Keywords
VFE - Progressive Masked Voice Exercises - PEEP - SOVTE
Learning Outcomes: As a result of this activity, the reader will be able to:
-
Explain Semi-Occluded Vocal Tract Exercises (SOVTEs) completed with and without Semioccluded Ventilation Masks (SOVMs) and purposes.
-
Explain the Progressive Masked Voice Exercise (PMVE) voice therapy program.
-
Describe the novel phonatory PEEP device and discuss its use in the PMVE voice therapy program.
Voice disorders can be experienced throughout the lifespan and 18 to 30% of adults within the general population currently report a voice disorder (Iliadou et al., 2024; Huston et al., 2024; Roy et al., 2005; Wang et al., 2023). The prevalence of voice disorders rises to 47% in the elderly (Roy et al., 2007), 44% in voice professionals (Oliveira et al., 2022), 46% in singers (Pestana et al., 2017), and 53.5% in children (Bhattacharyya, 2015). Without intervention, voice disorders can hinder livelihoods, impair social participation, and negatively impact quality of life. Evidence-based voice therapies have been found to be beneficial depending on etiologies and contributing factors with some being vocal hygiene and hydration, vocal function exercises, laryngeal manual therapy, and respiratory exercises (Desjardins et al., 2017; Garabet et al., 2024; Van Stan et al., 2015).
Typical evidence-based voice therapies, the Semi-Occluded Vocal Tract Exercises (SOVTEs), incorporate a wide variety of methods including lip and tongue trills, raspberries, humming, straw phonation, water-resistant resonant tube phonation, voiced fricatives, and cup phonation (Calvache et al., 2020; Cox & Titze, 2023; Hijleh & Pinto, 2021; Kapsner-Smith et al., 2015; Nam et al., 2019; Story et al., 2000). Performers and clinicians incorporate them based on levels of resistance and steady versus fluctuating characteristics (Andrade et al., 2014; Titze, 2006). Phonatory resistance and perceptions of increased back pressure prompt matching laryngeal impedance resulting in greater vocal economy and efficiency (Titze, 2006). Thus, the vocal mechanism can be driven harder without harmful effects by capitalizing on mechano-acoustic source–filter interactions. Favorable vocal tract configurations and effects include epilaryngeal tube narrowing, increased supraglottal and intraglottal air pressure, decreased vibratory amplitude and glottal separation, reduced phonatory threshold pressure, and diminished sound emission (Titze, 2009). Vocal Function Exercises (VFE; Stemple et al., 1994), a type of SOVTE program, have also been widely studied and found to be effective on acoustic, auditory-perceptual, and laryngeal function measures in persons with voice disorders (Angadi et al., 2019; Bane et al., 2019; Guzman et al., 2013; Kaneko et al., 2015; Lim et al., 2009; Pedrosa et al., 2016; Roy et al., 2001). Improvements have also been noted in the self-reported Voice Handicap Index-10 (VHI-10) (Jafari et al., 2017; Kaneko et al., 2015; Lim et al., 2009; Pedrosa et al., 2016) and structurally, and physiologically during videolaryngostroboscopic (VLS) examinations (Kaneko et al., 2015). A review by Barsties von Latoszek et al. (2023b) has also found the effects of VFE on aerodynamic variables such as maximum phonation time (MPT). VFEs are designed to realign the interconnected contributions of the subsystems of respiration, phonation, and resonance to optimize function and quality utilizing auditory, proprioceptive, and kinesthetic feedback (Roy et al., 2001; Stemple et al., 2020). A key limitation of VFEs and other types of SOVTEs, however, is that productions are limited to nonspeech tasks that vary in duration, pitch, and loudness.
SOVTEs have also been done in the presence of semioccluded ventilation masks (SOVMs). The use of SOVTEs coupled with SOVMs has grown and multiple variations have emerged (Awan et al., 2019; Fantini et al., 2017; Frisancho et al., 2020; Gartner-Schmidt et al., 2022; Gillespie et al., 2022; Guzman et al., 2023; Kissel et al., 2023; Meerschman et al., 2020). In addition to the partial occlusions created by the lips and tongue, SOVMs can be placed over the mouth and nose and semioccluded with various items including the hands, plastic plugs with variably sized openings, and water-resistant resonance tubes. In addition to the typical nonspeech SOVTEs, SOVMs allow for connected speech and singing due to this external positioning (Awan et al., 2019; Fantini et al., 2017; Gillespie et al., 2022; Meerschman et al., 2020).
The investigators of this current study have noted positive clinical results using an adjustable positive end-expiratory pressure or PEEP valve fitted into the aperture of an SOVM when coupled with a newly conceptualized clinical voice therapy protocol known as the Progressive Masked Vocal Exercises (PMVE) program. Improved acoustic and perceptual clinical outcomes have been anecdotally seen in patients with dysphonia using this accessible device such as greater ease of voicing, and improved voice quality (Gartner-Schmidt et al., 2022). This novel approach offers potential advantages similar to the SOVTEs including overtraining effects of airflow resistance and laryngeal impedance, incremental adjustments to resistance and impedance via titrations of PEEP valve settings, direct or specific training of the speaking and singing voice, and more complete integration of the lungs into the voicing mechanism (Titze, 2006). This contrasts with vocal hygiene programs that reduce vocal load (Roy et al., 2001) or other therapy programs that target the physiological balance of voice subsystems (Angadi, et al., 2019; Stemple et al., 1994, Stemple et al., 2020; Van Stan et al., 2015; Van Stan et al., 2021). To date, an adjustable PEEP valve has not been combined with a SOVM protocol that includes progressive resistance. Therefore, the purpose of this study was to compare the effectiveness of the PMVE therapy program using an innovative phonatory PEEP device to Stemple's VFE program on acoustic, auditory-perceptual, aerodynamic, and self-report outcomes measures. Due to the established efficacy of the VFE program, it was hypothesized that the objective acoustic, aerodynamic, subjective auditory-perceptual, and self-report outcomes of the PMVE therapy program would be similar to the VFE program after receiving therapy.
