The maximum phonation time as marker for voice treatment efficacy: A network meta‐analysis

There is a diversity in treatment approaches for voice therapy in which aerodynamic treatment effects between the approaches are lacking. The evidence of voice treatments on the maximum phonation time (MPT) was quantified using the statistical approach of a network meta‐analysis (NMA).


| INTRODUCTION
A comprehensive clinical voice assessment examines the processes involved in phonation, including laryngeal imaging analysis, auditoryperceptual judgement, aerodynamic analysis, acoustic analysis, and self-assessment. 1,2 These multidimensional measures are also acquired to assess pre-to-post treatment changes. The clinical application of aerodynamic measures such as mean airflow, estimates of subglottal pressure, and temporal phonation measures (e.g., maximum phonation time, or MPT) are supported by a strong evidence base. Aerodynamic parameters provide valuable information about the interaction between respiratory and phonatory mechanisms of speech with a detailed view of the underlying physiology of phonation, the degree of effort required for voicing, and compensatory behaviours used by patients with voice disorders. 3 MPT is considered an aerodynamic assessment which measures glottal efficiency by estimating the ability to control the pulmonary aerodynamic forces and the myoelastic forces of the larynx. 4 For MPT performance, the patient must sustain a vowel (e.g., /a/) after maximum inhalation for as long as possible, in which the total sound duration is measured in seconds. To be able to phonate a steady tone for a long time, healthy lung function along with adequate lung capacity and flow velocity of the air is a pre-requisite. 4 The alignment of subglottic pressure and glottal resistance is also crucial for sustained phonation over time. If the MPT duration is low, it is assumed that the dosage of breathing is difficult or that the glottal resistance, depending on the laryngeal pathology, is too low, which might have consequences in voice quality (e.g., a breathy voice). 5 Further influencing factors on MPT score include the number of trials, 6-8 gender, 6,9 and body mass index. 9 MPT is a measure which is often used in voice assessments due to its non-invasiveness, fast feasibility, and low cost. 7 In addition, MPT has been used to objectify the degree of severity of dysphonia, 10 and to determine the effects of voice treatments. 11 Voice treatment is considered effective using MPT if the duration post-treatment is at least 1.41 s longer than the pre-treatment score. 12 With regard to a testretest procedure in vocally-healthy participants, an extension of the MPT duration can also be expected at 1.01 s. 3 Dysphonia arises from many aetiologies. 13,14 These determine the primary treatment approach, which may include behavioural voice treatment, 11,15-17 phonosurgery, 18,19 and/or pharmacological therapy. 20 Successful behavioural voice treatment depends on various factors. The underlying aetiology (e.g., neuropathologic, emotional, improper vocal habits, and structural pathology), the maintaining factors, patient motivation, the treatment approach, and the skill of the clinician are all factors that can influence treatment outcomes. There are many behavioural voice treatment approaches available to a voice clinician. To compare more than two treatments in a single analysis a statistical approach of a network meta-analysis (NMA) can be utilised.
NMA is an extension of pairwise meta-analysis to compare three or more treatments for a given medical or healthcare condition, based on combining information from multiple existing comparisons among subsets of the treatments. 21 This statistical method of a meta-analysis can effectively compare the efficacy of voice treatments using the same outcome parameters and study designs.
A prior NMA for voice treatments investigated the outcome measure of voice-related handicap based on the voice handicap index (VHI). 17 That study was able to combine data from 13 randomised controlled and clinical trials (RCT) which collectively studied the effect of nine different treatment approaches on self-perceived VHI measures. Of those nine interventions, four were found to effectively improve VHI measures (e.g., at greater than chance levels based on meta-analysis statistics) from pre-to-post therapy. The analysis of pooled data via the NMA design provided high-quality evidence.
The purpose of this meta-analysis was to apply a NMA focused on RCTs evaluating the efficacy of voice treatment approaches on measures of MPT. As with our prior study, we also sought to establish an effectiveness ranking for different voice treatment approaches.
By extending knowledge of the response of MPT to treatment approaches through a NMA design, clinicians may be better informed of the potential differential effectiveness of varied treatment approaches for a given voice disorder. This information may also help to interpret future clinical studies which include MPT as an outcome variable. As with our previous NMA investigating treatment effects on VHI, MPT is also a common outcome measures in clinical practice and clinical research, and justification for its use through investigation using high-quality research designs such as NMA is needed.

