Changes in a Daily Phonotrauma Index After Laryngeal Surgery and Voice Therapy: Implications for the Role of Daily Voice Use in the Etiology and Pathophysiology of Phonotraumatic Vocal Hyperfunction

Abstract

Purpose

This study attempts to gain insights into the role of daily voice use in the etiology and pathophysiology of phonotraumatic vocal hyperfunction (PVH) by applying a logistic regression-based daily phonotrauma index (DPI) to predict group-based improvements in patients with PVH after laryngeal surgery and/or postsurgical voice therapy.

Method

A custom-designed ambulatory voice monitor was used to collect 1 week of pre- and postsurgery data from 27 female patients with PVH; 13 of these patients were also monitored after postsurgical voice therapy. Normative weeklong data were obtained from 27 matched controls. Each week was represented by the DPI, standard deviation of the difference between the first and second harmonic amplitudes (H1–H2).

Results

Compared to pretreatment, the DPI significantly decreased in the patient group after surgery (Cohen's d effect size = −0.86) and voice therapy (d = −1.06). The patient group DPI only normalized after voice therapy.

Conclusions

The DPI produced the expected pattern of improved ambulatory voice use across laryngeal surgery and postsurgical voice therapy in a group of patients with PVH. The results were interpreted as providing new objective information about the role of daily voice use in the etiology and pathophysiology of PVH. The DPI is viewed as an estimate of potential vocal fold trauma that relies on combining the long-term distributional characteristics of two parameters representing the magnitude of phonatory forces (neck-surface acceleration magnitude) and vocal fold closure dynamics (H1–H2). Further validation of the DPI is needed to better understand its potential clinical use.

Phonotraumatic vocal hyperfunction (PVH) characterizes a group of voice disorders that show obvious signs of tissue trauma on the medial contact surfaces of the vocal folds (e.g., nodules, polyps; Van Stan et al., 2020). It is believed that phonotraumatic lesions are caused by or associated with aberrant vocal behavior in daily life, such as phonating at inappropriate fundamental frequencies (f _o), producing excessive vocal intensity/sound pressure level (SPL), voicing too often without adequate rest, and/or phonating inefficiently (e.g., voicing produced with higher-than-normal levels of vocal fold collision for a given SPL; Hillman et al., 1989; Karkos & McCormick, 2009; Kunduk & McWhorter, 2009; Leonard, 2009). While patients with phonotraumatic lesions may be offered voice therapy as a primary treatment option (Holmberg et al., 2001), a common treatment approach is surgical removal of the lesions followed by voice therapy (Bouchayer & Cornut, 1992; Zeitels et al., 2002). Postsurgical voice therapy is often recommended because patients are thought to be at significant risk of recurrence based on the assumption that their daily (habitual) phonotraumatic vocal behavior has not been modified (Hillman et al., 1989; Van Stan et al., 2020). Therefore, it is believed that the only curative option for these patients after surgery is voice therapy focused on minimizing persistent phonotraumatic behaviors in daily life.

Most studies that have investigated differences in phonation between patients with PVH and a normative population throughout the course of treatment analyzed short-duration voice recordings in the laboratory (Bouchayer & Cornut, 1992; Cooper, 1974; Hirano et al., 1991; Holmberg et al., 2003; Ju et al., 2013; L. Lin et al., 2014; Petrovic-Lazic et al., 2015; Schindler et al., 2013; You et al., 2017; Zeitels et al., 2002), which may not accurately represent voice use in daily life. This is particularly true with respect to characterizing the cumulative (e.g., daily) impact of phonatory mechanisms believed to contribute to progressive tissue damage associated with phonotraumatic lesions. Furthermore, the studies based on these in-laboratory recordings provide conflicting results for multiple measures of interest (e.g., SPL, f _o, other objective measures). For example, some in-laboraotry studies reported increased f _o variability after surgery and/or postsurgical voice therapy (Bouchayer & Cornut, 1992; Wang et al., 2019; Zeitels et al., 2002), whereas others showed no change (Ju et al., 2013; You et al., 2017) or even decreased f _o variability (Caffier et al., 2017; Petrovic-Lazic et al., 2015; Toran & Lal, 2010; Uloza, 1999)_. Thus, the true impact on daily voice use of combining surgery and voice therapy to treat phonotraumatic vocal fold lesions is not well understood, nor has the relationship between daily voice use and phonotrauma been empirically established.