Methods
Participants
Recruitment spanned from May to September 2020, following the coronavirus-2019 pandemic, at two ENT clinics near Salt Lake City, Utah, during routine clinical evaluations. Patients 18 years or older who were diagnosed with a voice disorder by their physician and the voice care team in the clinic were assessed for eligibility. Individuals were excluded if they were judged to be inappropriate for the voice therapy conducted in this study, already receiving office-based therapy, reporting unresolved episodes of syncope, or previously diagnosed with a neurodegenerative disease. In addition to chart reviews and patient interviews, evaluation procedures included VLS, aerodynamic, recorded samples for computerized acoustic analyses, and auditory-perceptual judgments. Recruitment efforts yielded 25 enrolled participants who performed the baseline assessments. Once enrolled, a coin flip was used only to determine the placement of the first participant into their group (heads = PMVE; tails = VFE), following which the rest of the participants were alternately assigned into either the PMVE or VFE group. See [Fig. 1] for the flow of participants within the study. Nine drop-outs were noted, and the final data analysis included eight individuals in each group.
Figure 1 Study flow cart. PMVE, Progressive Masked Voice Exercise Program; VFE, Vocal Function Exercise.
Procedure
The Rocky Mountain University (RMU) Institutional Review Board approved this study (2003024-02), and all enrolled patients gave informed consent. Pre- and post-data were collected on acoustic, aerodynamic, perceptual, and self-report measures in the clinic at the time of initial diagnosis visit and after completion of 6 weeks of home-based therapy. See [Fig. 1] for the study flow. Participants were not financially compensated and received instructions for either the PMVE or VFE home-based trainings at the initial visit. The independent home-based delivery model was adopted to reduce transmission of the SARS-CoV-2 virus (Castillo-Allendes et al., 2021). Participants were blinded on their group assignments, while the assessors and treating clinicians were not blinded. This study utilized a within- and between-subject repeated measures experimental design.
Outcome Measures
Recording and Processing of the Audio Samples
For acoustic analysis, recordings were captured using a Blue USB Yeti desktop condenser microphone (Baltic Latvian Universal Electronics, LLC) with a 20-Hz to 20-kHz frequency response range at a sampling rate of 44,100 Hz. A cardioid pick-up pattern with no gain was used. It was positioned in clinic rooms 10 inches in front of the participants, secured, and tilted in its tabletop stand. It was connected to a Dell 3590 laptop Intel Core i5-8250U CPU at 1.60 to 1.80 GHz with a Windows 10 Pro operating system, and recordings were saved in the “.wav” format. The noise was minimized by not bumping or using the microphone stand/table and avoiding rooms with elevated nonstationary noise such as hallway chatter, running fans, or buzzing fluorescent lights. The recorded samples consisted of the all-voiced sentence, “We were away a year ago” (Kempster et al., 2009) for continuous speech (CS), and a 5-second sustained “ah” for sustained voice (SV). The sentence and sustained “ah” were recorded altogether with a 2-second silent pause between them. Participants were instructed to speak in their everyday voice and loudness.
Recordings were securely transferred and saved on the secure network at the Center for Communication Disorders (CCD) at RMU. Acoustic editing and phonatory analysis were conducted by author P.R.S. using a Microsoft Surface with an 11th Gen Intel Core i7-118G7 at 3.00 GHz and a 64-bit operating system. For acoustic edits and analysis, the freeware of Audacity (version 3.4.2; https://www.audacityteam.org/), Praat (version 6.2.23; www.praat.org), and VOXplot (version v2.0.1; Lingphon, Straubenhardt, Germany; ttps://voxplot.lingphon.com) were used. Edited acoustic samples were also analyzed using the Analysis of Dysphonia in Speech and Voice (ADSV; Model 5109, KayPentax). See [Table 1] for the acoustic measures.
Table 1
Summary of the acoustic measures and associated norms in this study
|
Measure
|
Unit
|
Norm
|
Interpretation/Definition
|
Program
|
|
HNR
|
Decibels (dB)
|
≥ 23.24
|
Higher = better voice quality.
A ratio of the voice signal to aperiodic noise within the voice
|
VOXplot
|
|
(Jitter) PPQ5
|
Percent (%)
|
≤ 0.29
|
Lower = better voice quality.
A measure of frequency stability
|
VOXplot
|
|
AVQI
|
Integers (+/−)
|
≤ 1.17
|
Lower = better voice quality.
A measure of calculated voice quality derived from acoustic characteristics
|
VOXplot
|
|
GNE
|
Ratio
|
≥ 0.89
|
Higher = better voice quality.
A ratio of the glottal voice signal to glottal noise
|
VOXplot
|
|
ABI
|
Integers (+/−
|
≤ 2.35
|
Lower = better voice quality.
A multiparameter measure that indexes the presence vs. absence and severity of breathiness
|
VOXplot
|
|
CPPS
|
Decibels (dB)
|
≥ 14.47
|
Higher = better voice quality.
A harmonic measure of the peaks in the cepstrum with the highest peak typically coinciding with the f
o
|
VOXplot
|
|
CSID
|
Integers (+/−)
|
≤ 24
|
Lower = better voice quality.
A multiparameter measure that indexes the presence vs. absence and severity level of dysphonia
|
ADSV
|
Abbreviations: ABI, Acoustic Breathiness Index; ADSV, analysis of dysphonia in speech and voice; AVQI, Acoustic Voice Quality Index; CPPS, Smoothed Cepstral Peak Prominence; CSID, Cepstral Spectral Index of Dysphonia; f
o, fundamental frequency; GNE, glottal-to-noise excitation ratio; HNR, harmonics-to-noise ratio; PPQ5, Jitter Pitch-Perturbation Quotient with a smoothing factor of 5 periods; VOXplot, Acoustic Voice Quality Analysis Program.
Signal-to-Noise Ratio Processing
Each sample was edited with Audacity to ensure an acceptable >30-dB signal-to-noise ratio (SNR; Deliyski et al., 2005) and to conduct uniform clippings of samples. The “Noise Reduction” tool within the “Effect” tab minimized ambient noise within each recording. Settings were optimized for the most noise removal without signal distortion or loss (Gailey, 2015), as instructed in the manual (https://manual.audacityteam.org/man/noise_reduction.html). Portions of waveforms without voice (noise only) were selected, and the noise profile was obtained. The entire sample was then selected, and the noise profile was applied. For each sample, the amount of volume reduction was 12 dB, sensitivity was at 6 (default), and the frequency smoothing (bands) value was at 0. Edited samples were exported from Audacity as “.wav” files at a sampling rate of 44,100 Hz and with signed 24-bit PCM encoding. Praat was then used on eight randomly selected pre- and post-edited SNRs based on similar methods employed by Pierce et al., 2021). The difference in the noise and signal intensities (dB) was calculated in the selected pre- and post-edited samples. Some pre-edited SNRs were below the recommended norm, ranging from 19.89 to 30.98 dB, with a mean and standard deviation of 26.57 dB (4.23). After the Audacity edits, SNRs were near or above the norm, ranging from 30.57 to 41.59 dB with a mean and standard deviation of 36.87 dB (4.18). See [Fig. 2] for the pre–post SNR processing summary.