| Data sources and searches
The preferred reporting items for systematic reviews and metaanalyses (PRISMA) extension statements for reporting of systematic reviews incorporating NMA of healthcare treatments was used to perform a systematic literature search in electronic databases. 22 The

Key points
• Various voice treatments were compared using maximum phonation time as marker for voice treatment efficacy.
• This is the second network meta-analysis on the treatment efficacy of dysphonia next to the evaluation of the voice-related handicap.
• One from eight voice treatments resulted in a significant improvement of maximum phonation time.
• Vocal function exercises have been identified in the present network meta-analysis as the only effective intervention.
• Vocal function exercises were evaluated in two network meta-analyses as effective in eliciting clinically significant treatment changes in voice handicap index and maximum phonation time.
MEDLINE, CENTRAL, and Speechbite electronic databases were searched from inception to November 19, 2021. Meaningful papers were identified by title and abstract. In addition, references were also reviewed in potential articles. The following search terms were used to identify the potential studies: voice disorders, dysphonia, voice therapy, voice treatment, and maximum phonation time.

| Study selection
Potential studies were included which evaluated in a RCT design a specific voice treatment approach with (a) comparisons between types of treatments or (b) control group without an intervention. Papers that examining the efficacy of voice treatments in participants with a neurological motor speech disorders or in vocally healthy participants were excluded. The use of medical or pharmacological treatments were also excluded. In addition, studies were excluded if they did not provide proper explanations of the voice treatment approaches or if the voice therapy procedure did not include a primary single approach. Lastly, interventions that included technical instruments in the application of voice treatment were excluded as well.

| Data extraction and risk of bias assessment
Two researchers (BBvL and KS) independently screened titles and abstracts. Full-text articles were then reviewed in duplicate and disagreements resolved by discussion together. We extracted data about: details of study design, types of voice disorder, description of voice treatment intervention, description of comparison group, completeness of outcome data, outcome measure of maximum phonation time, and statistics. If not reported, study authors were contacted. We assessed the risk of bias of the included studies using the PEDro scale.
The PEDro scale is an 10-item scale for assessing the methodological quality of RCT studies. The total score of the PEDro scale ranges from 0 to 10 points. 23 A score of 9-10 corresponds to excellent methodological quality, a score of six to eight to good quality, and a score of four to 5 to low quality. To assess the quality of studies, the PEDro scale considers the following factors: inclusion and exclusion criteria, randomised and concealed assignment of groups, association of all study participants and investigators, key outcome parameters, and statistical group comparisons. 23 F I G U R E 1 Course of study selection for the meta-analysis

| Statistical analysis
The mean differences of the pre-post treatment (MD) and SD results of the MPT values for each treatment arm were extracted from each study. When standard deviations were lacking, the p values of the pre-post MD were used to calculate them. For random-effects NMA, the R package netmeta from the open statistical programming environment R was used. 24,25 Results were demonstrated as MD between pre-post outcomes with 95% confidence intervals (CI). In addition, the ranking of interventions was related on the p score variable. This is a critical ranking score that can be viewed as a frequentist analogue to the area under the Bayesian cumulative ranking curve, without the need for resampling methods. 26 It is an easy analytic method based on frequentist point estimates and their standard errors. p-scores result in a ranking on a scale from 0 to 1, with 0 being the worst and 1 being the best. This p-score can be interpreted as a measure of the mean level of confidence that a treatment is better than a comparable treatment. Results for the comparison with the control group are shown in the forest plot. This was created with the R package metafor. The mean difference is presented with the 95% confidence interval, as well as the p-score.

| Study characteristics
The selection of papers for this meta-analysis is presented in the PRISMA chart ( Figure 1). The study characteristics of the included RCT's are presented in Table 1 Table 2).

