The mean and 95% CI are shown. The horizontal line indicates zero mean change from before to after treatment. The 95% CI is shown by the whisker plot around the mean. EVT indicates external vibration therapy; F0, fundamental frequency; JITT, percentage jitter; and SHIM, percentage shimmer.
The mean difference and 95% CI are shown. The horizontal line indicates zero change. EVT indicates external vibration therapy; SVT, soft vocal task; and VRP, voice range profile.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Anderson J, DeLuca M, Haines M, Merrick G. Immediate Effects of External Vibration vs Placebo on Vocal Function Therapy in Singers: A Randomized Clinical Trial. JAMA Otolaryngol Head Neck Surg. 2018;144(3):187–193. doi:10.1001/jamaoto.2017.2679
Does external vibration therapy have an immediate effect on voice production in trained singers?
In this randomized clinical trial of 27 participants, results showed a more consistent change in acoustic metrics after external vibration therapy than placebo treatment. Effort ratings improved after both external vibration therapy and placebo treatment; however, the results were not found to be meaningfully different between the groups.
External vibration therapy was found to have a more predictable change in acoustic metrics after treatment than placebo; subjective effort to perform 6 voice tasks evaluated in the study was not found to be different between the external vibration therapy and placebo treatment.
External vibration therapy (EVT) has been widely used in chronic pain conditions, musculoskeletal rehabilitation, and athletic training. Vibration therapy has been suggested to enhance vocal performance and has been popularized in social media. However, there is no evidence to support its effect on vocal function.
To evaluate the immediate effects of EVT in trained singers using acoustic and self-assessment parameters.
Design, Setting, and Participants
Prospective, randomized, placebo-controlled interventional study at St Michael’s Hospital Voice Clinic, affiliated with the University of Toronto. Data collection and analysis were performed by investigators who were blinded to the group assignment of the participants. Study participants were randomized to EVT or a placebo (control) group. The study dates were September 2015 to December 2016.
Participants attended the voice laboratory at St Michael’s Hospital, where a standardized data collection protocol was performed, including acoustic parameters, voice range profile, and soft voice tasks, followed by subjective rating of vocal effort or discomfort. The EVT group underwent EVT to 5 neck sites bilaterally. The placebo group underwent the same protocol with a modified device. After the intervention, the participants repeated the standardized data collection.
Main Outcomes and Measures
The primary outcome in this study was acoustic analysis (jitter, shimmer, and pitch range) compared before and after treatment. In addition, secondary outcomes included perceived effort or discomfort evaluated by participants after 4 voice tasks proposed to investigate more subtle voice properties. Within and between groups, data sets were statistically analyzed for potential treatment effect.
Among 27 participants (age range, 18-50 years; all female), 14 were randomized to the intervention group and 13 to the placebo group. Comparison of the treatment effect on the vowel token acoustic parameters evaluated showed that, after EVT, participants had a more cohesive change with a restricted 95% CI compared with placebo. The mean change in fundamental frequency after intervention was 5.00 Hz in both groups but the 95% CI was much wider after placebo (−30.30 to 19.20) than after EVT (−18.10 to 7.50). After EVT, the effect size was notable in the vowel (0.83) and SVT3 (0.79) task.
Conclusions and Relevance
In this study, EVT demonstrated a more predictable change in acoustic metrics compared with the placebo treatment. Effort ratings for 6 voice tasks evaluated in this study were not found to be different after EVT compared with the placebo treatment.