Figure 2 Summary of signal-to-noise ratios pre- and post-acoustic processing summary. S/N, signal-to-noise ratio.
Acoustic and Perceptual Measures
Once edited, samples were uniformly clipped and separated into individual CS and SV files. All noise (non-signal) segments were removed from each sample. Additionally, the middle 3 seconds were uniformly clipped from the sustained vowels to avoid irregular onset/offset artifacts (Olszewski et al., 2011). The uniformly edited and clipped CS and SV samples were entered into VOXplot simultaneously per participant. Measures computed were the multiparametric indices of the Acoustic Voice Quality Index (AVQI) (Maryn et al., 2014, 2010) and the Acoustic Breathiness Index (ABI) and their contributing measures including the harmonics-to-noise ratio (HNR; dB), jitter of the five-point perturbation quotient (PPQ5; %), glottal-to-noise-excitation (GNE) ratio, and the smoothed cepstral peak prominence (CPPS; dB) (Barsties von Latoszek et al., 2023a). The ADSV computed the Cepstral Spectral Index of Dysphonia (CSID) for the same sustained vowel (CSIDSV; Awan & Roy, 2009; Awan et al., 2016) and all voiced CAPE-V sentence samples (CSIDCS; Awan et al., 2009; Awan et al., 2016; Watts, 2015). See [Table 1] for a summary of the acoustic measures.
Perceptually, two non-study volunteer expert listeners familiar with the GRBASI scale (Dejonckere et al., 1996; Hirano, 1981; Yamauchi et al., 2010) were asked to rate each sample used for the acoustic analysis over a 2.5-hour session in the clinic. They were faculty-certified speech-language pathologists (SLPs) from RMU with extensive academic and clinical experience in treating voice disorders and were blinded to all study aspects. To calibrate, each expert listener independently rated six nonstudy recordings of dysphonic speakers. They listened to each sample three times with one additional playback if requested and rated each parameter (G = grade, R = roughness, B = breathiness, A = asthenia, S = strain, and I = instability) and severity (0 = normal, 1 = mild, 2 = moderate, and 3 = severe). Once rated, they compared responses. Discussion ensued for disagreements until a consensus was reached; no rating differed more than one point. This process was maintained for each recording of the actual study samples. The expert listeners were positioned 6 feet apart and 8 feet from the facing loudspeaker. Samples were randomized and played at a comfortable volume as indicated by the listeners, alternating between CS and SV productions without reference to time, treatment, or participants.
Aerodynamic Measure
The MPT, an objective aerodynamic assessment with strong evidence base in voice therapy, gauges glottal efficiency based on the vibratory characteristics of the true vocal folds set into motion by the exhalatory pulmonary airstream (Solomon et al., 2000). It is measured in seconds whereby, after a maximal inhalation, a sustained /a/ vowel is held until breath runs out. The current participants performed MPT as directed in their everyday voice and loudness. The VFE program has been previously found to be beneficial in improving MPT (Barsties von Latoszek et al., 2023b), and voice therapy is deemed to be effective with an MPT increase of 1.41 seconds from pretherapy levels (Rodríguez-Parra et al., 2009).
Self-Report Measures
Two self-report assessments including the VHI-10 (Rosen et al., 2004) and an informal five-item voice questionnaire arranged by author M.K.F. during the clinic sessions were utilized. The VHI-10 consists of 10 items assessing perceptions of the recent functional, physical, and emotional aspects of voice. Each response is indicated on a scale of 0 to 4, where 0 = never, 1 = almost never, 2 = sometimes, 3 = almost always, and 4 = always. The total score is calculated, with higher scores representing greater difficulties. According to Arffa et al. (2012), scores of 11 or higher should be considered abnormal. The novel, five-item questionnaire, with dichotomous 1 = yes, 2 = no responses, gauges perceptions about how the voice works and if the function is satisfactory to daily life, settings, and situations. As shown in [Fig. 3], “yes” responses are favorable except for item 3.
Figure 3 Novel intake/exit voice quality and use questionnaire.
Training Interventions
After completing the baseline assessments, an in-person 90-minute study-related education and training session (PMVE or VFE) was conducted by the authors M.K.F. and S.P. during the first week of participation. Both authors M.K.F. and S.P. are licensed SLPs with 13 years of combined experience in the specialty of voice disorders. Each participant, first, regardless of their assigned therapy group, received education on the structures and functions of voice production to raise awareness about the vocal folds, the importance of breath support, and the best laryngeal hygiene practices. Resources used to raise awareness included models, handouts, demonstrations, images, and videos, including their LVS examinations. General understanding was demonstrated via accurate paraphrasing of the material. Participants were then trained to accurately complete the home-based PMVE or VFE treatment programs during the 90-minute session. See [Fig. 4] for comparative details of each therapy program. See Appendix A for a full description of the PMVE program. Both groups were provided a written guidebook containing their exercise protocol, and the required dosages, each program including four different exercises to be performed twice a day for a total of 6 weeks (see [Fig. 4]). Learner-centered rapid cycle deliberate practice (RCDP) methods (Perretta et al., 2020) were incorporated to enhance the acquisition and retention of the protocol and exercises, a crucial factor for increased fidelity of home-based completion of the therapies. Guided practice included the identification of errors and deficiencies and repetitions with visual and verbal cues to redirect and shape accurate techniques and performances provided by the trainers (Chancey et al., 2019). At the end of the 90-minute training session, the participants were able to perform their entire program accurately and independently without corrective feedback, suggesting a successful transition to the 6-week home-based therapy. The 6-week VFE therapy has been previously recommended by Roy et al. (2001), Sauder et al. (2010), Stemple (2005), and Stemple et al. (1994) and was utilized and compared with PMVE as the treatment duration. Participants were instructed to track sessions via daily treatment logs and were sent weekly emails to remind them of the training steps and instructions. Participants were also instructed to contact the study team as needed. At the end of the 6-week home-based therapy, each participant was asked to return to the clinic for a follow-up visit with their physician and voice care team to complete the post-assessments.
