Examining Therapy Duration in Adults With Voice Disorders

Abstract

Purpose:

This study examined the number of voice therapy sessions and the number of weeks in treatment to achieve desired voice outcomes in adults with voice disorders. Factors that may predict therapy duration were examined, as was the percentage of patients returning to the clinic for additional voice therapy after initial discharge.

Method:

An observational cohort design was utilized. Data from 558 patients were extracted from the University of Wisconsin–Madison Voice and Swallow Outcomes Database. Patients diagnosed with muscle tension dysphonia, vocal fold paralysis, benign vocal fold lesions, laryngospasm/irritable larynx, and presbyphonia were examined. Patient demographics, auditory-perceptual assessments, acoustics, aerodynamics, videostroboscopy ratings, self-reported scales, and medical comorbidities were collected.

Results:

Patients required an average of 5.32 (SD = 3.43) sessions of voice therapy before voice outcomes were sufficiently improved for discharge. Average number of sessions ranged from 4.3 for presbyphonia to 6.7 for benign vocal fold lesions. Baseline overall Grade Roughness Breathiness Asthenia and Strain rating (p < .001), Dysphonia Severity Index (p < .001), Voice Handicap Index score (p < .01), age (p = .006), and occupational voice user status (p < .001) significantly predicted the number of therapy sessions required. Overall, 14.5% of patients returned for additional voice therapy following an initial discharge from treatment.

Conclusions:

Findings inform our understanding of how many sessions patients with voice disorders require to achieve desired voice outcomes. Additional research is needed to optimize the efficacy of voice treatment and determine how recurrence of dysphonia might best be prevented.

For as many as 17.9 million adults in the United States, voice disorders (dysphonia) inhibit effective communication and detract from quality of life (Bhattacharyya, 2014; Cohen et al., 2012; Roy et al., 2004). Voice disorders can be costly for patients in both medical care and days missed from work (Van Houtte et al., 2011; Williams, 2003). Voice disorders often have negative emotional side effects as they can inhibit social engagement (Krischke et al., 2005; Merrill et al., 2011, 2013) and impact the manner in which individuals are perceived by their peers (Amir & Levine-Yundof, 2013; Maryn & Debo, 2015). Etiologies of dysphonia are varied and include functional voice patterns (Aghadoost et al., 2020; Shembel et al., 2021), morphological vocal fold changes (Behrman et al., 2004; Hron et al., 2019), neurologic pathologies (Wang & Song, 2022), and even age-related laryngeal changes (Desjardins et al., 2020; Martins et al., 2014).

Best practice dictates that dysphonia should be managed by laryngologists and speech-language pathologists (SLPs; Aronson & Bless, 2009). Although surgical interventions may be indicated to address some laryngeal pathologies, voice therapy with an SLP is often the primary form of treatment (Casper & Murry, 2000; Desjardins et al., 2017). Even when surgery is required, procedures are often coupled with voice therapy to ensure safe voice use and prevent further laryngeal injury (Tang & Thibeault, 2017; Tibbetts et al., 2018). Fortunately, voice therapy can result in improvement demonstrated by both self-reported and objective measures of voice (Holmberg et al., 2003; Schindler et al., 2013).

The gains observed in objective measures following voice therapy have spanned diagnoses and outcome measure. In patients with vocal fold lesions, voice therapy improves perturbation measures (Petrovic-Lazic et al., 2015; Schindler et al., 2013) as well as functional auditory-perceptual voice assessments (Sahin et al., 2018). Emerging evidence suggests that voice therapy improves aerodynamic voice measures in this population as well (Holmberg et al., 2003). These positive changes suggest that both voice quality and the efficiency of laryngeal physiology improve following voice therapy. Similar improvements in perturbation measures (D'Alatri et al., 2008; Schindler et al., 2008), auditory-perceptual assessment (Miller, 2004), and perception of voice impairment have also been observed in individuals with vocal fold paralysis (Busto-Crespo et al., 2016). Additionally, voice therapy has been effective for functional voice disorders such as muscle tension dysphonia (Roy, 2008) and puberphonia (Alam et al., 2012; Desai & Mishra, 2012). In patients with dysphonia secondary to vocal hyperfunction, voice therapy has improved cepstral/spectral measures and relative fundamental frequency (f _o; Stepp et al., 2011; Watts, Diviney, et al., 2015; Watts, Hamilton, et al., 2015) as well as auditory-perceptual voice assessments (da Cunha Pereira et al., 2018). These findings further indicate that voice therapy improves voice quality and increases the efficiency of vocal function. Improvements in acoustics and auditory-perceptual voice outcomes are also evident for patients diagnosed with presbyphonia or age-related voice changes (Mau et al., 2010; Ziegler et al., 2014) as well as laryngospasm (Murry & Sapienza, 2010). Taken together, numerous studies provide compelling evidence that voice therapy is an efficacious tool in the treatment of voice disorders (MacKenzie et al., 2001; Ramig & Verdolini, 1998).

