro). The single WBV session and multiple sessions' effects were assessed. Study characteristics, study population, intervention protocols, effects of WBV sessions, and adverse events were investigated with a descriptive analysis. Results The search reported 365 articles and after screening and removal of duplicates, 11 manuscripts with PEDro score≥6/10 were selected (391 poststroke patients). Study characteristics, study population, intervention protocols (frequencies, amplitude of vibration, and peak acceleration), effects of a single or multiple WBV sessions, and adverse events were analyzed. They have been investigated with particular attention to bone turnover, structure and muscle functions, spasticity, postural control and risk of falls, functional mobility, somatosensory threshold, and activity and participation. Comparing WBV group with control group no significant benefits emerged. Discussion This systematic review included studies involving participants with non homogeneous characteristics, just considering the incorporation of studies on individuals with chronic and postacute stroke. Despite these limits, WBV treatment has no significant risks for patients and shows interesting effects of WBV treatment in Structure and muscle functions, Spasticity and Postural control. Conclusions Even though treatment with WBV appears safe and feasible, there is insufficient evidence to support its clinical use in poststroke rehabilitation at this point. More studies assessing other functional tests and with more specific treatment protocols are needed.


Introduction
Recently new technologies and new techniques, such as Whole Body Vibration (WBV), have been introduced by the health and fitness industry to pursue therapeutic or physical performance goals. Basic neurophysiological studies have shown that vibration can alter sensory and motor function by mostly activating the primary spindle endings, although secondary spindle endings, such as Golgi tendon organs, Pacinian, and Meissner corpuscles can also be activated [1].
Several types of Whole Body Vibration platforms can be found in literature [2][3][4].
Currently, there are three commercial typologies of vibration platforms. The first one, Galileo5, has a teeterboard that produces asynchronous sinusoidal side-alternating vertical vibrations.
Key descriptors of vibration devices include the frequency (number of complete movement cycles per second, measured in hertz), the amplitude (displacement of oscillatory motion, measured in mm), the acceleration (measured in m/s 2 or g), and the duration (exposure time) of the vibration exposure [6]. The intensity of vibration is determined by varying both frequency and amplitude; accordingly it may be possible to get a training program tailored to the needs of the person, or to adapt it to different goals.
The vibration devices can differ with frequency ranges from 0 to 60 Hz, amplitudes from 0 to 12 millimeters, and peak acceleration from 0 to 20,1 g. In a typical session, the user stands on the device doing static or dynamic exercises while the platform produces sinusoidal oscillations. In most cases, the vibration session consists of several bouts of vibration exposure (each lasting from less than a minute to several minutes) separated by rest periods.
The growing interest in vibrations started from animal research in the 1990s and early 2000s when a correlation between vibration and bone deposition was reported [7,8].
Other studies demonstrated that WBV training causes a continuous proprioceptive stimulation which increases neuromuscular receptivity [9]. Many studies have highlighted the possibility of WBV training to improve sport performance, increasing range of motion, and to be a beneficial supplementary training technique in strength programs for athletes [10][11][12][13][14][15][16].
The aim of this systematic review is to investigate the effectiveness of single or multiple WBV sessions, alone or in association with traditional rehabilitation, compared to traditional rehabilitation therapy or with sham therapy in patients with a stroke.

Study Design and Eligibility
Criteria. This systematic review was conducted and reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA statement). We have used PICO method (Patients/Population, Intervention, Comparison, Outcomes) [31] and a qualitative analysis focused on the differences between the selected studies. We followed PICO variables: persons with stroke (P); WBV training (I); comparison between WBV therapy and the same exercises performed without WBV, comparison between WBV therapy and other physical activities or sham therapy (C); outcomes measuring body functions and structures, activities, and participation (O) as reported in International Classification of Functioning, Disability and Health (ICF) Stroke Brief [32]. We investigated the effects of WBV therapy on patients with ischemic or hemorrhagic stroke. Only Randomized Control Trials (RCT) and controlled clinical trials written in English were selected. The single WBV session and multiple sessions effects were assessed.
We excluded studies on animals; not about stroke; based on focal vibration treatments; with a PEDro score<6 [5,33,34], or where the full-text was not available in our institutional University Library System.
The search query, based on the PICO strategy, included both ischemic and hemorrhagic stroke. The string used for PubMed was launched in the first week of January 2018 and contained at least one of these terms: "Nervous System Disease", "Stroke", Whole Body Vibration", "Vibration", "vibration platform", "sham therapy", "rehabilitation therapy", "gait", "balance", "muscle performance", "spasticity", "bone turnover", "postural control", and "muscle strength".
Those keywords were used in several combinations with Boolean operators (AND/OR) and modified for other databases.