Figure 4 Side-by-side comparison of voice therapies. PMVE, Progressive Masked Voice Exercise Program; VFE, Vocal Function Exercise Program.
Vocal Function Exercises Treatment Program
The VFE program was advanced by Dr. Joseph Stemple and has been extensively described in the previous literature (Roy et al., 2001; Sauder et al., 2010; Stemple, 2005; Stemple et al., 1994, 2020). As participants learned to perform the four exercises, emphasis was placed on achieving facial mask (the anatomical triangle with the apex at the bridge of the nose, the sides running along the nasolabial folds, and the base at the chin) vibrotactile feedback during vocal productions as a token of accurately focused resonance while also maintaining appropriate lip approximation and protrusion. To facilitate the accuracy of target frequency productions by the participants, lead investigator M.K.F. recorded 5-second audio samples that were representative of the prescribed pitch targets. Sample recordings were also captured with the use of a pitch pipe. All recordings were confirmed to be within 5 Hz of the prescribed frequencies. These recordings were played and practiced during the training sessions and then emailed to participants to facilitate appropriate home practice.
Progressive Masked Vocal Exercises Program
Participants in the PMVE group were trained according to the new standard protocol developed by lead investigator M.K.F. The voicing tasks are completed across progressively increasing resistance settings on the phonatory PEEP device that presumably increases laryngeal impedance incrementally limiting risk to voice quality or even the voicing mechanism itself. It is a top-down therapy, meaning conversational voice and/or singing voice is targeted from the outset presumably benefitting voice improvements in higher levels of discourse (Gartner-Schmidt et al., 2022). This is different from a bottom-up approach, where attempts would be made to gradually transition to higher levels of discourse from non-speech phonatory tasks. By way of analogy, this is like preferentially running long distances to build endurance when training for participation in a marathon race rather than prioritizing weight training sessions, useful as those might be to a well-rounded athlete. The VFE and PMVE protocols are further described in [Fig. 4], and Appendix A.
Data Analysis
All analyses were conducted with parametric statistics for variables that met assumptions of normality, and with nonparametric statistics for variables that violated the assumptions. Baseline demographics were compared between the therapy groups (PMVE vs. VFE) prior to the treatment to determine differences at baseline using independent t-tests (for age, parametric acoustic outcome measures, and aerodynamic measure of MPT), Mann–Whitney U-tests (for nonparametric acoustic, perceptual, and self-reported outcome measures), and Fisher's exact tests (for gender). For normal acoustic variables and aerodynamic measure of MPT, two-way, repeated measures analysis of variance (ANOVA) analyzed the effects of the training program with two time points (pre- and post-intervention) × 2 groups (PMVE vs. VFE), and eta-squared partial calculated the effect size (ES). ES of 0.1 to 0.24 indicated a small effect, 0.25 to 0.36 indicated a medium effect, and ≥0.37 indicated a large effect. Additionally, post hoc tests were performed to determine differences for main effects. For non-normal acoustic variables and voice and VHI questionnaires, two-tailed Mann–Whitney U-tests were used to compare the change scores between the PMVE and the VFE groups. Additionally, two-tailed Wilcoxon's signed-rank tests were performed to determine the pre- versus post–within-group therapy differences in each group individually. For auditory-perceptual variables, Fisher's exact tests were performed for between group (PMVE vs VFE) post-intervention, and for pre- versus post–within-group therapy differences in each group individually. All statistical tests were performed using the Intellectus Statistics software 2023 version (Palm Harbor, Florida). The level of significance (alpha) was set at 0.05, where a p< 0.05 was considered statistically significant.
Results
Participants in PMVE and VFE groups were similar in baseline characteristics (please refer to [Table 2]).
Table 2
Baseline (initial visit) demographic characteristics and comparisons between the participants
|
Sample characteristics
|
PMVE
n = 8
|
VFE
n = 8
|
p-Value
|
|
Gender
|
|
Male (n)
|
3
|
2
|
NS
|
|
Female (n)
|
5
|
6
|
NS
|
|
Age [mean (SD)]
|
51.63 (16.9)
|
48.63 (13.3)
|
NS
|
|
Laryngostroboscopic diagnosis
|
|
MTD (n)
|
3
|
4
|
–
|
|
VF nodules (n)
|
0
|
1
|
–
|
|
VF paresis (n)
|
3
|
2
|
–
|
|
VF paralysis (n)
|
2
|
1
|
–
|
|
Acoustic variables
|
|
HNRdB [mean (SD)]
|
16.96 (6.99)
|
19.81 (5.51)
|
NS
|
|
PPQ5 [mean (SD)]
|
0.46 (0.48)
|
0.35 (0.22)
|
NS
|
|
AVQI [mean (SD)]
|
5.12 (2.80)
|
4.18 (2.12)
|
NS
|
|
GNE [mean (SD)]
|
0.54 (0.19)
|
0.63 (0.19)
|
NS
|
|
CPPS [mean (SD)]
|
9.72 (4.75)
|
11.46 (3.02)
|
NS
|
|
ABI [mean (SD)]
|
6.20 (2.37)
|
5.50 (2.22)
|
NS
|
|
CSIDCS [mean (SD)]
|
25.15 (23.37)
|
19.50 (12.35)
|
NS
|
|
CSIDSV [mean (SD)]
|
56.96 (28.36)
|
40.87 (27.13)
|
NS
|
|
Auditory-perceptual variables GRBASI scale [mean rank]
|
|
Grade
|
8.31
|
8.69
|
NS
|
|
Roughness
|
7.25
|
9.75
|
NS
|
|
Breathiness
|
9.75
|
7.25
|
NS
|
|
Asthenia
|
9.88
|
7.12
|
NS
|
|
Strain
|
10.06
|
6.94
|
NS
|
|
Instability
|
8.31
|
8.69
|
NS
|
|
Aerodynamic variable
MPT [mean (SD)]
|
11.51 (4.84)
|
11.85 (3.74)
|
NS
|
|
Voice questionnaire [mean rank]
|
9.31
|
7.69
|
NS
|
|
VHI-10 [mean rank]
|
8.25
|
8.75
|
NS
|
Abbreviations: ABI, Acoustic Breathiness Index; AVQI, Acoustic Voice Quality Index; CPPS, Smoothed Cepstral Peak Prominence (dB); CSIDCS, Cepstral Spectral Index of Dysphonia (connected speech); CSIDSV, Cepstral Spectral Index of Dysphonia (sustained vowel); GNE, glottal-to-noise excitation ratio; HNRDB, harmonics-to-noise ratio (dB); MTD, muscle tension dysphonia; N, number of participants; NS, not significant; SD, standard deviation; VF, vocal fold; VHI, Voice Handicap Index-10.