Although the efficacy of voice therapy is well established, important aspects of this treatment remain poorly understood. For example, the required duration of voice therapy has not been widely investigated. Initial study indicates that voice patients in the United States and Canada receive an average of 12.52 sessions over the course of 7.62 weeks, and those in Europe receive an average of 10.99 sessions over 10.12 weeks (De Bodt et al., 2015). These numbers, however, do not account for patient diagnosis and may be influenced by individuals receiving longer-term behavioral interventions such as the Lee Silverman Voice Treatment (LVST; De Bodt et al., 2015; Ramig et al., 2004). Other studies have examined the course of voice therapy within specific patient populations (Fujiki et al., 2022). Voice therapy for muscle tension dysphonia, for example, has produced gains in vocal function in as little as 1–3 hr of intervention (Roy & Leeper, 1993). It has been reported that patients with benign vocal fold lesions require between two and 16 sessions of voice therapy (Lockhart et al., 1997; McCrory, 2001), but reasons for this wide range in session number have not been thoroughly explored. For patients with unilateral vocal fold paralysis, it has been reported that patients received an average of 12.6 sessions over approximately 6 weeks to achieve voice gains (Schindler et al., 2008). It has also been reported that children with benign vocal fold lesions achieve desired voice outcomes in approximately eight sessions of voice therapy (Fujiki & Thibeault, 2022). As many of the aforementioned studies either have relatively small Ns or do not allow for significant generalization outside the examined population, additional investigation is warranted.

Treatment duration has direct financial and emotional implications for patients. Previous study indicates that overall health outcomes improve when patients have clear expectations of treatment (Adams, 2010). More specifically, the study of intervention duration in other SLP specialties indicates that understanding how many therapy sessions are likely to be needed to achieve a desired outcome is important information as it allows patients to prepare for the monetary, transportation, and cognitive demands of treatment (Cherney, 2012). Understanding the number of required sessions over a given time period is also important as it allows patients to anticipate how many weeks, months, or years they can tentatively expect to be in therapy (Baker, 2012; Warren et al., 2007). Since many factors may influence progress in treatment, it is also important for patients to understand what percentage of patients may return to the voice clinic for additional therapy after an initial discharge. Currently, however, it is difficult for clinicians to give evidence-based predictions regarding the number of therapy sessions and the number of weeks spent in therapy. This is unfortunate because patient commitment and motivation might be enhanced if clinicians could better prepare them to anticipate what resources they must devote to their voice treatment program.

Currently, it is unclear what factors may predict or influence the number of voice therapy sessions required to achieve desired voice outcomes. Initial work indicates that more severe auditory-perceptual voice assessment or Voice Handicap Index (VHI) scores at baseline may be associated with a longer course of therapy (Smith et al., 2010). Patients with more complex medical histories have been less likely to complete prescribed voice therapy sessions (Smith et al., 2010). It has been reported that insurance issues and distance to provider may impact patient decisions to start voice therapy in the first place (Portone et al., 2008). Similarly, occupational voice use may affect the number of sessions needed. Other potential factors including sex, age, diagnosis, and comorbid behavioral health diagnosis remain largely unexplored. In summary, factors predicting the number of sessions required to achieve desired voice outcomes warrant further study.

This investigation examined the number of voice therapy sessions and number of weeks of treatment required to achieve patient desired voice outcomes in adults with voice disorders. Voice disorders included benign vocal fold lesions, muscle tension dysphonia, vocal fold paralysis, laryngospasm/irritable larynx, and presbyphonia. Additional factors (i.e., severity of dysphonia, occupational voice user status, age, sex) that may predict the number of required sessions were also examined, as was the frequency of patients returning to the clinic for additional voice therapy after an initial discharge. It was hypothesized that increased baseline severity of dysphonia, baseline VHI score, and occupational voice user status might increase the number of sessions needed to achieve desired voice outcomes.

Method

Study Design

A retrospective observational cohort design was employed. Data were pulled from the University of Wisconsin (UW)–Madison Voice and Swallow Clinics Outcome Database. This database is monitored by the Wisconsin–Madison School of Medicine and Public Health Institutional Review Board. Patients provide consent for participation in this database when they present to the Voice and Swallow Clinics of a large tertiary hospital. Data are then prospectively gathered as these patients receive further health care. At present, the database includes information from approximately 6,868 patients of varying diagnoses.

Eligible patients presented to the clinic between April 2009 and June 2021. Additional inclusion criteria were as follows. Patients were required to (a) present with a primary voice complaint, (b) be over 18 years of age, (c) complete a voice evaluation with an SLP and laryngologist, (d) complete a course of voice therapy with an SLP, and (e) be discharged from voice therapy.