We have used a custom-designed voice ambulatory monitoring system, called the voice health monitor (VHM), to investigate differences in daily voice use between patients with PVH and matched controls before any treatment (Mehta et al., 2015; Van Stan et al., 2020; Van Stan, Mehta, Zeitels, et al., 2015). The VHM uses a neck-placed accelerometer as the phonation sensor, which connects to an Android smartphone that is used as the data collection platform (Mehta et al., 2012). Most previous studies using ambulatory voice monitoring have failed to find large group-based differences in overall average daily SPL, f _o, cepstral peak prominence, H1–H2, and vocal dose measures (Titze & Hunter, 2015) between patients with PVH and matched vocally healthy controls (Cortés et al., 2018; Ghassemi et al., 2014; Mehta et al., 2015; Van Stan, Mehta, & Hillman, 2015). However, in a recent study, measures extracted from ambulatory data were used to develop a two-feature logistic regression model that classified an unseen test set of 90 subjects (45 pairs of patients with PVH and matched controls) with 76.7% accuracy (area under the receiver operating characteristic curve = 0.84; Van Stan et al., 2020). The two features included in the model were H1–H2 standard deviation (patients had lower variability than controls, especially less variability toward higher values) and SPL skewness (patients were more negatively skewed than controls).

Values of H1–H2 extracted from the neck-surface acceleration signal and inversed filtered oral airflow signal are highly correlated (Mehta, Espinoza, et al., 2019). Thus, both are interpreted similarly as representing the abruptness of vocal fold closure during phonation, with lower values reflecting more abrupt closure and vice versa (Klatt & Klatt, 1990; Mehta, Espinoza, et al., 2019; Stevens, 1998). The more restricted distribution of H1–H2 (smaller standard deviation with less variation toward higher values) for the PVH patient group in a recent study was interpreted to indicate a higher prevalence of more abrupt glottal closure compared to the matched control group (Van Stan et al., 2020).

The other feature of the logistic regression model was SPL skewness, estimated from the distribution of neck-surface acceleration magnitude (NSAM) that was calibrated to SPL (Švec et al., 2005). However, the logistic regression model in this article keeps the NSAM in physical units of vibration (cm/s²) instead of converting to SPL for two reasons: (a) The SPL calibration procedure has a known estimation error as high as ± 5–6 dB, which could introduce more uncertainty (noise) into the model, and (b) the accuracy of the logistic regression is the same when using either SPL skewness or NSAM skewness (test set total classification accuracy = 78.3% and area under the receiver operating characteristic curve = 0.84; Van Stan et al., 2020). The NSAM could be viewed as generally representing the magnitude of laryngeal forces associated with phonation. Therefore, the pretreatment negative skew of NSAM for the patient group in the previous study is interpreted as reflecting the daily accumulation of higher phonatory forces. This interpretation is based on evidence that the NSAM is correlated with low-bandwidth (subglottal pressure; Fryd et al., 2016) and high-bandwidth (peak-to-peak glottal airflow, maximum flow declination rate; Zañartu et al., 2013, 2012) aero-acoustic parameters, as well as mechanical forces generated by the tissue-to-tissue contact associated with vocal fold vibration (Coleman, 1988; Wokurek & Pützer, 2009, 2011, 2013). Thus, the NSAM probably represents a combination of these aero-acoustic and mechanical forces. However, it is currently not possible to determine the relative contribution of each force to the NSAM, as these are hypothetically expected to vary according to how the subject is voicing (e.g., at different f _o, intensities, level of periodicity, amount of vocal fold contact; Jiang & Titze, 1994).

