We searched PubMed, Embase, and Cochrane Library databases from inception to September 2022. All RCTs on WBVT in patients with stroke were searched. Search strategies for different databases are presented in Supplementary Table S1. The relevant reference lists in previous studies were checked to avoid possible omissions.

Two evaluators independently screened articles for eligibility based on the following criteria: 1) Participants: patients with stroke. 2) Interventions: WBVT defined in the experimental group. 3) Comparisons: routine rehabilitation (no WBVT or sham WBVT). 4) Physical function included spasticity, motor function, balance function, and gait performance; spasticity was measured by the Modified Ashworth Scale (MAS); motor function was assessed by the Fugl-Meyer Assessment (FMA), Manual Function Test (MFT), Rivermead Mobility Index (RMI), or Wolf Motor Function Test (WMFT); balance function was estimated by the Berg Balance Scale (BBS), Korea Berg Balance Scale (K-BBS), Activities Specific Balance Confidence (ABC), Tinetti Performance Oriented Mobility Assessment (TPOMA), Functional Reach Test (FRT), Modified Functional Reach Test (MFRT), or Mini Balance Evaluation Systems Test (Mini-BESTest); gait performance was measured by the Timed-Up-and-Go test (TUGT), 10-m Walk Test (10MWT), 6-Minute Walk Test (6MWT), Functional Ambulatory Category (FAC), or Tinetti Performance Oriented Mobility Assessment (TPOMA). 5) ADL was evaluated by the Barthel Index (BI), Korea modified Barthel Index (K-MBI), Functional Independence Measure (FIM), Frenchay Activity Index (FAI). 6) QOL was measured by the Stroke Impact Scale (SIS), Short-Form 12 Health Survey (SF-12). 7) Study design: RCTs. 8) Language: articles were published in English. Exclusion criteria included: 1) Unextractable data. 2) Abstracts or congress conference papers.

Two evaluators independently designed the electronic information form and extracted the following variables: 1) Study characteristics. 2) Participants characteristics. 3) Characteristics of WBVT. 4) Outcome data (the mean and standard deviation of scale scores). Non-compliant data formats were converted using the Cochrane Handbook’s recommended method. For studies with multiple WBVT groups with different frequencies, these groups were considered independent trials. If a multi-arm trial included WBVT with another intervention, we excluded the additional experimental group from analysis. When conducting an overall combined analysis of a functional outcome, if a study used multiple similar assessment scales, we selected the most commonly used, widely accepted, and reliably validated scale for inclusion in the overall analysis. This approach increased the comparability and consistency of the findings. Discrepancies between evaluators were resolved by discussion or input from a third evaluator.

The Cochrane Collaboration tool was developed for evaluating the bias risk of each eligible study (Sterne et al., 2019). Two evaluators independently used this tool to estimate the risk of selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases in each study.

The data were analyzed using Stata (version 17.0, Stata Corp, United States). The effect sizes of WBVT on functional outcomes were demonstrated using standardized mean difference (SMD) and 95% confidence interval (CI). The random effects model was applied to measure mixed effects, with p < 0.05 considered statistically significant. Heterogeneity was assessed using I ² statistics, with I ² > 50% indicating significant heterogeneity among studies. Publication bias was evaluated through funnel plots and Egger’s test, with asymmetric or a probability value of p < 0.05 indicating significant publication bias. Sensitivity analysis was performed by removing each study individually to assess the reliability of results. Subgroup analyses were mainly performed to explore the factors (frequency mode and vibration direction) influencing the effects of WBVT on physical function.

The PRISMA screening process was illustrated in Figure 1. A total of 928 records were retrieved from the database, and one record was identified from the references of previous studies. After removing duplicates, 607 remained. Following the scanning of title and abstract, 515 records were excluded, leaving 92 studies for full-text review. Ultimately, 25 articles met the conditions of quantitative meta-analysis.