Patient medical diagnoses were determined by a laryngologist and were divided into five categories: (a) muscle tension dysphonia, (b) vocal fold paralysis, (c) benign vocal fold lesions, (d) voice-related complaints associated with laryngospasm/irritable larynx, and (e) presbyphonia. Muscle tension dysphonia was defined based on previous study characterizing the condition as dysphonia occurring in the absence of neurological or organic etiologies (Payten et al., 2022; Verdolini et al., 2006). Videostroboscopic findings associated with this diagnosis included supraglottic constriction, laryngeal hyperfunction, or vocal fold hypoadduction upon phonation (Verdolini et al., 2006). Exclusionary criteria included diagnoses associated with neurological diseases such as Parkinson's disease, laryngeal cancer, or recurrent respiratory papilloma. These diagnoses were excluded because the treatment model for voice therapy for patients with Parkinson's disease differs in duration and intensity when compared with other voice disorders (Ramig et al., 2004). Thirty-five patients with neurological voice disorders were initially included in the sample but were excluded as they were enrolled in either LVST or SPEAK OUT!, the duration of which are often prescribed at baseline. Laryngeal cancer and papilloma patients were excluded as the primary treatment for these conditions was generally surgical (Davis et al., 2022; Taliercio et al., 2015). Additionally, patients with a primary complaint of paradoxical vocal fold movement were also excluded, as breathing-focused treatment differs from dysphonia-focused therapy (Burke, 2012). Patients with histories of in-patient laryngeal surgeries or procedures were also excluded. All patients underwent laryngeal imaging using rigid endoscopy or flexible endoscopy with chip tip camera. Laryngeal structures were observed under both halogen (steady-state) and xenon light (allowing for stroboscopy). Using both methods, videostroboscopy was used to assess the vibratory parameters of phonation.

Treatment

Consistent with best practice, voice therapy was individualized for each patient (van Leer & Connor, 2010), but similar elements were included in all patients' treatment programs. Therapy consisted of a combination of direct and indirect techniques. Direct voice therapy techniques included resonant voice exercises, semi-occluded vocal tract exercises (SOVTE), vocal function exercises, flow phonation, and manual circumlaryngeal techniques (Meerschman et al., 2019; Roy et al., 1997; Titze, 2006; Verdolini & Li, 2019). Additionally, vocal hygiene education was provided (Behrman et al., 2008). If patients had high-level voice needs such as singing, acting, or public speaking, these needs were also addressed. Therapy sessions were conducted in person and lasted approximately 50 min. Patients were assigned voice exercises to do at home between therapy sessions. SLPs specialized in the treatment of voice disorders provided all therapy.

Discharge From Therapy

Discharge criteria for all patients were tailored to their specific vocal needs (Gillespie & Gartner-Schmidt, 2018). Before discharge from therapy, all patients demonstrated sufficient proficiency with the voice exercises introduced. Exercise proficiency was determined by the treating clinician and was documented in patient's treatment notes. In addition, all patients indicated that they felt confident in their ability to address any reoccurrence of symptoms. All patients reported that their voice met their daily needs and that they were satisfied with their vocal range, endurance, intensity, and f _o. Patients also expressed understanding of vocal hygiene and indicated that they felt comfortable being discharged from therapy. Discharge status was determined based upon clinician documentation indicating that the above conditions were met and that therapy course was concluded.

Predictors of Therapy Duration

For each patient, the number of voice therapy sessions completed between initial voice evaluation and discharge from therapy was calculated. Additionally, it was noted which patients returned to the clinic for additional voice therapy sessions after discharge from therapy. The following factors were extracted from the database as potential predictors of therapy course.

Demographics

Patient demographics were collected including diagnosis, age, sex, employment status, and number of years smoked (0 if patient had no smoking history). Diagnoses were sorted into the following categories: (a) muscle tension dysphonia, (b) vocal fold paralysis, (c) benign vocal fold lesions, (d) voice-related complaints related to laryngospasm or irritable larynx, and (e) presbyphonia. Patients also indicated whether their voice was crucial for their occupation and their response was collected (yes or no).

Patient-Reported Voice Symptoms

Time since onset of dysphonia and onset type (sudden, gradual) were also collected. It was also recorded whether patients endorsed increased vocal effort (yes/no). Voice therapy schedule was also categorized as either weekly or other (if longer than a week passed between visits). Additionally, patients indicated whether their voice complaint had caused them to reduce their habitual voice use in a percentage (0%, 25%, 50%, 75%, 100%).

Auditory-Perceptual Assessment

GRBAS (Hirano, 1981) ratings made by the evaluating SLP at the time of evaluation were extracted from the medical record.

Videostroboscopic Findings

Laryngeal imaging findings were also pulled from the database. Relevant strobe findings included amplitude of vocal fold vibration, mucosal wave, and glottic closure as well as medial and anterior compression. Amplitude and mucosal wave were rated as normal, mildly, moderately, or severely decreased. Glottic closure pattern was determined using the classifications previously proposed in the literature (Poburka & Patel, 2021; Poburka et al., 2017). Compression was rated as none present, mild, moderate, or severe.

