Study selection and quality appraisal

A total of 1327 eligible studies were identified and seven studies (Arias et al., Reference Arias, Chouza, Vivas and Cudeiro2009; Chouza et al., Reference Chouza, Arias, Viñas and Cudeiro2011; Ebersbach et al., Reference Ebersbach, Edler, Kaufhold and Wissel2008; Gaßner et al., Reference Gaßner, Janzen, Schwirtz and Jansen2014; Kapur et al., Reference Kapur, Stebbins and Goetz2012; Kaut et al., Reference Kaut, Allert, Coch, Paus, Grzeska, Minnerop and Wüllner2011, Reference Kaut, Brenig, Marek, Allert and Wüllner2016) (196 patients) satisfied the inclusion criteria (Fig. 1). The characteristics of the selected studies are presented in Table 1. Out of the seven studies with available data, one study (Ebersbach et al., Reference Ebersbach, Edler, Kaufhold and Wissel2008) compared the effect of WBV vs. conventional physical therapy while the other compared WBV with a non-vibration control. One study (Kapur et al., Reference Kapur, Stebbins and Goetz2012) delivered vibration stimulus by a physioacoustic chair, while the other studies required a vibrational platform. Disease severity as defined by Hoehn and Yahr stage was reported in four studies (Gaßner et al., Reference Gaßner, Janzen, Schwirtz and Jansen2014; Kapur et al., Reference Kapur, Stebbins and Goetz2012; Kaut et al., Reference Kaut, Allert, Coch, Paus, Grzeska, Minnerop and Wüllner2011, Reference Kaut, Brenig, Marek, Allert and Wüllner2016). The Hoehn and Yahr scale, a gold-standard staging assessment that evaluates the severity of PD based on bilateral motor dysfunction and the compromise of gait and balance. It ranges from 1 to 5 with 1 being mild symptoms and 5 being the most severe cases bound to wheelchair (Zhao et al., Reference Zhao, Wee, Chan, Seah, Au, Lau and Tan2010). The baseline evaluation on Hoehn and Yahr stage of PD patients was 1–4 in two studies (Kaut et al., Reference Kaut, Allert, Coch, Paus, Grzeska, Minnerop and Wüllner2011, Reference Kaut, Brenig, Marek, Allert and Wüllner2016), 2–3 in one study (Gaßner et al., Reference Gaßner, Janzen, Schwirtz and Jansen2014), and 2 in Kapur’s study. All selected studies with available data scored 5 or higher on PEDro scale (Table 2).

Figure 1.

PRISMA 2009 flow diagram.

Table 1.

Profiles of Included Studies

NR, not reported.

Table 2.

Supplementary Materials

Synthesis of results

Six trials, which included 172 participants, investigated the effect of WBV on UPDRS-III at EOT. The pooled mean difference of six trials with 172 patients was −1.75 with 95% CI [−5.40, 1.90], while no significant heterogeneity was detected (p = 0.10; I2 = 45.8%). As shown in Fig. 2(a), the synthesis result from the UPDRS-III indicated that WBV was not significantly better than placebo or conventional physical therapy at alleviating symptoms of PD.

Figure 2.

Forest plots of outcome measures at EOT for comparing WBV vs. control in PD patients; (a) UPDRS-III; (b) Time Up and Go test; (c) Functional Reach Test; (d) Berg balance test and Tinetti score; (e) walking velocity.

Five studies reported data concerning balance; three of these studies (Arias et al., Reference Arias, Chouza, Vivas and Cudeiro2009; Chouza et al., Reference Chouza, Arias, Viñas and Cudeiro2011; Gaßner et al., Reference Gaßner, Janzen, Schwirtz and Jansen2014) used functional reach test, two (Ebersbach et al., Reference Ebersbach, Edler, Kaufhold and Wissel2008; Kaut et al., Reference Kaut, Brenig, Marek, Allert and Wüllner2016) used the Tinetti test, and one (Arias et al., Reference Arias, Chouza, Vivas and Cudeiro2009) used the Berg balance scale. Given these scales were established with different measures, data analysis was conducted separately concerning functional reach test and other balance scales. Gaßner et al. (Reference Gaßner, Janzen, Schwirtz and Jansen2014) modified the functional reach test with participants reaching forward to grasp a pen, while Arias et al. (Reference Arias, Chouza, Vivas and Cudeiro2009) and Chouza et al. (Reference Chouza, Arias, Viñas and Cudeiro2011) had patients displace a movable stick attached to the anthropometer. It should be noted that the values of functional reach test in Gaßner et al. (Reference Gaßner, Janzen, Schwirtz and Jansen2014) as seen in Table 1 were much higher as it included the length of upper limb. Therefore, standardized mean difference was calculated on functional reach test and no statistical difference was detected (SMD = 0.21, 95% CI, −0.29, 0.71) (Fig. 2(c)). Pooled data on Berg balance scale and Tinetti score from three studies, including 96 patients, indicated that there was no significant difference in balance scores between WBV and control groups (SMD = 0.39, 95% CI, −0.01, 0.80) (Fig. 2(d)). The test for heterogeneity showed that these data were largely homogeneous (I ² = 0%, Fig. 2(c) and (d)).