Levels of Evidence.
Study quality was assessed according to the guidelines of the Oxford Centre for Evidence-Based Medicine [38]; we have assigned a level of evidence 2 to all the studies included in this systematic review. To assess the methodological quality of the selected studies we used the PEDro scale [39], considering only high quality studies (score≥6). The results of methodological quality assessment are displayed in Table 1.

Data Extraction.
Articles were initially screened by title and abstract. Articles unclear from their title or abstract were reviewed according the selection criteria through full-text. Three authors (F. P., L. S., and R. G.) independently extracted data from the studies that met the inclusion criteria and they were blinded to each other's review. In case of disagreement, a fourth opinion (C. C.) could have been requested. Conference abstracts were evaluated but deemed not suitable because of the limited body of data related to the study.

Data Synthesis and Analysis.
We performed a descriptive analysis of the measures of WBV effects on each outcome selected. The heterogeneity of outcomes, participants, and intervention protocols made it impossible to draw up a metaanalysis. In the articles with significant outcomes we calculated the changes among the groups using the values of SES (Standardized-Effect-Size) concerned. The calculation was performed using the average values and standard deviations. The effect size was considered, according to Hedges [51], small (for values of SES = 0.2), medium (SES = 0.5), and large (SES = 0.8).
Rehabilitation Research and Practice 3 Table 1: Results of the quality assessment of the included studies.

Comparison ( studies)
The Control Group make exercise under the same condition as the Whole-Body Vibration Group, but without Whole-Body Vibration or with sham vibration The Control Group make other forms of exercise or physical activity Tihanyi et al, [40] Lau et al, [41] and Pang et al, [42] Brogårdh et al, [43] Chan et al, [44] Marín et al, [45] Liao et al, [46] Choi et al, [47] Choi W et al, [48] Van Nes et al, [49] Tankisheva et al, [50] eligibility criteria (it does not contribute to total PEDro score)   Figure 1 describes each step of our database research. Our initial search on PubMed produced 249 records, plus 90 records from Cochrane Library and 26 from PEDro Database. After removing 43 duplicates, an assessment was performed on headlines, abstracts, and full texts, which resulted in the removal of 304 records that left 18 eligible articles. Among the remaining 18 eligible articles 2 were not RCTs and 5 had PEDro scores<6/10. Therefore in this systematic review were included 11 articles [40][41][42][43][44][45][46][47][48][49][50] (10 studies). The two reports by Lau et al. [41] and Pang et al. [42] are based on identical data.

Study Characteristics.
To assess the methodological quality of the selected studies we used the PEDro scale [39], considering only high quality studies (score≥6) ( Table 1).

Intervention Protocol for WBV Group.
There are significant differences in the WBV protocols (  [46] investigated the effects of vibration intensity in poststroke patients. Two groups performed exercises on the same vibrating platform, with the same amplitude but with different frequencies and acceleration (respectively, 20 and 30Hz and 1.61 and 3.62g). Six studies used a vertical synchronous vibration [40,41,43,44,46,50] and four studies used an asynchronous vertical sinusoidal vibration transmitted alternately to the left and right side of the body [45,[47][48][49].
Five studies [40,[43][44][45]49] have provided only static exercises on WBV. The most common static exercise used was the semisquat with knee flexion at 30 ∘ and 60 ∘ while standing on the vibratory platform. Five other studies [41,42,[46][47][48]50] provided a set of static and dynamic exercises. In Marín et al. [45] the participants performed the exercises with WBV in addition to the daily conventional rehabilitation therapy. In Choi W et al. [48] participants performed the exercises with WBV combined with Treadmill Training.
In the Lau et al. [41] and Pang et al. [42] papers, participants completed 1.5 minutes of warm-up exercises in a sitting posture. Sessions in Choi W et al. [48] were preceded by 15 minutes of gentle stretching, while sessions were preceded by 10 minutes of warm-up and followed by 10 minutes of cooldown exercises in the Liao et al. [46] paper (Table 3).