Acoustic Voice Measures

Acoustic measures extracted from the database were collected from recordings made using the Computerized Science Lab developed by KayPENTAX (Model 4500). Acoustic data were analyzed on the middle 1 s of sustained /a/ vowels using the Multi-Dimensional Voice Program (Choi et al., 2012). Acoustic voice measures included jitter%, shimmer%, noise-to-harmonic ratio, maximum phonation time, f _o, Dysphonia Severity Index (DSI), and maximum and minimal f _o of vocal range.

Aerodynamic Voice Measures

Aerodynamic measures collected from the database included phonation threshold pressure (PTP), subglottal pressure, laryngeal airway resistance, and transglottal airflow. Measures were collected using the Phonatory Aerodynamic System (Model 6600) developed by Kaypentax.

Patient-Reported Scales

The VHI, Glottal Function Index (GFI), and Reflux Severity Index (RSI) were also extracted. The VHI is a 30-question measure designed to assess the impact of voice disorders on patients' lives, with greater scores being indicative of more impaired function (Jacobson et al., 1997). On the GFI, possible scores range from 0 to 20, with scores over 4 being indicative of a possible voice disorder (Bach et al., 2005). On the RSI, scores over 14 are potentially indicative of reflux (Belafsky et al., 2002).

Comorbidities

Comorbid complaints extracted from the database and categorized as either present or not (yes or no) included dyspnea, dysphagia, cough, globus sensation, asthma, allergies, self-reported history of reflux, or a comorbid behavioral health diagnosis.

Statistical Analysis

Predictors of voice therapy duration were identified using linear regression. To identify the factors that should be included in the regression model, all factors listed above were entered into a correlation matrix. Factors significantly correlated with the number of therapy sessions at an r ≥ .2 were included in the final regression model. Number of therapy sessions was regressed on age, diagnosis, whether voice use was crucial for employment, PTP, overall grade of voice quality on the GRBAS, DSI, mean glottal airflow, VHI, and history of a behavioral health diagnosis. Pearson's correlations or t tests with Bonferroni corrections were used for post hoc analyses. Significance was determined by α < .05. t tests and Fisher's exact tests (for odds ratios [ORs]) were used to analyze demographic data and comorbidities. All statistical analyses were completed in SPSS (Version 28).

Results

Data from 558 patients with diagnosed voice disorders (M = 52.33, SD = 16.9, women = 355, men = 203) were included, and the distribution of diagnoses, patient demographics, and proportion of patients reporting various comorbidities is presented in Table 1. The majority of patients presented with either muscle tension dysphonia (n = 174, 31.1% of study population) or benign vocal fold lesions (n = 160, 28.6%). The average number of voice therapy sessions ranged from 5.3 (presbyphonia) to 6.7 (benign vocal fold lesions) over the course of 6–7 weeks. Means and standard deviations for number of voice therapy sessions across diagnoses are presented in Figure 1. Number of therapy sessions did not differ statistically across diagnoses. Therapy schedule was weekly for most patients, with only 40 patients waiting two or more weeks between voice therapy sessions. No patients attended more than one session in a week.

Table 1.

Patient demographics and prevalence of comorbidities.

Variable	Diagnoses
	Muscle tension dysphonia	Vocal fold paralysis	Benign vocal fold lesions	Laryngospasm/ irritable larynx	Presbyphonia
Number of patients	174	102	160	68	54
Patient demographics
Age (SD)	51.5 (17.8)	58.7 (14.9)	46.4 (14.9)	48.4 (15.1)	72.3 (10.6)
Sex (female/male)	128/46	53/49	112/48	53/15	33/21
Months since onset (SD)	12.9 (16.1)	12.2 (13)	13.4 (4.8)	22.4 (24.4)	10 (11.8)
Work status (% employed)	68.3%	70.5%	76.2%	72.0%	34.1%
Occupational voice users	52.2%	65.1%	69.0%	30.8%	24.8%
Percentage of smokers	23.5%	32.3%	35.2%	35.1%	35.2%
Percentage of patients reporting comorbidities
Dyspnea	10.3%	11.7%	9.3%	48.5%	12.3%
Dysphagia	5.1%	14.7%	3.7%	6.8%	8.2%
Reflux (patient-reported)	35.0%	26.4%	36.2%	42.6%	28.9%
Cough/throat clear	13.7%	10.7%	15.6%	47.0%	10.4%
Allergies	19.5%	5.8%	19.3%	35.2%	19.5%
Behavioral health	8.0%	4.9%	4.0%	25.0%	8.1%
Patient-reported scale scores
VHI (SD)	31.9 (23.7)	36.3 (10)	41.3 (24.9)	36.5 (20.0)	53.4 (21.4)
GFI (SD)	8.9 (5.2)	8.4 (4.8)	9.07 (5.5)	9.41 (6.3)	6.75 (5.1)
RSI (SD)	15.4 (9.3)	36.5 (23)	12.3 (8.3)	17.0 (15.1)	11.8 (8.7)

Note. VHI scores > 30 indicate moderate or severe voice-related impairment (Jacobson et al., 1997). GFI scores ≥ 4 indicate possible voice problem (Bach et al., 2005). RSI scores ≥ 12 indicate possible reflux (Belafsky et al., 2002). VHI = Voice Handicap Index; GFI = Glottal Function Index; RSI = Reflux Severity Index.