To date, the two-feature regression model has only classified patients with phonotraumatic lesions before treatment (Van Stan et al., 2020). Therefore, it was not possible to determine to what extent the model only captured how patients vocally compensated for the presence of lesions or also detected voice use associated with the cause of phonotrauma. The assumed pathogenesis of phonotraumatic lesions includes the concept of a “vicious cycle” in which worsening tissue trauma triggers increases in compensatory forces (e.g., increased vocal fold adductory forces and subglottal air pressure) that contribute to more trauma (Hillman et al., 1989). Further insights into these cause-and-effect relationships could be gained by tracking treatment-related changes in the ambulatory voice use of patients with PVH. Ambulatory voice monitoring after laryngeal surgery (before voice therapy) could determine the extent to which measures have normalized or are persistently abnormal without the potentially confounding influence of the vocal fold lesions. It is reasonable to assume that measures which normalize (i.e., approach values exhibited by vocally healthy individuals) after surgery are likely to reflect compensatory behavior due to altered voice use in the presence of vocal fold lesions. On the other hand, measures that do not normalize until after postsurgical voice therapy are more likely to reflect voice use that could have been associated with the etiology of the phonotrauma. In the long term, identifying objective ambulatory measures of voice use that are associated with the etiology and pathophysiology of PVH would greatly improve clinical efforts to prevent, diagnose, and treat these disorders.

The current study used an inverse transformation of each probability estimate from the previously developed regression model, referred to as the logit (L), to create a daily phonotrauma index (DPI). This transformation enables the meaningful interpretation of differences/changes between values on the DPI (see Statistical Analysis section in Method).

The purpose of this study is to gain further insights into the role of daily voice use in the etiology and pathophysiology of PVH. This will be accomplished by applying the DPI to predict group-based improvements and individual subject variability in the ambulatory voice use of patients with PVH after laryngeal surgery and/or postsurgical voice therapy. It is hypothesized that the DPI, compared to pretreatment, will be significantly decreased after laryngeal surgery and voice therapy but will only reach the normative range after voice therapy. All data were collected as part of a larger, ongoing project aimed at attaining a better understanding of the etiology and pathophysiology of hyperfunctional voice disorders. The governing institutional review board approved all experimental aspects related to the use of human subjects for this study.

Method

Participants

Only female participants were selected to be in this study to provide a homogenous sample of a group that has a significantly higher incidence of phonotraumatic vocal fold lesions (Goldman et al., 1996; Herrington-Hall et al., 1988). Thirty-four female patients with an initial diagnosis of vocal fold nodules or polyps were recruited through sequential convenience sampling. Diagnoses were based on a comprehensive team evaluation (laryngologist and speech-language pathologist) at the Center for Laryngeal Surgery and Voice Rehabilitation at Massachusetts General Hospital (MGH Voice Center) that included (a) the collection of a complete case history, (b) endoscopic imaging of the larynx, (c) completion of the Voice-Related Quality of Life (V-RQOL) Questionnaire (Hogikyan & Sethuraman, 1999), (d) an auditory-perceptual evaluation using the Consensus Auditory-Perceptual Evaluation of Voice (CAPE-V; Kempster et al., 2009), and (e) aerodynamic and acoustic assessments of vocal function (Patel et al., 2018). During the comprehensive team evaluation, there is a patient-centered discussion between the patient and the clinicians regarding the patient's diagnosis, vocal demands, current vocal functioning, and treatment options (e.g., surgery, therapy, combination of both). Generally, but not always, voice therapy is suggested as the first treatment approach at the MGH Voice Center. Ultimately, it is the patient that chooses her course of treatment after thoroughly discussing the options.