For mean scores of acoustic variables including HNRDB, AVQI, GNE, ABI, CSIDCS, and CSIDSV, no interaction group × time effects and the main effect of the group were observed. There was a main effect of time in all the variables except GNE (p < 0.05), showing a significant improvement from the pre- to post-training program. The PMVE group showed pre- to post-training improvements in HNRDB and CSIDSV, while the VFE group showed pre- to post-training improvements in AVQI, ABI, and CSIDCS variables (p < 0.05). Please refer to [Table 3] for details.
Table 3
Between-group (PMVE vs. VFE) differences for study variables that meet parametric assumptions
|
Acoustic variables
|
Group
|
Pre-training
[mean (SD)]
|
Post-training
[mean (SD)]
|
Time effect
[p-value (ES)]
|
Group effect
[p-value (ES)]
|
Time × group effect
[p-value (ES)]
|
Pre- to post-training differences within each group
[p-value]
|
|
HNRdB
|
PMVE
VFE
|
16.96 (6.99)
19.81 (5.51)
|
27.74 (10.95)
23.75 (5.40)
|
0.01 (0.39)[a]
|
0.85 (0.00)
|
0.19 (0.12)
|
0.008[a]
0.28
|
|
AVQI
|
PMVE
VFE
|
5.12 (2.80)
4.81 (2.12)
|
3.83 (1.48)
2.66 (1.17)
|
0.009 (0.40)[a]
|
0.25 (0.09)
|
0.80 (0.00)
|
0.07
0.03[a]
|
|
GNE
|
PMVE
VFE
|
0.54 (0.19)
0.63 (0.19)
|
0.62 (0.14)
0.70 (0.15)
|
0.08 (0.20)
|
0.28 (0.08)
|
0.83 (0.00)
|
–
–
|
|
ABI
|
PMVE
VFE
|
6.20 (2.37)
5.50 (2.22)
|
5.71 (1.64)
4.22 (1.27)
|
0.04 (0.26)[a]
|
0.23 (0.10)
|
0.33 (0.07)
|
0.40
0.04[a]
|
|
CSIDCS
|
PMVE
VFE
|
25.15 (23.37)
19.50 (12.35)
|
18.55 (19.82)
6.33 (6.01)
|
0.007 (0.42)[a]
|
0.27 (0.09)
|
0.31 (0.07)
|
0.15
0.01[a]
|
|
CSIDSV
|
PMVE
VFE
|
56.96 (28.36)
40.47 (27.13)
|
37.79 (16.50)
26.97 (24.84)
|
0.006 (0.43)[a]
|
0.25 (0.09)
|
0.61 (0.02)
|
0.02[a]
0.07
|
|
Aerodynamic variable
|
|
MPT
|
PMVE
VFE
|
11.51 (4.84)
11.85 (3.74)
|
12.99 (4.41)
13.57 (4.39)
|
0.098 (0.18)
|
0.818 (0.004)
|
0.894 (0.001)
|
0.181
0.290
|
Abbreviations: ABI, Acoustic Breathiness Index; AVQI, Acoustic Voice Quality Index; CSIDCS, Cepstral Spectral Index of Dysphonia (connected speech); CSIDSV, Cepstral Spectral Index of Dysphonia (sustained vowel); GNE, glottal-to-noise excitation ratio; MPT, maximum phonation time; PMVE, Progressive Masked Voice Exercises; SD, standard deviation; VFE, vocal function exercises.
a Indicates p-value significant at 0.05.
For median scores of acoustic variables including PPQ5 and CPPS, no between-group differences were observed. The VFE group showed pre- to post-training improvements in PPQ5 (p < 0.05). Please refer to [Table 4] for details.
Table 4
Between-group (PMVE vs. VFE) differences for study variables that do not meet parametric assumptions
|
Variables
|
Group
|
Pre-training
[median or frequency “n”]
|
Post-training
[median or frequency “n”]
|
[∂, mean rank]
|
Between group
p-Value
|
Within group
p-Value
|
|
Acoustic variables (MDN)
|
|
PPQ5
|
PMVE
VFE
|
0.20
0.32
|
0.17
0.11
|
8.88
8.12
|
0.75
|
0.12
0.05
a
|
|
CPPS
|
PMVE
VFE
|
9.25
11.29
|
12.07
13.27
|
9.12
7.88
|
0.64
|
0.09
0.07
|
|
Auditory-Perceptual variables GRBASI Scale (postcomparison frequencies only)
|
PMVE
|
1/2/5/0
|
0/4/4/0
|
–
|
0.29
|
1.00
|
|
Grade (0–3)
|
VFE
|
0/3/5/0
|
1/6/1/0
|
–
|
|
0.64
|
|
Roughness (0–3)
|
PMVE
VFE
|
1/3/3/1
0/2/4/2
|
0/3/5/0
2/4/2/0
|
–
–
|
0.29
|
1.00
1.00
|
|
Breathiness (0–3)
|
PMVE
VFE
|
3/2/3/0
5/2/1/0
|
5/3/0/0
5/3/0/0
|
–
–
|
1.00
|
0.68
0.64
|
|
Asthenia (0–3)
|
PMVE
VFE
|
4/2/2/0
7/0/1/0
|
4/4/0/0
5/3/0/0
|
–
–
|
1.00
|
0.66
0.37
|
|
Strain (0–3)
|
PMVE
VFE
|
1/3/4/0
2/5/1/0
|
0/5/3/0
2/6/0/0
|
–
–
|
0.10
|
0.14
0.04a
|
|
Instability (0–3)
|
PMVE
VFE
|
1/2/5/0
0/3/5/0
|
0/5/3/0
1/7/0/0
|
–
–
|
0.20
|
1.00
0.37
|
|
Voice questionnaire
|
PMVE
VFE
|
2.00
1.00
|
2.50
3.00
|
6.62
10.38
|
0.11
|
0.46
0.03a
|
|
VHI-10
|
PMVE
VFE
|
14.50
21.00
|
11.50
16.00
|
9.69
7.31
|
0.32
|
0.57
0.03a
|
Abbreviations: ABI, Acoustic Breathiness Index; CPPS, Smoothed Cepstral Peak Prominence (dB); CSIDCS, Cepstral Spectral Index of Dysphonia (connected speech); CSIDSV, Cepstral Spectral Index of Dysphonia (sustained vowel); MTD, muscle tension dysphonia; n, number of participants; NS, not significant, ∂, delta or (post-mean − pre-mean); PPQ5, Pitch-Perturbation Quotient with a smoothing factor of 5 (Jitter ppq5%); SD, standard deviation; VF, vocal fold; VHI, Voice Handicap Index-10.