Figure 1.

Means and standard deviations for number of therapy sessions required prior to discharge.

Several differences were observed across diagnostic groups. As might be expected, patients in the presbyphonia group were, on average, significantly older than patients with other diagnoses (p = .02); no other significant differences in age were observed. Patients in the presbyphonia group were less likely to be employed (OR = 0.40, p < .01) or to be occupational voice users (OR = 0.28, p < .01). Patients in the laryngospasm group were also less likely to be occupational voice users, but this did not reach statistical significance (OR = 0.59, p = .06). Laryngospasm patients were more likely to report dyspnea (OR = 4.6, p < .01) or cough symptoms (OR = 3.4, p < .01) and have comorbid behavioral health diagnoses (OR = 3.1, p < .01) than the other groups. Laryngospasm patients also had longer periods of time between symptom onset and diagnosis (p < .01). Patients with vocal fold paralysis were more likely to report dysphagia (OR = 2.8, p < .01) than patients in the other groups. No other statistically significant differences were observed.

Correlations are presented in Table 2. Based on this correlation matrix, linear regression was run as previously indicated. Overall, the regression model had an r ² of .54, indicating that the model accounted for 54% of the variation in the number of voice therapy sessions required prior to discharge. Table 3 contains a statistical summary for the regression model. Age was found to be a significant predictor of number of voice therapy sessions (β = 0.030, t = 2.79, p = .006) and was significantly correlated with the number of voice therapy sessions required (r = .21, p < .001). Patients over 52 years of age required significantly more sessions than those who were younger (p = .04; < 52 = 4.8 sessions, ≥ 52 = 5.7).

Table 2.

Correlations used to determine factors for linear regression.

Factor	Pearson correlation	p value	Factor	Pearson correlation	p value
Age	.214	< .01 *	Laryngeal procedure	.073	.083
Diagnosis	.276	< .042 *	Maximal f o	−.069	.149
Behavioral health diagnosis	−.202	.016 *	Minimal f o	.093	.062
Dysphonia Severity Index	−.291	< .01 *	Mean glottal airflow	.098	.078
Glottal airflow	.222	< .01 *	Maximum phonation time	.033	.489
Overall GRBAS	.529	< .01 *	Noise-to-harmonic ratio	.102	.117
Occupational voice user status	.282	< .01 *	Onset type (gradual, sudden)	.029	.569
Phonation threshold pressure	.343	.006 *	Increased vocal effort	.085	.098
Voice Handicap Index	.248	< .01 *	% reduced voice use	.062	.207
Allergies	−.091	.032*	Reflux (patient report)	−.063	.138
Asthma	−.082	.054	Reflux Severity Index	−.064	.136
Dyspnea	−.102	.017*	Returned following discharge	.163	< .01*
Dysphagia	−.038	.380	Schedule (weekly, other)	.195	< .01*
Employment status	.030	.473	Sex	.105	.415
f o range	.026	.547	Shimmer	.157	< .01*
Glottal Function Index	.058	.232	Subglottal pressure	.098	.078
Globus	−.067	.203	Time since onset	.025	.558
Glottic closure pattern	.065	.183	Throat clearing/ cough	−.021	.622
Glottal vibratory characteristics	.075	.652	Voice Handicap Index (Emotional)	.043	.313
Jitter	.026	.587	Voice Handicap Index (Functional)	.061	.155
Laryngeal airway resistance	.082	.143	Voice Handicap Index (Physical)	.062	.147
Laryngeal hyperfunction	.111	.039*	Years smoked	−.021	.630

Note. Bolded font indicates that factor was included in linear regression. f _o = fundamental frequency; GRBAS = grade, roughness, breathiness, asthenia, strain.

^*Indicates statistical significance at p < .05.

Table 3.

Statistical summary.

Factor	Coefficient	CI [2.5%, 97.5%]	SE	t	p value
Age	0.030	[.009, .051]	.011	2.79	.006*
Diagnosis	0.046	[−.163, .254]	.106	0.432	.666
Occupational voice user	1.55	[.823, 2.29]	.373	4.17	< .001*
Phonation threshold pressure	0.024	[−.055, .104]	.040	0.599	.550
GRBAS	2.65	[2.14, 3.17]	.264	10.08	< .001*
Dysphonia Severity Index	−0.279	[−.351, .208]	.036	7.68	< .001*
Voice Handicap Index	0.021	[.006, .035]	.007	2.85	.005*
Glottal airflow	0.084	[.044, .125]	.021	4.08	.055
Behavioral health diagnosis	−0.339	[−1.73, 1.05]	.709	−0.479	.633

Note. CI = confidence interval; SE = standard error; GRBAS = grade, roughness, breathiness, asthenia, strain.