clinicaltrials.gov Identifier: NCT02083341
External vibration has been widely used as a therapeutic tool to treat a variety of musculoskeletal injuries and chronic pain conditions and to promote muscle strengthening.1-3 Vibration therapy has been suggested to enhance vocal performance and has been popularized in social media. However, the mechanisms of action are not well understood. Delivery of external vibration therapy (EVT) has been primarily through either direct cutaneous treatment over a specified site with a portable device or using a whole-body platform.1-5 The immediate effects of external vibration that have been reported include pain reduction3,4,6,7 (vibratory analgesia), increased muscle perfusion,1 and prevention or reduction of delayed-onset muscle soreness4 after exercise. According to a systematic review1 of the literature published in 2013, one of the immediate effects in skeletal muscle after EVT is a dramatic increase in muscle perfusion or oxygenation (mean, 29%). In addition, passive transcutaneous vibration over skeletal muscles causes a reflex contraction (also known as the tonic vibration reflex), and EVT before exercise may prevent pain, stiffness, and loss of range of motion due to delayed-onset muscle soreness.4 External vibration therapy has also been shown to improve skeletal muscle resistance strength and flexibility in athletes through various mechanisms, including increased hormone production, muscle hypertrophy, and activation of the tonic vibration reflex.1,2 Vibratory analgesia is as yet poorly understood but has been largely attributed to the gate control theory of inhibition first described by Wall and Cronly-Dillon7 and by Melzack and Wall.6 It has been shown that, in the presence of vibration, the activated mechanoreceptors inhibit dorsal horn pain signaling fibers in the spinal cord. Other mechanisms involved are central in nature, with mutual antagonism in the somatosensory cortex between pain and mechanical stimuli.5-8
Certain desirable acoustic qualities in a singer’s voice are only accessible when the larynx and its extrinsic muscles are in a relaxed state.9,10 Muscle tension and vocal strain are common problems in performers. Professor David Ley, a vocal pedagogist, has developed a Vibrant Voice Technique10 based on EVT. He proposes that the use of EVT may reduce tension, improve range and projection in singers by the release of chronic tension, and warm up the musculature.10-12 We could not identify any other reference to EVT and vocal function in an English-language literature search. However, there was sufficient evidence in the literature describing the use of EVT for other skeletal muscle sites1-4,6,7 to justify investigating its possible effects on voice production.
In vocalists, other techniques used to facilitate voice production that have been reported include manual laryngeal reduction technique (posited as increasing relaxation and flexibility)9 and vocal warm-up exercises.13 Acoustic parameters, such as perturbation measures and pitch range, have been used to evaluate the effect of these techniques.9,13 Similarly, in our study, acoustic data were collected to evaluate possible immediate effects of EVT on these parameters.
The inability to produce soft, high-pitched phonation has been correlated with increased vocal effort and higher phonation threshold pressure.14 To investigate potentially subtle, immediate effects of EVT, the participants in our study were asked to rate effort and discomfort after performing 4 soft, high-pitched voice tasks.15,16
The objective of this study was to evaluate the immediate effects of external vibration in trained singers using acoustic and self-assessment parameters. The study design was a prospective randomized clinical trial comparing acoustic parameters and a self-reported rating of effort or discomfort before and after EVT or placebo treatment. The trial protocol is available in the Supplement. The study was carried out at St Michael’s Hospital Voice Clinic, which is affiliated with the University of Toronto.
The study was approved by the St Michael’s Hospital Research Ethics Board. All participants provided written informed consent for enrollment in the study.
Although anecdotal reports of this form of vibration therapy have been reported in the media,10-12 there is neither a standardized protocol for its use nor a validated study of the potential effects on voice production. The St Michael’s Voice Clinic team (J.A., M.D., M.-E.H., and G.M.) formed a focus group led by a senior vocal pedagogist, 2 laryngologists, and 2 experienced voice-specialized speech-language pathologists to develop a pilot protocol of EVT. The protocol was designed to potentially assist singers and treat voice disorders, such as muscle tension dysphonia. Several studies have evaluated optimal frequency settings, and 30 to 75 Hz has generally been used for musculoskeletal and pain reduction effects.1 The vibratory device (SIRI; LELO), vibration frequency (50 Hz), duration, and treatment sites were determined based on the literature and a pilot study,17 along with specific therapy instructions. The protocol targets muscle sites in the jaw, neck, and strap muscles, which are common areas of excessive tension in singers, and certain forms of dysphonia. To test the effect of the vibration therapy protocol, a prospective randomized clinical trial was carried out.
Participant recruitment was through local poster advertisements in the St Michael’s Hospital Voice Clinic, the Canadian Opera Company in Toronto, the Faculty of Music at the University of Toronto, and various local choirs. Participants were consented by a research coordinator for participation and then randomized (using RANDOM.ORG, an online list randomizer) to EVT or a placebo (control) group. The study dates were September 2015 to December 2016. Twenty-seven trained singers were successfully recruited (14 EVT and 13 placebo) and enrolled in the study. Three other individuals were excluded based on laryngeal pathology on videostroboscopy. Sample size estimates were conducted using an online power estimator for paired-samples t tests.18 Parameter estimates were based on observed differences before and after vocal warm-ups in female singers.13 Power analysis (90%) revealed that, before and after treatment, differences may be observed when the sample size was greater than 10.