Fifteen patients had tried voice therapy (45- to 60-min sessions, once a week) before opting for surgery, with a mean of eight sessions (ranging from two to 21 sessions). All surgeries were performed using state-of-the-art phonomicrosurgical techniques (Zeitels, 2019; Zeitels et al., 2002). Laryngeal diagnoses of the consented patients were confirmed or revised after surgery. If, during surgery, the lesion diagnosis was changed to something other than nodules or polyp, the patient was terminated from the study (e.g., cyst, papilloma). Also, patients were terminated from the study if additional diagnoses were made during surgery (e.g., sulci). Four patients were terminated when they were diagnosed with vocal fold sulci during surgery. Two patients were terminated when their diagnoses changed during surgery to cyst, and one patient was terminated when her diagnosis changed to papilloma based on histology. Of the 27 remaining patients, 19 were diagnosed with bilateral vocal fold nodules, three were diagnosed with a unilateral vocal fold polyp, three were diagnosed with a unilateral vocal fold polyp and reactive vocal fold nodule, one was diagnosed with bilateral vocal fold polyps, and one was diagnosed with bilateral vocal fold nodules and a left vocal fold polyp. For the 13 patients who completed postsurgical voice therapy, the treatment consisted five sessions (mean) and ranged from two to eight sessions. All participants were engaged in occupations considered to be at a higher-than-normal risk for developing a voice disorder (Verdolini & Ramig, 2001). Patient occupations included 13 singers, two nurses, two psychologists, two consultants, one retired salesperson, one marketer, one administrator, one teacher, one music teacher, one event planner, one physical therapist, and one speech-language pathology assistant. The average (standard deviation) age of the patient group was approximately 29 (11) years (range: 18–57 years).

To serve as matched controls, 27 subjects with healthy voices were recruited through snowball sampling. The snowball sampling approach asked patients enrolled in the study to identify a colleague with no history of voice disorders, approximately the same age (± 5 years), the same sex, and the same occupation. The normal vocal status of all participants in the control group was verified via interview and a laryngeal stroboscopic examination. During the interview, the control candidates were specifically asked if they had any voice difficulties that affected their daily life, and a speech-language pathologist evaluated the auditory-perceptual quality of their voices. If the control candidate indicated voice difficulties or demonstrated a nonnormal voice quality, they were excluded from study enrollment and did not undergo a laryngeal stroboscopic examination. Due to the matching paradigm, the normative group's occupations and age—29 (11)—were the same as the patient group.

Table 1 reports subscale scores for the self-reported V-RQOL and clinician-judged CAPE-V ratings for the participants in the patient group both before and after surgery. There are missing data points underlying the mean ± standard deviation reported in the table since these measures were extracted from a clinical database during the course of standard care (not all measures were taken before/after interventions on all patients). Specifically, the estimates at each time point are composed of the following number of patients: before treatment (V-RQOL based on 26 patients; CAPE-V based on 27 patients), after surgery (V-RQOL based on 15 patients; CAPE-V based on 26 patients), and after therapy (V-RQOL based on eight patients; CAPE-V based on 13 patients). These subjective scales are reported only to generally describe the severity level of the patient group, not for statistical analysis or results reporting. Therefore, reliability was not addressed. V-RQOL scores are normalized ordinal ratings that lie between 0 and 100, with higher scores indicating a higher V-RQOL. CAPE-V scores are visual analog scale ratings that range from 0 to 100, with 0 indicating normality and 100 indicating the most extreme example of deviance for a particular voice quality characteristic. The CAPE-V measurement for each patient came from one rater—the treating speech-language pathologist's single rating during a routine clinical evaluation using the CAPE-V standard reading and sustained vowel samples. Both perceptual scales qualitatively indicate that, compared to before treatment, the patients' voice quality and V-RQOL improved after surgery and therapy, with the largest improvements following surgery.

Table 1.

Patients' mean (± standard deviation) self-reported quality of life impact due to their voice disorder using the Voice-Related Quality of Life (V-RQOL) subscales and the perceived qualities of their voice as judged by a speech-language pathologist using the Consensus Auditory-Perceptual Evaluation of Voice (CAPE-V) form.