^* p < .05.

Whether patients reported that their voice use was crucial for their employment was also a significant predictor of number of therapy sessions (β = 1.55, t = 4.17, p < .001). Patients who required their voice for their occupation completed significantly more voice therapy sessions (M = 6.14) than those who did not require voice use for work (M = 4.06, p = .01). Additionally, total VHI score significantly predicted number of sessions (β = 0.021, t = 2.85, p = .005). VHI score was significantly correlated with number of voice therapy sessions needed (r = .248, p < .001), as patients who perceived greater voice-related handicap generally completed more sessions of therapy.

Overall grade of dysphonia as measured by the GRBAS scale (rating for g) significantly predicted voice quality (β = 2.65, t = 10.08, p < .001). GRBAS score was significantly correlated with number of voice therapy sessions required (r = .52, p < .001), as more severe voice quality was associated with more therapy sessions. Additionally, DSI score was also a predictor of number of sessions required (β = −0.279, t = 7.68, p < .001). DSI was inversely correlated with number of sessions needed (r = −.291, p < .001).

Neither diagnosis (β = 0.046, t = 0.432, p = .666), PTP (β = 0.024, t = 0.599, p = .550), mean airflow (β = 0.084, t = 4.08, p = .055), the presence of a comorbid behavioral health diagnosis (β = −0.339, t = −0.479, p = .663) were significant predictors of the number of therapy sessions required.

Of the 558 patients examined in this study, 81 (men = 31, women = 50) returned to the clinic for a second course of therapy following an initial discharge from therapy. The percentage of patients returning for additional therapy across diagnoses was as follows: muscle tension dysphonia = 15%, vocal fold paralysis = 13%, benign vocal fold lesions = 18%, laryngospasm or irritable larynx = 15%, and presbyphonia = 5% (see Figure 2). ORs indicated that patients of no one diagnosis were significantly more likely to return for additional therapy (all p values ≥ .19); however, patients with presbyphonia were 16% less likely to return (OR = 0.84, p = .03). Additionally, occupational voice users were twice as likely to return for additional therapy than those who indicated that voice use was not essential for their work (OR = 2.04, p < .01). There were no statistical differences in age (p = .18) or time since onset (p = .54) between patients who returned for additional therapy and those who did not. Additionally, neither men nor women were significantly more likely to return for additional therapy (p = .38).

Figure 2.

Percentage of patients returning for additional voice therapy following initial discharge across diagnoses.

Discussion

Voice therapy is the first line of treatment for many voice disorders. Even when surgical intervention is indicated, evidence suggests that voice outcomes are optimized when procedures are coupled with voice therapy (Béquignon et al., 2013; Johns, 2003; Rosen et al., 2000). Objective, self-reported, and auditory-perceptual measures of voice indicate that therapy improves vocal function (MacKenzie et al., 2001; Schindler et al., 2012; Speyer, 2008). Considering the importance of voice therapy in managing dysphonia, it is unfortunate that relatively little is known about voice therapy duration. This lack of understanding makes it difficult for SLPs to prepare voice patients for how long they might expect to be in therapy. Additionally, this makes it challenging to study how voice therapy efficiency might be optimized.

This study examined the number of voice therapy sessions needed to achieve treatment goals for patients with muscle tension dysphonia, benign vocal fold lesions, vocal fold paralysis, laryngospasm/irritable larynx, and presbyphonia. Overall, the average number of voice therapy sessions was 5.32 (SD = 3.43). Across individual diagnoses, the number of sessions ranged from 4.3 (presbyphonia) to 6.7 (benign vocal fold lesions) before clinical assessment indicated that discharge from therapy was appropriate (see Figure 1). Dose frequency was one session per week for most patients. Factors that significantly predicted the required number of voice therapy sessions included baseline GRBAS rating, VHI score, DSI, age, and whether the patient felt that their voice was crucial for their job.

The average number of sessions reported in this study differs from previous estimates. For example, Portone-Maira et al. (2011) reported that patients completed an average of 4.8 sessions prior to discharge. Methodological differences could be partially responsible for this finding, as Portone-Maira examined predominantly two clinicians who evaluated patients presenting with a wide range of diagnoses. It is also possible that this difference reflects individual practice patterns. For example, the rate at which clinicians advance patients from SOVTE to resonance voice on the word and sentence level may differ between individuals. Furthermore, some clinicians may directly begin therapy with word- or sentence-level tasks. Alternatively, De Bodt et al. (2015) reported that patients completed an average of 10.87 sessions of voice therapy. This higher estimate may reflect the fact that De Bodt et al. considered all voice-related diagnoses together. As such, patients undergoing treatment for neurological voice disorders (such as the voice symptoms associated with Parkinson's disease) may have inflated averages as they pursued treatments such as LSVT. Such regimens follow intensive therapy schedules, generally consisting of a minimum of 16 sessions over the course of 4 weeks (Sharkawi et al., 2002). Thirty-five patients who completed an average of 10.02 (SD = 3.4) sessions were eliminated from this investigation because they presented with neurological voice problems. This would support the hypothesis that diagnosis was responsible for this difference in findings. It should be noted, however, that patients with neurological voice disorders made up a small portion of the study completed by De Bodt et al. Thus, this difference could also be driven by the fact that De Bodt et al. examined data published between 1975 and 2013—during which time practice patterns likely evolved.