All participants attended the voice laboratory at St Michael’s Hospital for a single session lasting between 1 and 2 hours, where a standardized data collection protocol was performed, including acoustic parameters, voice range profile (VRP), and soft voice tasks, followed by subjective rating of vocal effort or discomfort. All participants underwent a videostroboscopy to rule out laryngeal pathology before acoustic recording.
The pretreatment data collected from each group were compared to establish that the participants were similar at baseline within each group. Consideration of homogeneity at the outset of the study between the placebo and EVT groups was extended to include examination of 95% CIs, as well as Cohen d (comparison of the effect size), to ensure that important clinical effects were not blinded by tests of significance. Given possible sizable breadth in standard deviations, within-group analyses were planned using paired-samples t tests for each outcome variable (before and after treatment differences).
Self-reported effort ratings for 6 voice tasks evaluated in the study were compared before and after treatment by calculating the mean change in rating for each task. Examination of the 95% CIs and Cohen d values was performed to ensure that important clinical information was included.
There were 4 criteria for inclusion in the study. These were (1) female singer with a minimum of 3 years of vocal training, (2) age between 18 and 50 years, (3) the ability to give informed consent, and (4) native English speaker.
There were 5 criteria for exclusion from the study. These were (1) a score greater than 10 on the Singing Voice Handicap Index, (2) recent upper respiratory tract infection, (3) a history of smoking, (4) systemic illness affecting the upper aerodigestive tract (ie, asthma or allergies), and (5) the presence of laryngeal pathology on stroboscopic examination.
After informed consent was obtained and the videostroboscopy was screened to ensure a normal examination, participants underwent a standardized digital voice recording in a sound-treated audiometric room conducted by an experienced voice-specialized speech-language pathologist, who was blinded to the treatment group of the participant. Acoustic data were collected using a multidimensional voice program (model 4500, version 3.4; Kay Elemetrics) and a VRP program (model 4326, version 3.3; Kay Elemetrics).
Participants were asked to produce the vowel token /a/ 3 times at a comfortable pitch and loudness for 5 seconds. Acoustic analysis (jitter, shimmer, and pitch range) compared before and after treatment was completed on trimmed vowel segments of 1000 milliseconds. Each vowel token was analyzed for fundamental frequency (F0), percentage jitter (JITT), and percentage shimmer (SHIM). Pitch range was recorded (excluding glottal fry): participants were asked to phonate the vowel token /a/, starting at a comfortable pitch, and to glide up to the highest pitch possible and then down to the lowest pitch possible without allowing the voice to become rough or gravely (glottal fry register). A limited VRP was collected by cuing the participants to produce the softest and loudest voice at the following semitone levels: habitual and 20th, 50th, and 80th percentiles of their pitch range. The F0 range and percentile levels are expressed in both hertz and semitones because keyboard cuing is used for the VRP. Participants were asked to produce the vowel token /a/ at the cued pitch level as loud as possible (3 trials), and the result was recorded as maximum intensity in decibels of sound pressure level. For minimal sustainable intensity, the participant was asked to produce the vowel as softly as possible, without whispering. After acoustic data collection, participants used a visual analog scale19 ranging from 1 (least) to 10 (extreme) to describe the effort required to perform the vowel token and the VRP.
To investigate subtle vocal properties, the Self-Administered Vocal Rating (SAVRa)16,20 was performed. The SAVRa is a combination of subjective vocal ratings of perceived effort or discomfort experienced after 4 specific voice tasks.
Four soft vocal tasks (SVTs)16 were requested to be done softly and at a high pitch. For SVT1, the participant was asked to produce and sustain the vowel /ee/ as softly as possible at a comfortable pitch for 5 seconds. For SVT2, the participant was asked to glide from low to high pitch as softly as possible on the vowel /ee/. For SVT3, the participant was asked to say “ee-ee-ee-ee-ee-ee-ee” staccato (very short) and at a high pitch. For SVT4, the participant was asked to sing a few bars of “Happy Birthday” extremely softly and at a high pitch.