Subjective measure	Before treatment	After surgery	After therapy
V-RQOL
Social-Emotional	68.2 ± 27.3	89.3 ± 18.7	96.1 ± 7.0
Physical Functioning	60.5 ± 22.7	85.3 ± 14.9	96.0 ± 3.5
Total score	63.1 ± 20.2	87.1 ± 15.2	94.1 ± 9.6
CAPE-V
Overall Severity	29.4 ± 12.9	10.7 ± 9.9	1.6 ± 3.5
Roughness	20.6 ± 12.1	4.5 ± 8.2	0.8 ± 2.8
Breathiness	15.6 ± 13.4	3.8 ± 6.2	0.5 ± 1.9
Strain	22.7 ± 13.5	7.8 ± 8.1	2.3 ± 3.5

Data Collection

The VHM (Mehta et al., 2012) was used to collect ambulatory voice data on all subjects in the study. The VHM employs a miniature accelerometer (model BU-27135, Knowles Electronics) attached via double-sided medical grade tape to the anterior neck (below the larynx and above the sternal notch) to sense phonation. The sensor is connected to a custom smartphone application as the data acquisition platform, and the system records the unprocessed acceleration signal at 11025-Hz sampling rate, 16-bit quantization, and 80-dB dynamic range to obtain frequency content of neck-surface vibrations up to 5 kHz. The VHM application provides a user-friendly interface for starting/stopping recording and periodic alert capabilities that include system checks.

All participants in the patient group were monitored for 1 week (7 days) at two different points: (a) before any surgical and/or therapeutic intervention and (b) after surgery and before voice therapy (once the laryngologist discontinued the order for complete voice rest). It was recommended that all patients in the study complete a course of postsurgical voice therapy, but only a subset of the patient group (n = 13) began and finished a postsurgical course of voice therapy. This subset of 13 patients was monitored for a third week after postsurgical voice therapy. Each control participant was monitored for one full week.

Data Analysis

The hours-long neck-surface acceleration recordings were divided into nonoverlapping frames of 50 ms in duration. As was done in previous studies (Mehta et al., 2015; Van Stan, Mehta, Zeitels, et al., 2015), each frame was considered voiced if it passed the following thresholds: (a) Amplitude was greater than 45 dB SPL, (b) the first non–zero-lag peak in the normalized autocorrelation exceeded a threshold of .6, (c) f _o (reciprocal of the time lag of the first nonzero autocorrelation peak) was between 70 and 1000 Hz, and (d) the ratio of low- to high-frequency energy exceeded 22 dB. These criteria were needed to eliminate several types of nonphonatory activity such as tapping or rubbing on the sensor, extremely high levels of environmental noise (e.g., very loud music), and electrical interference/artifacts.

As this study used the previously developed regression model (Van Stan et al., 2020), the only two features extracted from all voiced frames were NSAM and H1–H2. To calculate NSAM, the root-mean-square of each 50-ms frame was transformed into physical units of cm/s² according to the linear mapping obtained from a calibration procedure for the specific miniature accelerometer used during recording. The calibration procedure was completed once for each accelerometer, where the accelerometer was calibrated to a reference accelerometer (4533-B, Brüel & Kjær) by applying a known chirp vibration signal covering the 10- to 5000-Hz spectrum using an electrodynamic vibration exciter (Mini-Shaker Type 4810, Brüel & Kjær) on a vibration isolation table (BT-2024, Newport Corp.). To calculate H1–H2, each 50-ms frame underwent one discrete Fourier transform. The H1–H2 for each frame was defined as the difference (in dB) between the amplitudes of the first and second harmonics in the frequency spectrum.