As shown in Fig. 2(b), the pooled analysis from four studies on the Time Up and Go test showed favorable results for WBV (WMD = −1.59, 95% CI, −2.90, 0.28; I ² = 0%). In comparison, pooled data from two studies with 38 patients showed no benefits for WBV on walking velocity (Fig. 2(e), WMD = −0.05, 95% CI [−0.17, 0.06]). The Cochrane Q test and I ² revealed that the data had no significant heterogeneity (p = 0.46, I ² = 0%). Taken together, these data show that WBV did not contribute any additional benefits to alleviating the balance and walking velocity symptoms of PD when compared with placebo or conventional physical therapy. However, the pooled analysis of four studies, including 116 patients, assessing Time Up and Go test showed favorable results for WBV. No side effects and falls were reported in the selected studies.

Sensitivity analyses and publication bias

Our results revealed that no single study statistically influenced the pooled WMD and 95% CIs (Fig. 3(a) and (b)). The p values of Begg’s test and Egger’s test were greater than 0.05, suggesting no publication bias in the present study (Fig. 4(a) and (b)). In other words, our results were comparatively credible and stable.

Figure 3.

Sensitivity analysis of each included study. (a) UPDRS-III for individual studies; (b) Time Up and Go test for individual studies.

Figure 4.

Begg’s funnel plots of the publication bias. (a) UPDRS-III for individual studies; (b) Time Up and Go test for individual studies.

Discussion

The present systematic review and meta-analysis identified seven clinical trials that included 196 patients, which compared WBV with placebo vibration or conventional physical therapy. The results of this review indicated that WBV may not be beneficial at alleviating UPDRS motor, balance, and walking velocity scores when compared placebo vibration or conventional physical therapy. Previous systematic reviews (Sharififar et al., Reference Sharififar, Coronado, Romero, Azari and Thigpen2014; Sitjà Rabert et al., Reference SitjàRabert, Rigau Comas, Fort Vanmeerhaeghe, Santoyo Medina, Roqué i Figuls, Romero-Rodríguez and Bonfill Cosp2012) have shown consistent results concerning the effect of WBV on these outcomes. Within group effects have been observed on motor scores in the UPDRS and balance for patients with PD in groups treated with WBV; therefore, this is likely a result of nonspecific placebo effects. In contrast, our pooled analysis yielded robust and consistent findings in favor of WBV on Time Up and Go test scores. The results are consistent across studies despite the fact that these studies differed in treatment duration, vibration frequency, and cumulative dose. The Time Up and Go test involves a sequence of activities that requires participants to start in a sitting position, stand, walk for 3 m, then turn around, and walk back to sit (Podsiadlo & Richardson, Reference Podsiadlo and Richardson1991). Our results indicated that WBV had favorable results on Time Up and Go test scores; however, no significant difference in walking velocity was detected. One plausible explanation for this is that WBV might have a positive effect on sit-to-stand transition or turning. This could be something worth research attention since previous studies showed that 50% of PD patients had difficulty turning and thereby increased risk of falls (Cheng et al., Reference Cheng, Yang, Wang, Wu, Cheng, Wang and Wang2014; Stack et al., Reference Stack, Jupp and Ashburn2004). However, the underlying mechanisms of WBV on patients with PD are poorly understood. It has been hypothesized that exercise training promotes neuroplasticity and neurorepair in patients with PD (Petzinger et al., Reference Petzinger, Fisher, McEwen, Beeler, Walsh and Jakowec2013). Studies using animal models have demonstrated that exercise can restore glutamate receptor expression (Van Leeuwen et al., Reference VanLeeuwen, Petzinger, Walsh, Akopian, Vuckovic and Jakowec2010), increase dendritic spine density in certain regions (Stranahan et al., Reference Stranahan, Khalil and Gould2007), and enhance hippocampal cerebral blood flow (Villar-Cheda et al., Reference Villar-Cheda, Sousa-Ribeiro, Rodriguez-Pallares, Rodriguez-Perez, Guerra and Labandeira-Garcia2009). By stimulating muscle contraction, WBV may elicit a similar mechanism to exercise training. A preliminary fMRI study also showed activation of basal ganglia in healthy individuals after WBV training, indicating the neuromodulative role of WBV (Kaut et al., Reference Kaut, Brenig, Marek, Allert and Wüllner2016). Besides, patients with PD often have trouble in postural transitions such as sit-to-stand and turning as they tend to trigger freezing (inability to move despite intention) compared to straight-ahead gait (Jacobs et al., Reference Jacobs, Lou, Kraakevik and Horak2009). This was attributed to impaired anticipatory postural adjustments (APAs), a stabilizing phase engaging postural control prior to step initiation (de Souza Fortaleza et al., Reference de Souza Fortaleza, Mancini, Carlson-Kuhta, King, Nutt, Chagas, Freitas Junior and Horak2017; Mezzarobba et al., Reference Mezzarobba, Grassi, Valentini and Bernardis2018). On the other hand, clinical studies have shown that WBV may contribute to better postural control in patients with PD (Kaut et al., Reference Kaut, Brenig, Marek, Allert and Wüllner2016; Turbanski et al., Reference Turbanski, Haas, Schmidtbleicher, Friedrich and Duisberg2005). The potential positive effect of WBV on sit-to-stand transition or turning could be a result of enhanced postural control in patients with PD.