Notes: aIndicates p-value significant at 0.05.
For median scores of voice and VHI questionnaires, no between-group differences were observed. The VFE group, however, showed pre- to post-training improvements in both variables (p < 0.05). Please refer to [Table 4] for details.
For auditory-perceptual variables including using the GRBASI scale, no between-group differences at posttreatment time points were observed. The VFE group showed pre-to-post training improvements in the strain variable (p < 0.05). Please refer to [Table 4] for details.
For mean scores of aerodynamic measures of MPT, no interaction group × time effects and the main effect of group and time were observed. The PMVE and VFE groups showed no pre-to-post training improvements in MPT (p < 0.05). Please refer to [Table 3] for details.
Discussion
We are unaware of any established SOVM program such as the PMVE that includes progressive use of resistance and laryngeal impedance as a primary clinical voice therapy approach using an adjustable PEEP valve. Thus, the purpose of this preliminary investigation was to establish efficacy of the innovative phonatory PEEP device and the PMVE voice therapy program and compare outcomes with the well-established, evidence-based VFE program (Angadi et al., 2019; Guzman et al., 2013; Jafari et al., 2017; Kaneko et al., 2015; Lim et al., 2009; Pedrosa et al., 2016; Roy et al., 2001; Stemple et al., 1994). The question addressed was, “Do acoustic, aerodynamic, perceptual, and self-report measures of both the PMVE and VFE programs differ significantly after therapy?” Our hypothesis was that the objective acoustic and aerodynamic measures and the subjective auditory-perceptual and self-report outcomes of the PMVE therapy program would be similar to the VFE program after receiving therapy. The results of this study suggest that there were no differences in the acoustic, aerodynamic, perceptual, and self-report measures between the PMVE and VFE programs after therapy, suggesting that both treatment therapies worked. When examined individually, the VFE program resulted in a wide variety of improvements posttherapy, while the PMVE program solely resulted in acoustic improvements posttherapy.
There are several reasons for these findings. The similarity of participants' responses for both the VFE and PMVE therapy programs is not surprising because both utilized SOVTEs which are commonly associated with beneficial mechano-acoustic, source–filter interactions and aerodynamic characteristics of voice production (Cox & Titze, 2023; Titze, 2006; Titze, 2009). These common effects have been demonstrated across a wide variety of traditional methods, materials, and protocols used for SOVTEs and with SOVTEs more recently achieved with SOVMs (Angadi et al., 2019; Awan et al., 2019; Calvache et al., 2020; Cox & Titze, 2023; Fantini et al., 2017; Frisancho et al., 2020; Gartner-Schmidt et al., 2022; Gillespie et al., 2022; Guzman et al., 2023; Hijleh & Pinto, 2021; Kapsner-Smith et al., 2015; Kissel et al., 2023; Meerschman et al., 2020; Nam et al., 2019; Story et al., 2000). Findings by Awan et al. (2019), Gartner-Schmidt et al. (2022), and Gillespie et al. (2022), can also offer insights due to their comparative study on variably occluded face masks (VOFMs) on aerodynamic and acoustic characteristics in people with and without voice disorders when producing syllables and sentences. For aerodynamics, they observed reductions in subglottal air pressure that presumably optimized phonatory airflow and decreased glottal resistance. Acoustically, favorable changes were also observed for the CPP and the CSID. These findings suggest a potential link between favorable airflow dynamics and favorable acoustic characteristics and suggest that SOVTEs of all types, including those incorporated into the PMVE therapy program utilizing the novel phonatory PEEP device, presumably impact vocal tract airflow dynamics and acoustic characteristics of voice production positively. Interestingly, the authors suggested that since favorable changes were realized for each variably sized outlet, it may be beneficial to have people with various voice disorders complete tasks across the hierarchy of opening sizes from smallest (most resistance) to largest and eventually without partial occlusion (least resistance) resembling a context of connected speech and modify task durations to determine maximal benefits (Gillespie et al., 2022). This gets at the progressive nature of the phonatory PEEP device in conjunction with the PMVE therapy program.
When we examine the within-group differences, PMVE only resulted in acoustic improvements, while VFE resulted in improvements across acoustic, perceptual, and participant self-report measures. These findings are in line with the extensive evidence of the VFE program typically noting improvements across multiple markers including laryngeal functions, movements and appearances, acoustic and aerodynamic measures, auditory-perceptual ratings, and self-report questionnaires (Angadi et al., 2019; Bane et al., 2019; Barsties von Latoszek et al., 2023a, b; Guzman et al., 2013; Kaneko et al., 2015; Lim et al., 2009; Pedrosa et al., 2016; Roy et al., 2001). This evidence supports the notion that the VFE program realigns the interconnected contributions of respiration, phonation, and resonance to optimize function and quality by utilizing auditory, proprioceptive, and kinesthetic feedback (Roy et al., 2001; Stemple et al., 2020). Although MPT outcomes were not significantly different from pre-versus-post therapy levels for both the PMVE and VFE groups, the mean MPT duration increased by greater than 1.41 seconds for both therapies. This suggests that both the PMVE and VFE therapies could be clinically effective (Rodríguez-Parra et al., 2009).