| Efficacy of voice treatment methods
There is a high heterogeneity in the results of NMA (I 2 = 81.8%) and

| DISCUSSION
This present NMA is the second of its kind to evaluate the treatment efficacy of voice treatment approaches in voice-disordered participants investigated in RCT study designs. Both network meta-analyses generally reported a low risk of bias for included studies. The prior NMA investigated the VHI as primary outcome parameter, which is a common used voice parameter measuring the impact from the view of the patient. 17 In that prior study, stretch-and-flow phonation revealed significant and clinically relevant treatment outcomes. This intervention ranked as the superior voice treatment across all other treatments, but RV, comprehensive voice rehabilitation program, and VFE approaches demonstrated statistically significant improvements on VHI score, as well. 17 Three of these four relevant voice treatments were included in the present NMA with the objective aerodynamic primary outcome parameter of MPT as well. However, VFE was the only treatment approach confirmed to yield significant improvement in MPT compared to other voice treatment interventions. Thus, two NMA studies have found VFE to be effective in eliciting clinically significant pre-to-post treatment changes in measures of VHI and MPT.
VFE as a treatment approach utilises four core exercises that were designed to strengthen and balance the laryngeal musculature and to calibrate airflow with muscular effort. 39 The treatment stimuli consist of multiple repetitions of a nasal vowel /i/ (exercise #1) or modified vowel /o/ phonation-called lip buzz (exercise #2 to 4), in which maximal sustained phonation is trained in exercises #1 and #4. 40 As such, VFE can be characterised as adhering to the specify principle for motor learning when relating these exercises to measures of MPT. That is, two of the four VFE exercises require maximum sustained phonation.
VFE is a physiologic vocal treatment concept that strive to improve the strength, balance, and stamina of laryngeal muscles under consideration of improving the balance among laryngeal muscle effort, respiratory effort and control, and the supraglottic placement of the tone. 39 A previous meta-analysis of VFE 16 showed comparable results to the present NMA in MPT mean improvements. In this previous metaanalysis seven studies with different study designs were included for the effect size calculation. 16

| Limitations and future directions
The caveats of the present meta-analysis relate to the generalizability of the results, but also provide direction for future research.
F I G U R E 3 Forest plot representing the treatment effect sizes of the maximum phonation time by specific interventions in comparison with the control group sorted at the order of treatment rankings of the p-score. The grey area between the two blue lines represents the equivalence region for the MPT score at 1. Third, many voice treatment approaches were not considered in the present study, although they may have been considered in prior systematic reviews or surveys on voice treatment. 11,12,15,17,40,45 Reasons for exclusion were based primarily on the lack of RCT design, where studies used other voice treatment outcomes than MPT, or dropped important statistical data for a NMA.
Fourth, the search strategy for relevant papers for this present study was conducted in two languages. Other languages such as Asian languages, Spanish, or French were omitted, which could also contain potentially relevant publications.
Fifth, further studies are needed to evaluate voice treatments for homogeneous groups of dysphonia. This meta-analysis has considered organic and non-organic voice disorders in a fairly balanced proportion. However, there is a high diversity between the numerous types of voice disorders and severity of dysphonia that may vary considerably the performance of the MPT. Evaluating the treatment effects of specific approaches for other voice disorders generally encountered in clinical practice, for example laryngitis, vocal fold nodules, vocal fold polyps, vocal fold oedema, vocal fold cyst, vocal fold paralysis, leucoplakia, carcinoma, and muscle tension dysphonia. 13,14 Sixth, the assessment of intensity and frequency of voice treatment approaches should be further explored, as treatment dose issues are poorly understood in the voice treatment literature.

| CONCLUSION
This NMA reported here provides evidence that VFE was effective in improving MPT. The effect on MPT showed in other voice treatment approaches no significant or clinically relevant treatment effects. Thus, the presented NMA reduced the number of effective interventions from a larger pool of voice treatment approaches based on an aerodynamic analysis, when MPT is being utilised as a clinical outcome measure. Otherwise, these findings may be especially useful for clinicians using diverse therapies to treat dysphonia to better assess method-specific expectations for improvement in aerodynamic measurements after MPT. Finally, the present study highlights that further contributions of high-quality intervention studies are needed to support clinical practice in laryngology.

AUTHOR CONTRIBUTIONS
Concept and design: all authors. Acquisition, analysis, or interpretation of data: BBvL and KS. Drafting of the manuscript: BBvL and CRW.
Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: SH.

CONFLICT OF INTEREST
None.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/coa.14019.

DATA AVAILABILITY STATEMENT
Data may be made available on request.

ETHICS STATEMENT
There is no ethic statement given the type of study (meta analysis), which does not require ethic committee.