Mean±SD presented unless indicated otherwise.
III Median (range).

IV
The results shown in this table refer to the difference between the WBV and comparison groups. The SES was calculated from the mean and standard deviation of the change scores unless indicated otherwise.

V
The SES was not reported because MAS is an ordinal variable.

VI
The SES for this particular outcome was reported in the text by the authors.

VII
The electromyographic amplitude data of individual muscles were not included because they were not normalized, making it difficult to compare groups.    performed conventional rehabilitation exercises with music or maintained habitual physical activity (Table 3).

Effects of a Single WBV Session. Tihanyi et al. 2007 [40]
and Chan et al. 2012 [44] (46 participants) investigated the immediate effects of a single WBV session. In Table 3 are summarized the outcome measures including significant findings about lower limb muscle strength, spasticity, postural control, and functional mobility.

Events during WBV Sessions.
A total of 211 participants were exposed to WBV. Six trials [41-43, 45, 46, 49, 50] reported slight to mild side effects, generally declining after the first therapeutic sessions. In Lau et al. [41], 5 of the 41 participants in the WBV group reported adverse symptoms potentially related to vibration: knee pain, fatigue, and dizziness. Brogårdh et al. [43] reported that 15 of the 31 participants, in both groups, reported a transient and mild muscle soreness or muscle fatigue. Tankisheva et al. [50] reported that some participants felt a tingling in the legs. Liao et al. [46] reported a participant with moderate knee pain after low-intensity WBV, 3 participants with fatigue after low-intensity WBV, and 2 after High-Intensity WBV. Two studies [45,49] have no side effects in all participants (38 persons) in the WBV group. In 3 studies [40,44,47] it is not clear whether any adverse events occurred.

Discussion
Our literature shows that WBV treatment presents no significant risks for patients, but in this review we cannot state an objective benefit in poststroke patients according to ICF (e.g., bone turnover, motor functions, balance, mobility, somatosensory threshold, risk of falls, and activities of daily life and participation).

Bone Turnover.
Literature shows an accelerated loss of bone mass in the paretic side [52], a high level of bone resorption, and a low level of markers of bone formation in poststroke patients [53].
In our review, Pang et al. [42] measured, with no significant results, biochemical markers of bone turnover (Ctelopeptide of type I collagen cross links and bone-specific alkaline phosphatase). Since the current literature may present beneficial results of WBV for bone mineral density, further studies are necessary to investigate WBV effects to the bone of poststroke patients. [41-43, 45, 46] did not show significant results. Tihanyi et al. [40] reported a variable muscle strength after a single WBV session: increase of maximum isometric knee extension torque (SES=0.50); increase of maximum eccentric knee extension torque (SES=0.46) on the paretic side; decrease of coactivation quotient of Biceps Femoral Muscle during isometric knee extension (SES=0.82) and Eccentric Knee Extension (SES=0.16). Liao et al. [46] examined 8 muscle strength parameters and 3 parameters for body functions and structures, with no significant results. Tankisheva et al. [50] reported better outcomes for the WBV than the control group: increase of isometric knee extension torque in paretic leg (week 6) (SES=1.74) and increase of Isokinetic knee extension strength (240 ∘ /s) in paretic leg (week 12) (SES=0.96), while in Van Nes et al. [49] both groups achieved similar improvements. This discrepancy is probably due to the difference in treatment duration and between the two control groups' treatments. In Van Nes et al. [49] we were not able to determine whether improvements are due to the conventional rehabilitation program (all participants took part in) or to additional WBV or to music therapy. Therefore we cannot say that WBV is a viable alternative to other types of therapy to deliver muscle strength improvements after stroke and other studies will be necessary to investigate the different effects varying WBV amplitude and duration.