Compared with the previous Cochrane review (Sitjà Rabert et al., Reference SitjàRabert, Rigau Comas, Fort Vanmeerhaeghe, Santoyo Medina, Roqué i Figuls, Romero-Rodríguez and Bonfill Cosp2012), we identified four additional RCTs and seven trials were included for meta-analysis. As the present review aimed to investigate the effect of WBV on motor function in PD, a study previously identified in the review was excluded as it examined the effects of WBV on proprioceptive performance (Haas et al., Reference Haas, Buhlmann, Turbanski and Schmidtbleicher2006b). The Cochrane review also included one study (Haas et al., Reference Haas, Turbanski, Kessler and Schmidtbleicher2006a) that adopted cross-over design without a wash-out period, which we set to avoid in the protocol. In this context, pooled analysis was conducted based on relatively rigorous design. However, these data should be interpreted with caution due to several limitations of the present study. The first limitation is that these findings are based on studies that have potential bias despite the relatively high score on PEDro scale. Most of the selected studies have an unclear risk of bias on random sequence generation and allocation concealment. Moreover, the trial conducted by Arias et al. (Reference Arias, Chouza, Vivas and Cudeiro2009) showed a high risk of selection bias because they performed a group distribution using individual patient orders to assess the gait test. In addition, six out of seven studies compared WBV with placebo vibration, while half the studies showed a high risk for performance bias. Gaßner et al. (Reference Gaßner, Janzen, Schwirtz and Jansen2014) used a placebo group that included participants in the same position as the vibration group while audible and noticeable signs of vibration were delivered. By contrast, Kaut et al. (2011, Reference Kaut, Brenig, Marek, Allert and Wüllner2016) administered participants in the placebo group with the lowest frequency of vibration. In three studies (Arias et al., Reference Arias, Chouza, Vivas and Cudeiro2009; Chouza et al., Reference Chouza, Arias, Viñas and Cudeiro2011; Kapur et al., Reference Kapur, Stebbins and Goetz2012), participants were required to use the same position as the experimental group while no vibration was administered, which compromised the blinding. Second, the number of studies included in the present review was too small to identify subgroups concerning the effect of WBV on patients with PD. Furthermore, few studies have examined the long-term effect of WBV on patients with PD.

The most robust finding to emerge from this study is that WBV may not improve motor scores of the UPDRS, balance, or walking velocity. However, WBV improved Time Up and Go test scores when compared with placebo vibration according to consistent pooled data. This result suggests that WBV might have favorable effects on sit-to-stand transition and/or turning, which awaits confirmation from clinical studies. The long-term effects of WBV and optimal parameters also call for further work.