This is the first study to incorporate the novel phonatory PEEP device into the novel PMVE therapy program. Mughal et al. (2005) defined PEEP as pressure in the alveoli of the lungs that is greater than atmospheric pressure as measured at the end of exhalation. However, an adjustable PEEP valve in conjunction with an SOVM was initially considered by the investigators clinically because the valve provides rapid and convenient advancement of resistance and matching levels of impedance (Titze, 2009) without requiring a transition to another device or protocol during SOVM phonation. This phonatory PEEP, as we have termed it, allows clients and clinicians effective and efficient control of vocal exercises as well as the desired titration of airway resistance (Awan et al., 2019; Gillespie et al., 2022), including overtraining in the form of progressive overload, a strategy familiar to other medical contexts (Todd et al., 2012). In the context of mechanical ventilation, PEEP enhances the recruitment of alveoli (Acosta et al., 2007), increases lung volume and transpulmonary pressure, improves oxygenation, and decreases the work of breathing (Smith & Marini, 1988). Furthermore, PEEP results in backpressure in the presence of airflow (exhalation; Dick et al., 1978), an important effect where phonatory SOVM tasks are concerned (Titze, 2006). Importantly, these valves are not simply expiratory resistance devices, as impedance with varied levels of resistance is produced independently of airflow (Blanch, et al., 2005). The effects of PEEP on voicing and outside of the context of mechanical ventilation are currently unknown. But if the widely accepted understanding remains that the respiratory system acts as the power source for phonatory vibration, it stands to reason that functionally increasing the capacity of this same power source through alveolar ventilation or recruitment might be expected to significantly impact the functional behavior of other voicing mechanism components or even to increase voice quality and function overall. In the schema presented here, the components of the respiratory system including the lungs are viewed as investigative targets of interest given the potential to have global impacts on both vocal function and voice quality (Titze, 2006). More data are needed to investigate lung function before and after PMVE practice with the phonatory PEEP device to examine its impact on phonation. The PEEP valves utilized in this study can be set to 22 to 23 cm H2O. As all participants started the protocol at 7 to 8 cm H2O initial practice level (IPL) based on their ability to maintain voicing and tolerate the treatment, none reached this maximum level. Future studies may reasonably aim to determine if higher resistance levels provide significant clinical benefits though maximizing resistance leads to four to five times the force typically used in conversational voicing. The PEEP valve is also assumed to have been tested to specific tolerances prior to clinical use, though some loss of precision or even functional resistance for a percentage of these calibrated devices appears to be within the realm of possibility. Anecdotal clinical use of the PEEP device and PMVE protocol has also indicated reductions in coughing for some patients. Given the role that respiratory issues can play in coughing, this is not surprising. Future research may therefore examine the role of this novel therapy in treating refractory chronic cough.
Limitations and Future Directions
There were several limitations noted in the study. Each treatment group (the PMVE vs. VFE) had only eight individuals diagnosed with different forms of dysphonia and were consecutively enrolled which could suggest a small and biased sample. To address this bias, the authors used the coin flip method for random assignment to minimize any baseline differences between the groups. Thus, future studies should include larger, randomly selected, representative samples including individuals with different forms of dysphonia. Another limitation could be the administration of 6 weeks of home-based therapy which might be inadequate training dosage to improve the study outcomes. These findings suggest that patients with dysphonia may require extended treatment programs due to a variety of factors including but not limited to inadequate compliance with home exercise programs, prolonged noncompliance with dosage-dependent therapies, the nature and severity of the disorder being treated, and illness or other medical issues arising during a course of treatment. Future studies should, therefore, examine different durations for PMVE to demonstrate its efficacy. A third limitation is that the voice recordings used for acoustic and perceptual assessments were captured in a clinic for this study, as opposed to the recommendations made by Patel et al. (2018), which impacted the quality of the recorded samples and, therefore, the SNR calculations. To improve the SNR quality, the authors used processing software to filter the audio recordings and assist with interpretation. Future studies should, thus, include a controlled environment for these assessments such as capturing acoustic samples in a soundproof audiometric booth. Lastly, this study did not determine the intra- and inter-rater reliability between the two GRBAS raters; therefore, the readers should interpret these assessment methods with caution.
Conclusion
Both the VFE and PMVE therapy programs performed similarly after 6 weeks of home-based training. This suggests the potential efficacy of the PMVE protocol. The VFE demonstrated pre- versus post-differences across a range of outcome measures, as compared to the PMVE program which resulted in acoustic improvements only. Therefore, PMVE may offer therapy alternatives for therapists working with patients to resolve dysphonia, especially in acoustic measures. Advantages for PMVE appear to include the ability to overtrain the voice at progressively higher levels of resistance without switching between therapy protocols or approaches, to target the speaking or singing voices with ease rather than being forced by a protocol to practice at lower levels of the discourse hierarchy, and to functionally unify the lungs with the rest of the voicing mechanism. Further research, however, is still warranted given the novelty of PMVE and the limitations noted in this study related to sample size and clinical representativeness.
The PMVE Program
The PMVE protocol uses either a size 5 or 6 ventilation mask (Ambu King Mask manufactured by Ambu A/S). These masks were chosen because they featured a soft, tacky cushion that was assumed to be comfortable for participant's use while also allowing for an adequate seal when pressed into the face. Sizing was determined by holding up a mask before the facial mask area and estimating the reasonable capacity of the mask to cover the nose and mouth while allowing for unimpeded articulation. Fitting was then attempted. In all cases, it was recommended that subjects use the index and middle fingers on each hand to press the mask cushion gently but firmly around the bridge of the nose during exercises. See [Fig. Appendix 1] for images depicting the proper wear approach. Masks were then fitted with a disposable PEEP 20 valve (Ambu A/S) using a compatible adaptor. See [Fig. Appendix 2] for an image of the phonatory PEEP device.