The characteristics of the 25 selected studies are presented in Table 1. The studies were conducted in different regions, including 17 in Asia (Chan et al., 2012; Lau et al., 2012; Nagino et al., 2013; Choi et al., 2014; Guo et al., 2015; Lee, 2015; Lee et al., 2016a; Choi et al., 2016; Liao et al., 2016; Choi et al., 2017; Lee et al., 2017; Hwang, 2018; Ahn et al., 2019; Dang and Chen, 2019; Lee et al., 2020; Kim and Lee, 2021; Wei and Cai, 2022), 5 in Europe (van Nes et al., 2006; Merkert et al., 2011; Brogårdh et al., 2012; Marín et al., 2013; Tankisheva et al., 2014), 1 in America (Silva et al., 2014), and 2 in Turkey (Alp et al., 2018; Sade et al., 2020). A total of 991 individuals were enrolled, with 6 studies not reporting specific stroke subtypes (Merkert et al., 2011; Choi et al., 2014; Choi et al., 2016; Choi et al., 2017; Alp et al., 2018; Dang and Chen, 2019), 3 articles not providing the side of lesion (Merkert et al., 2011; Guo et al., 2015; Ahn et al., 2019), and 3 articles not providing disease duration (Merkert et al., 2011; Choi et al., 2016; Lee et al., 2020). Most studies reported subacute and chronic stages. Frequency mode of WBVT was categorized into fixed frequency (van Nes et al., 2006; Brogårdh et al., 2012; Chan et al., 2012; Nagino et al., 2013; Silva et al., 2014; Choi et al., 2016; Liao et al., 2016; Lee et al., 2017; Alp et al., 2018; Lee et al., 2020; Kim and Lee, 2021; Wei and Cai, 2022) and variable frequency (Merkert et al., 2011; Lau et al., 2012; Marín et al., 2013; Choi et al., 2014; Tankisheva et al., 2014; Guo et al., 2015; Lee, 2015; Lee et al., 2016a; Choi et al., 2017; Hwang, 2018; Ahn et al., 2019; Dang and Chen, 2019; Sade et al., 2020). According to the vibration direction, WBVT can be divided into side-alternating (van Nes et al., 2006; Marín et al., 2013; Nagino et al., 2013; Choi et al., 2014; Lee, 2015; Lee et al., 2016a; Choi et al., 2016; Choi et al., 2017; Hwang, 2018; Ahn et al., 2019; Dang and Chen, 2019; Lee et al., 2020; Wei and Cai, 2022) and vertical vibration (Merkert et al., 2011; Brogårdh et al., 2012; Chan et al., 2012; Lau et al., 2012; Tankisheva et al., 2014; Guo et al., 2015; Liao et al., 2016; Lee et al., 2017; Alp et al., 2018; Sade et al., 2020; Kim and Lee, 2021).

The assessment of bias risk in the included studies is presented in Figure 2. For the random sequence generation method, 14 studies were rated as low risk, and 1 study were rated as high risk. Eight studies used allocation concealment. The risk of blinding in participants was unclear in most studies. Eleven studies implemented a blind method for outcome evaluators and all studies reported a low risk of incomplete outcome data.

We conducted subgroup analysis based on the frequency mode and the vibration direction of WBVT. Regarding the parameters of WBVT, variable frequency vibration and side-alternating vibration exhibited significant efficacy in reducing spasticity, as well as improving motor and balance functions. Conversely, fixed frequency vibration and vertical vibration did not show significant therapeutic effects on any of these physical functions. (Table 2).

In the physical function domain, including balance function and gait performance, the combined assessment of the funnel plot symmetry distribution (Supplementary Figure S5) and Egger’s test results (p = 0.0154 and p = 0.0556, respectively) suggested a possible publication bias. However, sensitivity analysis using the leave-one-out method did not show any significant changes in the overall conclusions, indicating the reliability of the results.

The underlying mechanism of WBVT remains unclear, but it is believed that vibration stimulation activates a variety of human skin and deep tissue receptors, especially class Ia and II afferent fibers, increasing sensory input to the central nervous system (Roll et al., 1989). This stimulation affects the excitability of functional brain areas such as the primary somatosensory cortex, primary motor areas, and premotor areas, promoting neuroplasticity (Murillo et al., 2014; Lapole et al., 2015; Lopez et al., 2017). Additionally, vibration stimulation on muscles can cause a tonic vibration reflex, leading to involuntary muscle contraction and affecting proprioception and motor performance (Murillo et al., 2014). Studies utilizing surface electromyography (sEMG) have demonstrated that WBVT can effectively activate muscles in individuals with conditions such as stroke and spinal cord injury (Dionello et al., 2017).

The results of our analysis demonstrated the anti-spasticity effect of WBVT, which was also confirmed in previous studies (Park et al., 2018; Yang et al., 2022). Since stroke patients may experience spasticity in various body parts (Wissel et al., 2010), and a recent meta-analysis has confirmed the effectiveness of WBVT in both upper and lower extremity spasticity (Zhang et al., 2023), it is essential to explore the effects of WBVT on spasticity in different body parts. However, most relevant trials did not report the specific spastic muscle groups assessed by MAS, making it difficult to determine the efficacy of WBVT for specific spastic muscles. Additionally, most included RCTs did not provide precise details regarding antispasmodic drug therapy, preventing us from further evaluating the possible role of these drugs in the anti-spasticity effect of WBVT. Future research should explore the effects of WBVT on spasticity in different body parts and the role of pharmacotherapy in the anti-spasticity effect of WBVT.