Ratings for effort and discomfort were done on a separate visual analog scale19 ranging from 1 (least) to 10 (extreme). The participants were requested to evaluate the ease of all soft voice production tasks, keeping in mind signs of poor voice quality (ie, voice breaks, roughness, delayed onset, instability of pitch or volume, and reduced pitch range).14
After completing the SAVRa, participants then underwent a standardized treatment session (10-15 minutes) with one of us (M.E.H.) using either the external vibration device or the placebo device. The placebo device was a modified version of the external vibrator, with the eccentric weight removed from inside the device, which was replaced with a piece of paper. The placebo device looked and sounded the same as the treatment device. The placebo device was based on a similar device described in the literature.15 Each participant was told that the vibration may be “subthreshold,” which was intended to imply that she may or may not be able to “feel” the vibration. Participants in the study underwent a standardized therapy to 5 neck sites bilaterally (see the Muscle Sites Assessed section below) using the vibratory device (SIRI; LELO) (50-Hz setting) or a modified placebo device. Each site was massaged for 1 minute while the participant lightly hummed a midrange note. Participants were instructed to take a breath whenever it felt comfortable. Each muscle group site was done in the same order and starting on the right side, followed by the left side.
There were 5 muscle sites assessed bilaterally. These included (1) the masseter (the participant was instructed to gently chew with lips closed), (2) the suboccipital area (site of cervical muscle insertion while the participant nodded her head gently), (3) the sternocleidomastoid (midsection), (4) the digastric anterior belly or tongue base (submental and submandibular area), and (5) the paralaryngeal or superolateral thyroid cartilage (strap and constrictor insertion sites).
After the intervention, the participants repeated the same acoustic data collection protocol, the SAVRa, and effort ratings of the tasks. The investigators (M.D. and G.M.) collecting the data were blinded to the group assignment of the participants.
All statistical analyses were performed using a software program (SAS, version 7.1; SAS Institute Inc). Paired-samples t tests18 were used to compare the group means for acoustic parameters. The data sets were judged to be significantly different at 2-sided P < .05. The 95% CIs were calculated for the parameters collected in the study to establish the expected change in before and after treatment measures. To quantify the effect size of treatment (ie, the standardized difference in before and after treatment measures), Cohen d was calculated.21 The effect size of the treatment was considered small at d = 0.2, moderate at d = 0.5, and large at d = 0.8. The data analysis for the EVT group was complete for 14 participants. In the placebo data analysis, one individual had incomplete acoustic data (due to a technical issue) and was excluded, resulting in 12 participants analyzed in that group.
Among 27 female participants (age range, 18-50 years), acoustic parameters analyzed included the mean F0 and perturbation measures JITT and SHIM. Figure 1 shows the flow scheme of all study participants. These parameters were averaged from the 3 vowel token /a/ samples. Pitch range was reported in semitones.
The pretreatment data collected from each group were compared to establish that the participants were similar at baseline within each group. The findings from this analysis (Table 1) revealed that the placebo and EVT groups did not differ on outcome measures. Comparison of the effect size between the groups (ie, Cohen d) indicated a small to moderate difference. Taken together, the similar 95% CIs and Cohen d values indicated that the placebo and EVT groups were similar at the outset of the study.
Within each group, the acoustic data sets were also evaluated before and after treatment by calculating the mean difference after the intervention. A t test (significance level, P < .05) was performed to compare the mean difference between the 2 groups (Table 2). The 95% CI and Cohen d value for each metric were also calculated. In the EVT group, the mean change in F0 was 5.00 Hz (95% CI, −18.10 to 7.35 Hz) compared with the same mean difference in the placebo group of 5.00 Hz (95% CI, −30.30 to 19.20 Hz); however, the latter 95% CI was much wider. Similarly, the 95% CIs were larger for JITT, SHIM, and semitone range in the placebo group than in the EVT group (Figure 2). The JITT increased by 0.23 (95% CI, −0.59 to 0.12) in the EVT group, which remained within the normal range, compared with 0.34 (95% CI, −0.84 to 1.20) in the placebo group (Table 2). In the EVT group, the semitone range demonstrated an increase of 0.40 (95% CI, −1.30 to 0.59) compared with 0.70 (95% CI, −3.20 to 1.90) in the placebo group.
There were 6 voice tasks in which participants rated their vocal effort (Table 2). In the EVT group, there was a reduction in the mean effort rating before and after treatment, with a large effect size observed in the vowel task (d = 0.83) and SVT3 (d = 0.79) and a small effect size observed in VRP (d = 0.32), SVT1 (d = 0.44), and SVT4 (d = 0.32).
In the placebo group (Table 2), a large treatment effect was observed in the vowel task (d = 1.19) and SVT4 (d = 1.11), which implies that, after treatment, the reduction in effort rating for these tasks is estimated at 1 SD from the pretreatment mean. Two other tasks showed a moderate treatment effect (d = 0.64 for VRP and d = 0.69 for SVT1). The SVT2 (d = 0.21) and SVT3 (d = 0.28) had a small effect size.