In this study, overall GRBAS ratings were correlated with therapy duration, suggesting that baseline severity of dysphonia predicted the course of voice therapy. This is perhaps not surprising as GRBAS ratings quantify severity of dysphonia from a functional perspective not captured by any single voice measure (Fujiki & Thibeault, 2021a, 2021b). Additionally, auditory-perceptual assessments can predict therapy duration in children (Fujiki & Thibeault, 2022) and are sensitive to the voice changes produced by voice therapy (Speyer, 2008). It is also likely that more severe auditory-perceptual voice assessment results often portend more severe laryngeal pathologies. It may also be that the increased number of sessions to discharge was partially clinician driven—as SLPs were unlikely to discharge a patient that they perceived to be significantly dysphonic.

DSI was a significant predictor of the number of sessions required prior to discharge, supporting the finding that severity of dysphonia affected duration of therapy. This is expected considering that of all the acoustic measures examined, DSI had the strongest relationship (inverse) with therapy duration. DSI is a composite measure calculated from multiple other measures such as vocal f _o range and jitter% (Awan & Roy, 2006; Wuyts et al., 2000); thus, it may have better captured overall vocal function than the individual acoustic measures examined in this study. Future study should include acoustic analyses of connected speech to determine if these measures would yield similar results.

The fact that VHI score predicted the number of sessions needed prior to discharge suggested that the extent to which an individual perceives their life to be disrupted by dysphonia was related to voice therapy duration. The reasons for this finding likely extended beyond vocal function as the VHI is only weakly correlated with dysphonia severity (Fujiki & Thibeault, 2021b) and uncorrelated with treatment adherence (van Leer & Connor, 2015). Rather, the VHI has been shown to reflect the psychosocial effects of dysphonia (Gartling et al., 2021; Maertens & de Jong, 2007), which may vary between patients and voice disorders. Thus, it may be that individuals who perceive greater influence of dysphonia on their social lives and emotional well-being are more willing to complete longer courses of therapy. Additionally, lower baseline VHI scores have been associated with increased likelihood of successful discharge from voice therapy (Smith et al., 2010), suggesting that individuals with higher VHI scores may have more difficulty achieving a functional voice and therefore require more time in therapy. Alternatively, they may hold themselves to a higher standard of what voice quality is acceptable. These issues warrant future study.

Whether individuals reported that voice use was crucial for their occupation was also a significant predictor of the number of voice therapy sessions required before discharge. There is ample evidence suggesting that occupational voice users are at increased risk for vocal pathology (Behlau et al., 2012; Pestana et al., 2017; Roy et al., 2004; Sivasankar, 2002) and likely experience greater vocal demands than individuals less dependent upon their voices for work (Williams, 2003). On average, individuals in this study who relied on vocal function for employment completed more sessions of therapy. These patients were highly motivated to avoid vocal relapse and to prepare for increased vocal load. Thus, it would be understandable if they requested additional therapy sessions because their livelihoods depended—at least to some extent—on achieving successful voice outcomes.

Age influenced the duration of therapy as older patients received more sessions than younger ones on average. This phenomenon was not driven solely by individuals in the presbyphonia group, but by older patients scattered throughout the various diagnoses. This may be because older patients were less likely to have familial and occupational demands that prevented therapy participation. Past study suggests that factors such as number of school-age children in the home may influence voice therapy initiation (Pasternak et al., 2022) and may therefore affect how eager patients are for discharge. Additionally, recent work indicates that rates of cognitive impairment may be as high as 25% in voice patients ≥ 65 (Leclerc et al., 2020), which could partially explain this finding. Cognitive status was not examined in this study but may be an important factor to examine in the future. Alternatively, it may be that these patients were less accustomed to the therapy context, or it could be that age-related tissue changes rendered healing from vocal fold injury and/or learning new patterns of vocal function more difficult (Kendall, 2007). As such, given the more advanced age of the presbyphonia group, it is interesting that these individuals did not require more therapy sessions than patients with the other diagnoses. It could be that the effect of age was counteracted by the lower proportion of occupational voice users in this group. More work is needed to explain this finding.

It was particularly interesting that diagnosis did not predict therapy duration considering that patients with muscle tension dysphonia have been known to achieve dramatic changes in voice quality in a single session (Mathieson et al., 2009; Roy et al., 1997). In the current population, significant gains were often achieved quickly in patients with muscle tension dysphonia; however, additional sessions were completed to prevent recurrence and ensure adequate vocal endurance. The fact that therapy duration was similar across diagnoses likely reflects the fact that patients were learning similar skills regardless of pathology. Thus, a similar amount of time was needed for skill acquisition and mastery.