Spasticity.
In Chan et al. [44] the WBV significantly reduced spasticity measured with the Modified Ashworth Scale (MAS) (p≤.001) [54] and Visual Analogic Scale (VAS) (SES=1.96), The Hmax/Mmax ratio decreased significantly more in the WBV group in the unaffected leg only (SES= 0.87), indicating a decrease in excitability of the stretch reflex pathway (Table 3). Participants were not "blind" to the treatment, so the increase of VAS can be a placebo effect. Of the 3 studies that measured spasticity after multiple WBV sessions [42,43,46,50], only Pang et al. [42] reported beneficial effects on knee spasticity, but no effects on ankle spasticity evaluated with MAS. Liao et al. [46] applied the Kruskal-Wallis-Test to knee and ankle MAS ordinal data, providing an interquartile range for these parameters and showing no significant difference between the three groups examined.
Literature shows that because of its ordinal nature and because it is related to muscular activity and resistance in response to passive movements [55,56], the MAS is probably not the best assessment for spasticity. To our knowledge this scale is the most used in selected studies, even if its results depend on the experience of the clinicians.
Evidences about the effects of WBV in reducing spasticity after stroke are insufficient in our review and it is impossible to declare the superiority of WBV compared to other rehabilitative processes.

Postural Control and Risk of Falls.
Chan et al. [44] reported beneficial effects of a single session of WBV on 14 Rehabilitation Research and Practice postural control; however this was assessed by only measuring the distribution of weight between the legs (increase of total body weight percentage on affected side, SES=0.87, and decrease on unaffected side, SES=0.87) disregarding other important parameters such as biomechanical constraints: sensory orientation, walking balance, etc. We cannot exclude a placebo effect, since the participants were not "blind" to the intervention. The effects of multiple WBV sessions on balance are insufficient. None of the 5 studies [43,45,46,49,50] that measured balance outcomes showed significant differences between the groups after a treatment period of 6-12 weeks, suggesting that WBV does not provide poststroke improvements in postural control. Brogårdh et al. [43] and Marin et al. [45] used the Berg-Balance-Scale (BBS) as the main balance outcome. In these studies the level of disability at baseline was quite moderate, probably due to the inclusion criteria (Table 2), reducing the significance of the improvements.
In Liao et al. [46] the balance performance in daily activities was measured by the Mini-Balance-Evaluation-System-Test [57], producing nonsignificant results about WBV effects. However, the data demonstrated a decisive time-effect on increased balance levels (P <.001) for all groups. Van Nes et al. [49] showed that postural control improvements produced by WBV were similar to other types of physical activities. Tankisheva et al. [50] asserted a superiority of WBV compared to usual physical activity for improving balance in an upright posture using a swaying platform (SES=1.47). However, the authors did not explain why they only reported this improvement without dismissing other balance outcomes. Only Choi et al. [47] analyzed balance control in the sitting position, reporting significant improvements in the Modified-Reaching-Functional-Test (MFRT) after WBV: Anterior reach (SES=0.51); Nonparetic reach (SES=0.60); Paretic reach (SES=0.38). Only one study [41] measured the incidence of falls and reported negative results. This was probably due to the fact that only 10% of the participants had at least one fall during the three months before treatment and the lack of any significant changes in motoneuron outcome variables.
The study of Lee G. [5], not considered by our systematic review because of an inadequate PEDro score (5/10), reported a significant increase in the equilibrium level measured with the Berg-Balance-Scale compared to pretreatment evaluations and the control group (difference of BBS score between baseline and follow-up: -6.00 ± 5.17 in the WBV group versus -0.56 ± 0.88 in the control group). We report this data because the research was conducted employing a platform that produced horizontal oscillations.
On the basis of these studies, we cannot recommend WBV to reduce the risk of falls in poststroke patients. (Lau et al. [41] reported a nonsignificant improvement in the incidence of falls during the period of follow-up between the WBV group and the control group who performed the same exercises, but without WBV.)