Statistical Analysis

Instead of weeklong estimates of voice use—as was done in previous studies (Mehta et al., 2015; Van Stan et al., 2020; Van Stan, Mehta, Zeitels, et al., 2015)—the summary statistics included in the regression model (NSAM skew and H1–H2 standard deviation) were computed from daylong distributions. Daily summary statistics (unlike weekly summary statistics) preserved individual subject variability at a daily level, and linear mixed-effects regression models provided adjusted group means that take this daily variability into account. Linear mixed-effects regression models were used to analyze the results across four repeated observations (0 = control, 1 = pretreatment, 2 = postsurgery, and 3 = posttherapy). To take full advantage of the matched patient–control study design, each control subject was used as the “0” observation for their matched patient. The models assessed differences among the regression values of the patient group (pretreatment, postsurgery, or posttherapy) and the normative group, as well as differences across stages of treatment (pretreatment vs. postsurgery, pretreatment vs. posttherapy, and postsurgery vs. posttherapy).

The regression analysis was applied to each individual patient's daily data to test the model (i.e., validation) and demonstrate overall improvements (movement toward normal classification) across treatment. The logistic regression model uses a daily estimate of NSAM skew and H1–H2 standard deviation to classify a subject's day on a probability scale of 0–1. On this scale, data are considered to be from a patient with probabilities ≥ .5 and from a control with probabilities < .5. The probability (p) of a subject's data being classified as coming from a patient or not results from a logistic transformation of the patient's daily NSAM skew (N _Skew) and H1–H2 standard deviation (H _SD), represented in Equation 1:

$$\rho =\frac{1}{1+e^{-\left({}^{\ast }-4.195\times N_{\text{Skew}}-1.254\times H_{\mathrm{SD}}+6.729\right)}}.$$ (1)

Both NSAM skew and H1–H2 standard deviation contributed nearly equally to classification with standardized odds ratios of 0.384 and 0.309, respectively. In other words, when NSAM skew or H1–H2 standard deviation increased 1 SD, the subject was 2.6 or 3.2 times more likely to be a healthy control subject, respectively. When assessing changes across treatment in the current study, the logit was used instead of the probability. As shown in Equation 2, the logit (L) is an inverse transformation (i.e., link function) of the nonlinear probability (p) estimate:

$$L=log\frac{\rho }{\left(1-p\right)}.$$ (2)

The logit was used because changes in L (ΔL) can be interpreted equally throughout the scale. In contrast, this is not true for the probability, where, for example, a reduction in p from .99 to .98 (ΔL = −.31) represents a much larger improvement than a reduction in p from .50 to .49 (ΔL = −.02). Note that negative values of L represent voice use in the normative range. The more readily interpretable logit values (L) comprise what is referred to as the DPI.

Statistical significance was based on effect sizes representing clinically meaningful differences instead of the traditional alpha value of .05. As in previous investigations (Van Stan et al., 2020), clinically meaningful differences were considered to be medium-to-large effect sizes (Cohen's |d| ≥ 0.5; Cohen, 1988). All statistics were completed using R 3.5.0 (R Core Team, 2018). The model-derived means and effect sizes were calculated using the emmeans package (Version 1.4.4; Lenth et al., 2020). Patterns of change in the DPI were only labeled as associated with treatment if three conditions were satisfied: (a) Pretreatment difference: There was a medium-to-large difference between patients and controls before any treatment. (b) Treatment difference: There was a medium-to-large difference between patients before and after the intervention or combination of both interventions. (c) Direction of treatment difference: The medium-to-large differences between patients before/after intervention were in the direction of the control group.

Results

Across all patient and control subjects, the monitoring system was worn for a mean duration of 73 hr 14 min (SD = 18 hr 6 min), and the subjects phonated for a mean duration of 6 hr 52 min (SD = 2 hr 45 min). The DPI significantly decreased (i.e., toward normal classification) across treatment for the patient group. Compared to before treatment, the patient group's DPI decreased after laryngeal surgery (adjusted mean ΔDPI = −0.66, d = −0.86) and after voice therapy (adjusted mean ΔDPI = −0.82, d = −1.06). Compared to the control group's average DPI (adjusted mean DPI = −0.41), the patient group was significantly higher pretreatment (adjusted mean DPI = 0.63, d = 1.36) and postsurgery (adjusted mean DPI = −0.03, d = 0.50), but not significantly different posttherapy (adjusted mean DPI = −0.19).