After proper mask fit was achieved, the initial practice level (IPL) was determined using the toleration protocol. The IPL refers to the beginning level of resistance provided by the PEEP valve when completing the therapeutic components of the program. For individuals in the PMVE group, resistance was initiated at 7 to 8 cm H2O by manually twisting the valve cap until the lower edge was midway between the 5 and 10 H2O lines marked on the PEEP valve. Participants were warned that they would not be able to inhale with the mask properly fitted over the nose and mouth. They were warned to keep the device out of the reach of unsupervised children or those who would be assumed to have difficulty manipulating the mask given this issue. Participants were also instructed to practice with another responsible adult present as standard clinical care targeting reduced risk of injury. Participants were then instructed to complete automatic speech tasks (e.g., counting, reciting the alphabet, and naming days of the week or months of the year in sequence). During these tasks, observations were made specifically assessing for signs of inappropriate mask fit, excessive effort, or intolerance. If inappropriate mask fit or intolerance were demonstrated, adjustments were made until an adequate seal and tolerance were achieved. If signs of excessive effort such as impounding of air into the cheeks or audible excessive leakage of air through the PEEP valve cap were demonstrated, efforts were made to cue diaphragmatic breathing to reduce effort. If faintness was reported, participants were coached through repetitive diaphragmatic breathing cycles after removing the mask and allowing for recovery to baseline breathing. If toleration was not achieved at the initial 7 to 8 cm H2O setting after the adjustments, participants were instructed to adjust the valve down to its least resistance setting at 5 cm H2O. This became the IPL for these participants. Conversely, if voicing was adequately maintained and no patient complaints were verbalized or observed at the initial 7 to 8 cm H2O setting, the PEEP valve was adjusted upward to 10 cm H2O. Toleration with automatic speech was again assessed and this became the IPL if no signs of difficulty maintaining voice or intolerance were demonstrated. If this higher impedance level proved too difficult for maintenance of voicing or intolerance was observed or reported, the valve was reduced back to 7 to 8 cm H2O, and this became the IPL for these participants. In all cases for this study, the IPL was 7 to 8 cm H2O, though the investigators were careful to observe if the transition from initial warm-up tasks to automatic voicing was tolerated differently. As a functional rule set, reducing the adjustable PEEP value resulted from (1) inability to maintain voicing and/or (2) inability to maintain voicing without excessive effort or toleration issues. Conversely, advancing the adjustable PEEP valve to a higher resistance level resulted from (1) demonstrated ability to maintain voicing and/or (2) maintenance of voicing without excessive effort or toleration issues.
A weekly schedule to advance the PEEP valve 2 to 3 cm H2O as tolerated was then provided in verbalized and written forms. With the mask in place fitted over the nose and mouth and at the prescribed IPL, the following protocol was completed:
-
Hold “ah” at a high pitch for as long as possible two times (F above Middle C for women or F below Middle C for men were offered as targets as requested).
-
Glide “ah” from your lowest pitch to your highest pitch two times.
-
Glide “ah” from your highest pitch to your lowest pitch two times.
-
Speak or sing for 3½ minutes (women) or 4 minutes (men).
General practice protocols involved the following instructions:
-
You should not use a loud voice or attempt to project your voice through the mask.
-
You should not be able to hear or feel air leaking around the mask near the nose or mouth.
-
You should not hear air leaking through the valve cap on the end of the mask.
-
You should not fill your cheeks with air.
-
You should not feel strain in your neck.
This standard regimen of voicing tasks of the Progressive Masked Voice Exercises (PMVE) program is completed in four minutes and practiced twice daily. For home practice, to increase accuracy, participants also received recordings of F above middle C for women or F below middle C for men.
General care instructions for the mask included wiping it down with a sanitizing cloth or alcohol pad. These were provided along with a plastic bag with a closure top though participants were instructed to allow the bag to remain partially open to allow the device to air dry.
Capillary Refill Protocol
A capillary refill protocol was provided to determine adequate hydration levels and avoid faintness as the primary concern for intolerance. Capillary refill has previously been suggested to assess circulatory status with at least one study indicating the upper normal limit as two seconds for children and adult men, three seconds for women, and four seconds for the elderly though it was allowed that environmental temperature or even the perception of habitually cold or warm hands impacted refill rates (Schriger & Baraff, 1988). In the present study, subjects were instructed to press the index and middle fingers into the flesh covering the lower two ribs hard enough to Blanche the tissue. Fingers were then to be removed with instructions to observe the return of typical tissue coloration. If coloration returned in under three seconds, subjects were then instructed to proceed with their assigned therapy protocol. If coloration returned at three seconds or more, participants were given instructions to drink water and repeat the task until adequate capillary refill was demonstrated, typically after 30 minutes. An analogous protocol used in the clinic to demonstrate this approach included using the index finger and thumb to blanche the flesh covering the first dorsal interosseous musculature on the opposite hand as the protocol delineated above was determined to be unnecessarily invasive to practice in the clinic as it would involve at least partial doffing of clothing.
Contraindications for PMVE
A single participant completing PMVE demonstrated a visual increase in the density and severity of feeder vessels to a benign left vocal fold lesion (the lesion was biopsied after follow-up assessment and histopathology indicated this was most likely a nodule with prior occult hemorrhage of an associated vessel). Though no subsequent hemorrhage was demonstrated, and voicing behaviors may have reasonably explained this result (e.g., excessive throat clearing or coughing), the investigators generally advise caution when attempting to use the device or its protocol with participants demonstrating prominent vocal fold or laryngeal vessels and complete avoidance where a vessel hemorrhage or hematoma have been demonstrated. With this risk considered, the overall capacity of this therapy to support wound healing in the context of increasing blood flow, supplying oxygen and other nutrients, and carrying off metabolic waste has not yet been examined. Individuals demonstrating dysplasia of the vocal folds or larynx, biopsy-proven cancer of the vocal folds or larynx, or positive margins for resected vocal fold cancer should not use the device or protocol as angiogenesis and blood supply issues are generally accepted to be factors in tumor growth. Metastatic disease or the presumption of any cancer in the respiratory system, throat, or nasal cavity would also eliminate reasonable application of this therapy for the same reasons. It has been found that PEEP use in the context of ventilation may reduce cardiac output and result in hypotension (Mughal et al., 2005). Other investigators addressed concerns that have been raised about the use of PEEP in participants with intracranial vascular abnormalities or those with significant medical histories for the lungs (Ranieri et al., 1993). Cardiopulmonary concerns have also been pointed out for individuals with COPD with barotrauma related to hyperinflation of the lungs among them (Mughal et al., 2005; Ranieri et al., 1993). As hyperinflation of the lungs is assumed to cause some level of at least indirect pressure in the esophagus (Blanch et al., 2005; Tuxen & Lane, 1987), those individuals with Barrett's esophagus or those presumed to have this issue and other disorders involving the esophagus may be at risk when using PEEP as proposed in this study. Clinicians are therefore cautioned to seek out physician approval before using PMVE or avoid use for any of these populations. Other populations that should not be considered for the use of PMVE would be those with significant cognitive impairment including those with dementia, pediatrics without total one-on-one supervision from a competent adult, individuals with known syncope issues or seizure disorders, or those without adequate manual strength to manipulate posturing of the device as these participants would presumably be at risk for suffocation. Because cardiovascular suppression is an issue, athletes or participants with exceptionally low resting heart rates may also not tolerate this therapy well unless compensatory strategies are utilized.