For motor function, previous meta-analyses mainly investigated the effect of WBVT on lower muscle strength, with a small effect size (Park et al., 2018; Yang and Butler, 2020). In contrast, our study aimed to directly explore the effect of WBVT on motor function using motor function scales. The meta-analysis revealed a significant effect size in favor of WBVT in improving motor function, particularly in the upper extremity. This result is consistent with a recent meta-analysis, although its credibility was limited due to language barriers (only two non-English reports were included) (Yang et al., 2022). However, variations in study designs prevented us from further analyzing the potential reasons for this conclusion. The significant effects observed on the FMA, indicating the positive impact of WBVT on motor function, highlight its potential as an effective intervention for stroke patients. Nonetheless, caution is advised when interpreting the non-significant effects found on the RMI and WMFT, considering the limited number of included studies. To validate these findings and explore the specific influence of WBVT on these motor function domains in stroke patients, future research with larger and more diverse samples is warranted. Furthermore, maintaining methodological consistency and using similar motor function scales in subsequent studies will facilitate accurate comparisons and enable integrated analyses.

Impaired balance and unstable posture increase the risk of falling in the majority of stroke patients, even with improved motor function. Poor balance control negatively affects the recovery of motor function and gait performance, which are crucial for restoring functional independence in stroke patients (de Haart et al., 2004; Lee et al., 2016b). Our meta-analysis supports previous findings that WBVT improvs balance scores in patients with stroke (Yang and Butler, 2020; Yin et al., 2022). The possible mechanisms of balance recovery after WBVT may be an increase in proprioception and improvement in weight-bearing to the hemiplegic side (Hwang, 2018). However, the impact of publication bias in the included studies should be considered when interpreting our results, although sensitivity analysis indicated that its impact may be limited. Therefore, further studies with larger sample sizes, standardized intervention protocols, and consistent outcome measures are needed to confirm our findings.

To gain a deeper understanding of the varying effects of WBVT on different balance assessment scales, we conducted a subgroup analysis specifically on the balance scale. Our findings indicate a significant therapeutic effect of WBVT in improving balance function in stroke patients when utilizing the BBS. This may be attributed to the high sensitivity and accuracy of the BBS in assessing balance function in this population. However, it is worth noting that the other five balance scales (ABC, K-BBS, MFRT, Mini-BESTest, and TPOMA) did not show significant effect sizes. The limited number of studies available for analysis and the potential influence of multiple factors on the results of a single study should be considered, which may impact the reliability of these conclusions. Therefore, additional studies are necessary to strengthen support for these findings. In conclusion, while WBVT has demonstrated positive effects in improving balance function in stroke patients, its applicability may vary across different balance assessment scales. Further research is needed to validate and substantiate these findings.

For gait performance, various clinical scales are used to evaluate the improvement in gait performance of individuals with stroke. These include the 6MWT to assess walking endurance (Fulk et al., 2008), the FAC with 6 evaluation levels of gait capacity (Mehrholz et al., 2007), the TPOMA with 7 gait evaluation items (such as starting, step length, gait symmetry, step continuity, walking path, and step width) (Abbruzzese, 1998), the TUGT to assess balance capability and functional mobility (Podsiadlo et al., 1991), and the 10MWT to reflect changes in walking speed (Holden et al., 1984). Our study demonstrated a small but significant effect size of the 10MWT, suggesting a limited but positive effect of WBVT on walking speed. The potential benefits of WBVT on walking speed might include the reduction of ankle spasticity and improvement in ankle joint control (Goldie et al., 2001; Lamontagne et al., 2002). However, our meta-analysis showed an insignificant effect size for the total random effect of gait scale scores, including the 6MWT, FAC, TPOMA, and TUGT. These results suggest that WBVT did not improve overall gait performance compared to controls, which is inconsistent with previous meta-analyses (Yang and Butler, 2020; Yin et al., 2022). The difference in results may be attributed to the mixed study design due to differences in inclusion exclusion criteria, study methodology, and other factors across studies.