Aerodynamic voice measures did not predict the number of sessions needed prior to discharge, even though PTP has been observed to predict the duration of voice therapy for children with vocal fold lesions (Fujiki & Thibeault, 2022). This difference in findings may be related to the fact that multiple diagnoses were examined in this study. Individuals with vocal fold paralysis, for example, often employ different airflow patterns than individuals with muscle tension dysphonia (Bielamowicz & Stager, 2006; Zheng et al., 2012). Additionally, as a group, patients with muscle tension dysphonia produce varied results on aerodynamic measures (Belsky et al., 2021). The strongest relationship was observed between therapy duration and glottal airflow (r = .22, p < .01), perhaps because glottal airflow may vary with vocal intensity and quality (Patel et al., 2022). Alternatively, it may be that aerodynamic measures of voice are simply too variable in patients with dysphonia to predict course of treatment (Holmberg et al., 2003). Future study should examine the relationship between aerodynamic measures, diagnosis, and course of treatment.

Between 10% and 15% of discharged patients returned for additional therapy. This return rate was lower than might be expected considering that vocal fold tissue may be more susceptible to the recurrence of lesions once morphological vocal fold changes have occurred (Lee et al., 2021; Lennon et al., 2014). This may be because prolonged inflammation can have downstream impact on vocal fold re-epithelialization and extracellular matrix organization (Welham et al., 2008); and epithelial injury renders laryngeal tissue more susceptible to environmental and mechanical stressors (Levendoski et al., 2014). Even in the absence of morphological vocal fold changes, muscle tension dysphonia—and even laryngospasm—have been documented to be somewhat cyclical in nature (Maceri & Zim, 2001). In fact, as many as 75% patients with muscle tension dysphonia may experience some recurrence of dysphonia following treatment (Dworkin et al., 2000; Roy et al., 1997). The fact that some patients returned to the clinic when dysphonia symptoms returned could be viewed as a positive finding. It was encouraging that patients knew where to seek help and had a sufficiently positive experience in voice therapy to seek more treatment. Still, it is unknown how most of the patients in this study faired over an extended period of time.

As voice therapy duration is a complex issue, several factors should be considered when interpreting these findings. First, the regression model accounted for 54% of the variation in the number of voice therapy sessions required prior to discharge. This r ² is relatively high for this type of study; however, there are clearly other factors that impact how many sessions patients require. In addition, it would be very difficult to capture all the factors that could affect therapy duration, and many of these fell outside the scope of this study. For example, possible influences include client–therapist rapport, familial supports, transportation, and cognitive functioning. Additionally, almost all patients examined in this investigation had health insurance, and thus they may not represent individuals paying for services out of pocket. These and other factors merit future study. It should also be noted that due to the retrospective nature of this study and the large N examined, objective posttherapy data were not available as patients successfully discharged from therapy often did not require repeat voice evaluations. Although the present findings are ecologically valid as they reflect actual patients' course of treatment, future study should include posttherapy voice measures to illustrate therapeutic outcomes. Additionally, future work should examine cepstral/spectral measures of voice, which may better predict therapy course than measures of perturbation. In addition, as the regression model examined multiple measures of vocal function, there was some correlation between factors. Between-factor correlations were weak; however, caution should be employed before interpreting the impact of any one factor in isolation. Additionally, it should be noted that patients in the current investigation did not present with gender-affirming voice goals. As such, future work is needed to examine therapy duration for patients pursing gender-affirming voice services. It should also be remembered that patients came from one treatment site and were discharged from voice therapy. Although this provided consistency in assessments and treatment techniques, additional study is warranted to determine if these findings reflect patterns observed in wider settings.

Conclusions

Patients required an average of 5.32 sessions of voice therapy (depending on diagnosis) before voice outcomes were sufficiently improved for discharge. As such, six sessions is likely a reasonable estimate for clinicians to give patients. A more conservative estimate may be required if patients are occupational voice users or if they present with severe dysphonia at baseline (as measured by GRBAS score or DSI) or a high baseline VHI score. In these cases, it may be safer to estimate eight sessions as 89% of patients required eight or fewer sessions. Overall, 14.5% of patients returned for additional voice therapy following an initial discharge. These data inform our understanding of how many sessions patients with voice disorders require to achieve desired voice outcomes and allow for initial estimates of how long patients can expect to be in treatment. Additional study is needed to optimize the efficacy of voice treatment and determine how recurrence of dysphonia might best be prevented.

Data Availability Statement

The data sets generated during and/or analyzed during the current study are not publicly available due to patient privacy concerns.

Funding Statement

Funding was provided by National Institutes of Health Grant T32-DC009401 supporting the first author (University of Wisconsin–Madison, PI: Susan Thibeault).