Figure 1 displays the performance of the DPI on an individual patient basis as their voice use was monitored before and after treatment for their voice disorder. Nearly all patients (25 of 27 patients) exhibited a reduction in DPI following laryngeal surgery and/or voice therapy (i.e., they ended below the dashed diagonal line). Fourteen of those patients exhibited such a large reduction that their DPI values following treatment were negative and thus demonstrated conversion to normative values (points located in the lower right quadrant of the graph). Nine of the patients with positive DPI values exhibited a reduction following treatment but did not cross the threshold of normality (their DPIs were still positive). One patient (represented by an “×” in the upper middle of Figure 1) exhibited an unexpected increased DPI following laryngeal surgery, which was followed by an expected decrease in DPI following successful voice therapy. Four patients exhibited negative DPI values and were thus classified as being in the normative range before any voice treatment was provided (lower left side of Figure 1 near the gray box).

Figure 1.

Scatter plot of the logits from the daily phonotrauma index (DPI) are plotted on the x-axis (pretreatment) and y-axis (postsurgery and/or posttherapy). Normal classification at all time points is represented as a gray box. No change across treatment is represented as a gray, diagonal, dashed line. Any data points above or below this diagonal line represent worsening or improvement of the logit, respectively. The cutoff between normal and patient is represented as a gray, horizontal, solid line; that is, when a marker is below this line, the patient's presurgery data were classified as “patient” and the patient's postsurgery data were classified as “control.” Each circle represents a single patient's average DPI (mean across 7 days of monitoring). One patient who significantly “worsened” after surgery and significantly improved after voice therapy is represented as an ×. The arrows represent the direction of change in the DPI from postsurgery (beginning of arrow) to posttherapy (end of arrow). Downward-pointing arrows represent improved classification, and upward-pointing arrows represent worse classification. Markers without arrows represent patients who were only monitored before and after laryngeal surgery (not after postsurgical voice therapy).

Discussion

The two purposes of this study were to gain further insights into how daily voice use may contribute to (a) the onset of vocal fold lesion formation (i.e., etiology of PVH) and (b) continued tissue damage because of compensatory adjustments to counteract diminished phonatory function associated with the presence of lesions (i.e., pathophysiology of PVH). Since patients already had phonotraumatic lesions when recruited for this study, inferences about etiological and pathophysiological mechanisms relied on the sequencing of treatment-related changes in daily voice use as measured by the DPI, as well as comparisons with the normal control group. We hypothesized that, compared to pretreatment, the DPI would significantly decrease toward normal classification after laryngeal surgery and voice therapy, but only decrease into the normative range after voice therapy.

The DPI performed well on data it had never seen (e.g., all of the postsurgical and posttherapy data). Consistent with our hypotheses, compared to pretreatment, the mean patient DPI significantly decreased after laryngeal surgery and voice therapy. Furthermore, significant differences between the patient and control groups only disappeared after voice therapy. The significantly decreased DPI after surgery (before voice therapy) is most likely associated with a reduction in compensatory adjustments that are no longer necessary because the vocal fold lesions had been eliminated by surgical removal. The fact that the DPI for patients postsurgery remained significantly higher than the control group (−0.03 vs. −0.41, respectively) supports the assumption that there was a persistence of vocal hyperfunction prior to voice therapy, probably reflecting conditions that contributed to the formation (etiology) of phonotraumatic lesions. This interpretation gains additional support from the finding that the difference in DPI between the patient and control groups disappeared only after therapy (posttherapy mean L = −0.19; i.e., therapy is interpreted as being responsible for reducing the persistent postsurgical vocal hyperfunction). Finally, even further support was provided by the results for individual patients, as seen in Figure 1, where most of the patients (10 out of 13) improved after therapy according to the DPI (arrows pointing downward).