Although clinical scales are commonly used to evaluate gait performance in patients with stroke, they lack objective and accurate indicators to reflect temporospatial variables of gait, such as step length, cadence, and walking speed. Previous studies have reported mixed results regarding the effects of WBVT on gait variables in stroke patients. Some studies have shown significant improvements in walking speed, step length, and limb support (Choi et al., 2017; Sade et al., 2020), while others have not observed significant differences in gait variables between the WBVT group and control group (Lee, 2019). A recent meta-analysis also preliminarily validated the significant improvement of WBVT in terms of stride length, walking speed, cadence and stride length in stroke patients (Yin et al., 2022). However, differences in measurement devices may lead to measurement biases that could affect the results of the analysis. Due to the limited number of RCTs using computerized gait analysis, more research is needed to draw reliable conclusions on the effectiveness of WBVT in improving gait in stroke patients.

The level of independence in performing ADL is crucial for the QOL of stroke patients (Clarke et al., 2000; Kim et al., 2014). Therefore, when developing rehabilitation programs, it is essential to address not only physical function improvement but also the reduction of ADL dependency and enhancement of QOL in an integrated manner. Unlike previous meta-analyses, our comprehensive study aimed to analyze the rehabilitative effects of WBVT in stroke patients, specifically examining its impact on ADL and QOL. However, our analysis of ADL scores revealed a non-significant overall effect size of WBVT, indicating insufficient evidence to determine its efficacy in improving ADL. It is important to acknowledge the considerable heterogeneity among the included studies, which may be attributed to variations in study design, sample characteristics, vibration treatment protocols, and ADL assessment tools employed, leading to inconsistency and uncertainty in the study results.

Interestingly, a study by Hwang (Hwang, 2018) reported improvements in walking ability and ADL in stroke patients who received WBVT, particularly among those showing enhanced balance scores. This finding suggests that enhancing balance has a significant impact on improving gait performance and ADL. Our findings indicate that WBVT may be effective in improving spasticity, motor function, and balance in stroke patients. However, it is important to note that positive outcomes in physical function do not always translate into improvements in ADL, as observed in some studies (Hwang, 2018). Another important consideration is that the recovery of walking ability is crucial for improving ADL and QOL in stroke patients (Xu et al., 2016; Wonsetler and Bowden, 2017), which may lead to conflicting outcomes. Therefore, future research on WBVT in stroke patients should focus on standardizing protocols, considering sample characteristics, utilizing validated outcome measures, improving study design, and investigating underlying mechanisms to enhance understanding of its effects on ADL.

Impaired physical function and mobility have consistently been linked to decreased QOL in stroke patients (Clarke et al., 2000; Geyh et al., 2007; Teixeira-Salmela et al., 2009; Gunaydin et al., 2011). Recognizing the importance of QOL, we aimed to assess the effectiveness of WBVT on QOL outcomes. However, the limited number of available RCTs in this specific area contributes to the uncertainty of our study’s findings. Future research should prioritize investigating the impact of WBVT on QOL to provide a more comprehensive understanding of its potential benefits in stroke patients.

The optimal parameters of WBVT for functional recovery in stroke patients remains uncertain. Our subgroup analysis showed that variable frequency and side-alternating vibration had a more significant therapeutic effect on reducing spasticity, improving motor and balance function, while evidence for fixed frequency and vertical vibration was lacking. Thus, variable frequency and side-alternating vibration of WBVT differentiated more effective and less effective trials, providing valuable information for designing future WBVT-based trial protocols. Perchthaler et al. (Perchthaler et al., 2015) also suggested that using side-alternating vibration and increasing the knee flexion position could reduce head acceleration and damage to the eyes and ears. However, it is worth noting that the WBVT used in the included trials was mostly individually prescribed by professionals to minimize the risk of potential injuries, such as falling during training. Therefore, when implementing precise, individualized WBVT in clinical practice, modulating factors of successful interventions should be taken into account. Distinguishing the characteristics of successful and unsuccessful interventions is crucial in optimizing the therapeutic benefits of WBVT for stroke patients.

This meta-analysis had several limitations. Firstly, the analysis did not include an examination of other parameters of WBVT, such as amplitude and stimulation time. This omission was due to the variations in vibration equipment and training programs across the included studies. Therefore, following comprehensive guidelines for reporting WBVT studies will help to improve the quality and comparability of future WBVT studies in stroke, thereby promoting the development and deepening of the clinical application of WBVT in stroke (Wuestefeld et al., 2020). Secondly, the limited number of RCTs assessing QOL outcomes restricted our ability to comprehensively evaluate the effectiveness of WBVT in this domain. Furthermore, the utilization of different clinical function scales among the included studies may have introduced measurement heterogeneity, despite efforts to minimize its impact through the calculation of SMD values. Moreover, the wide confidence intervals and the quality of the studies included in this meta-analysis may limit the generalizability and interpretation of the results. Therefore, further well-designed studies are warranted to validate and confirm the findings presented in this meta-analysis.