In the EVT group, the mean change in effort rating after intervention for the vowel task was 0.73 (95% CI, −0.91 to 1.24) similar to the placebo group of 1.17 (95% CI, −0.15 to 1.67). Both groups showed with a significant effect size in this task with a Cohen d score of 0.83 and 1.19, respectively. The effort rating after placebo treatment for the VRP task was found to have a wider 95% CI (−0.31 to 3.05) than the EVT (−0.51 to 1.58). However, the treatment size effect was modest in both with a Cohen d of 0.64 (placebo) and 0.32 (EVT).
The 95% CIs calculated around the mean effort ratings for the remaining 4 tasks (SVT 1-4) were similar with minimal or moderate effect indicated by a Cohen d score below 0.7. The exception was a significant size effect of 1.11 for the SVT4 in the placebo group. These results are shown in Figure 3.
There is convincing evidence that EVT can improve musculoskeletal performance and has been therapeutic in various disorders, including chronic pain conditions. Singers can be viewed as vocal athletes, with maximum performance requirements of intrinsic and extrinsic laryngeal muscles. In singers, vocal fatigue and muscular strain with associated dysphonia are common issues. It has been proposed that external vibration may improve vocal production.10 However, to date, no relevant substantiating reports have been published. In this study, acoustic parameters, including perturbation measures and pitch range, were evaluated in singers after a single EVT session and compared with a placebo group. After EVT, the change in acoustic parameters showed a narrower 95% CI than in the placebo group, indicating that the EVT treatment effect was more consistent than the placebo treatment.
Although the mean SHIM for both groups was in the normal range both before and after treatment, other studies13,22 have demonstrated that perturbation measures improve (decrease) after a mean of 10 minutes of warm-up vocalization. It is possible that the vocalization protocol performed in our study may have been responsible for the improvement in JITT and SHIM in the placebo group and that the addition of EVT altered this effect.
Although a reduction in vocal effort (improvement) after the intervention was observed in both the placebo and EVT groups, the treatment effect was found to be large in the placebo group for different tasks, including the vowel task and SVT4 in the placebo group and SVT1 and SVT3 in the EVT group. The 95% CIs for the change in effort ratings were similar for both groups in most of the SVTs, implying that there is not a consistent meaningful difference in vocal effort after EVT compared with the placebo treatment.
The mechanisms for the reported effort reduction may relate entirely to the treatment protocol, which was based on vocal warm-ups.13,22 The addition of EVT did not influence the improvement in effort or ease of voice production more than the placebo device. Nevertheless, the role of perceived reduced effort may have had an effect on vocal performance from the perspective of a placebo effect.
Possible limiting factors in the study include the sample size (n = 27) given our power analysis requiring 10 participants in each group and the particular parameters chosen. Also, we evaluated these parameters after a single vibration session vs a repeated treatment protocol. The rationale for EVT and its demonstrable benefit in chronic pain conditions may be relevant in a repetitive strain model and may have a role in pain management in voice patients, which is an area for future investigation.
Acoustic metrics were found to be more consistently influenced by EVT than a placebo treatment. However, there was no meaningful difference in the treatment effect of EVT on effort ratings compared with the placebo group.
Accepted for Publication: October 24, 2017.
Corresponding Author: Jennifer Anderson, MD, MSc, FRCS(C), Department of Otolaryngology–Head and Neck Surgery, St Michael’s Hospital, 30 Bond St, Ste 8CC, Toronto, ON M5B 1W8, Canada (firstname.lastname@example.org).
Published Online: January 4, 2018. doi:10.1001/jamaoto.2017.2679
Author Contributions: Dr Anderson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Anderson, DeLuca, Haines.
Acquisition, analysis, or interpretation of data: DeLuca, Merrick.
Drafting of the manuscript: Anderson, Haines.
Critical revision of the manuscript for important intellectual content: All authors.
Obtained funding: Anderson.
Administrative, technical, or material support: Anderson, Merrick, Haines.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest, and none were reported.
Funding/Support: This study was funded by a St Michael’s Hospital Alternate Funding Plan Innovation Grant.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: William To, MSc (St Michael’s Hospital), was the research coordinator on the project. He received no compensation for his contribution beyond the normal course of his employment.