Functional Mobility.
Only Chan et al. [44] investigated changes in functional mobility however, there were profound differences among groups before treatment with the WBV group having a greater level of disability than the control group (longer Timed "Up&Go" Test (TUG) and 10-Meter-Walk-Test (10MWT) times). The initial differences between groups may have influenced the outcome results, decrease of TUG (SES=1.80) and increase of 10MWT (maximal speed) (SES=0.79), since there may have been more room for improvement in individuals with more severe mobility limitations. Three studies [42,43,46] produced outcomes related to mobility, indicating that WBV does not confer advantages in this regard. This may be due to the fact that the exercises involved only part of functional components associated with gait, given the limitations of the vibratory devices. One study [48], combining WBV with Treadmill training, measured improvements by GAITRite (CIR systems Inc., USA, 2008) in Walking speed (SES=0.241), Step length of affected side (SES=0.337), and Stride length (SES=0.318). Although these results were positive, they need to be supported by other studies with larger sample sizes. Based on the available evidences it is not possible to draw positive conclusions regarding the effects of WBV therapy to improve mobility poststroke.

Somatosensory Threshold.
The study by Van Nes et al. [49] showed improvements of somatosensory threshold in both WBV and control groups. No significative differences between groups were found.
All participants did the conventional rehabilitation program; so it was not possible to determine if the improvement of somatosensory threshold was due to conventional program, the additional use of WBV, or the music therapy program.

Activity and Participation.
The initial intention of the study was to explore the literature based on the ICF Stroke Brief. Unfortunately, it was very challenging and very little information was available about the activity and participation (especially d599 self-care, d729 general interpersonal interaction, or d230 carrying out daily routine) in selected studies. However, the effects of WBV on participation in social activities were investigated by Brogårdh et al. [43] with negligible differences in scores of the Stroke-Impact-Scale between groups. Van Nes et al. [49], comparing WBV and music therapy, reported nonsignificant differences in the assessments of functional mobility and daily activities. Liao et al. [46] investigated several outcomes, but without any reported improvements; therefore it can be concluded that WBV therapy does not improve participation in social life for people with stroke.

Limitations.
This systematic review included studies involving participants with nonhomogeneous characteristics, since studies with individuals with chronic and postacute stroke (disability level at baseline higher for the latter) were incorporated. Only 3 studies provided physiological explanations of the intervention protocol [41,42,46,49]. Two studies [40,50] had very low numbers of participants (≤20) reducing statistical power. In one study [44] there are profound differences between groups in impairment at the baseline. In other studies [42,43,46] there were detectable inadequacies in the protocols and instruments leading to poor correlations between Interventions and Outcomes. Some outcomes were described by ordinal variables, for which no data were provided on statistically significant improvements, only allowing a simple descriptive analysis.

Conclusions
By comparing WBV groups performing exercises during single or multiple sessions (4-12 weeks of treatment) to poststroke patients after the same exercises without WBV or other types of rehabilitation treatment, we are unable to demonstrate any significant systematic benefits from WBV treatment. This was mainly due to the heterogeneity of the studies completed to date. Though treatment with WBV appears safe and feasible and favourable in several outcomes, to our knowledge there are no sufficient evidences to support the integration of WBV in poststroke rehabilitation programs.
We are not able to highlight the differences between a synchronous and asynchronous vibration treatment, because there were no studies designed to investigate this aspect. Future RCTs may consider this topic and also the other parameters of the vibration platform, by continuing the research started by Liao 2016 [44] who investigated the effects of different stimulus intensities.
Future studies need to use outcome measures with good psychometric properties such as multiple measures for the same outcome, a statistically useful number of participants, and homogeneous disability characteristics for participants.