The two measures that comprise the DPI, H1–H2 standard deviation and NSAM skew, did not individually improve as hypothesized. Specifically, H1–H2 standard deviation increased and normalized after surgery alone, whereas NSAM skew only improved after surgery and did not normalize. This contrasts with the DPI results, which combined the two measures in a way that showed significant improvement after surgery and voice therapy compared to pretreatment, as well as normalization only after therapy. Based on these findings, we currently consider DPI to be a measure of the potential for vocal fold trauma that relies on combining the long-term (daily) distributional characteristics of two parameters (H1–H2 and NSAM) that represent the magnitude of phonatory forces (NSAM) and vocal fold closure dynamics (H1–H2). Therefore, it is reasonable to view DPI as reflecting the relative level of PVH.

The interpretation of NSAM can potentially be further refined through future work that identifies the underlying individual/relative contributors of various aero-acoustic and mechanical forces. This could include aerodynamic modeling (e.g., subglottal impedance-based inverse filtering; Zañartu et al., 2014, 2013), careful examination of the relationship between subglottal pressure and NSAM (Fryd et al., 2016; J. Z. Lin et al., 2019; Marks et al., 2019), and simultaneous measurements of intraglottal vocal fold tissue contact and NSAM during voicing (Gunter et al., 2005; Jiang et al., 2001; Mehta, Kobler, et al., 2019; Motie-Shirazi et al., 2019).

Of note, although this study focused on using the DPI to measure treatment-related changes in voice use, a host of other traditionally used objective voice measures were tested with this data set, and none of them was significantly associated with treatment except f _o variability. Ambulatory f _o variability metrics (standard deviation, mid-90% range, 95th percentile) differed from normal before treatment and normalized only after voice therapy—although none of the f _o variability measures made it into the DPI regression model. Surprisingly, the patients did not exhibit a medium-to-large improvement in f _o variability after surgery despite a presumptive increase in physiological ability to voice at higher f _o without the lesions. This may be due to the clinical observation that patients often display guarded/overly careful voice use during their postsurgery appointments (e.g., talking softer, stating that they don't want to “overdo it” despite being cleared to return to full voice use). The improved f _o variability after therapy could be expected since many voice treatment approaches incorporate pitch glides, vocalizes, and other variations in pitch during practice (e.g., see Stemple et al., 1994; Verdolini-Marston et al., 1995).

The other objective measures tested were SPL and NSAM (mean, median, mode, standard deviation, 5th percentile, 95th percentile, kurtosis), f _o (mean, median, mode, 5th percentile, skew, kurtosis), cepstral peak prominence (mean, median, mode, 5th percentile, 95th percentile, mid-90th range, skew, kurtosis), H1–H2 (mean, median, mode, skew), vocal doses (percent phonation, cycle dose, distance dose), and logarithmically spaced bins of phonatory as well as nonphonatory segments (Titze et al., 2007). Although none of these measures were associated with changes across treatment, they may still be useful in characterizing clinically important behaviors in individual patients or subgroups of patients; for example, patients who are 3 SDs away from a normative group on any singular measure. One reason for the lack of differences associated with treatment in these statistics could be that the metrics do not individually estimate the amount of vocal fold contact or collision during voicing, which is the hypothesized causative and/or associative feature of phonotraumatic lesions. Thus, these results call for future work to develop measures (especially vocal doses) that incorporate key etiological factors of phonotrauma, such as estimates of vocal fold collision or additional vocal fold closure parameters.

Patients in this study received state-of-the-art treatment, and outcomes were generally positive. However, because strict protocols for surgery and voice therapy were not established/used, this study was not intended to be a formal study of treatment efficacy, effectiveness, or clinical management of patients with PVH. However, future work should not only continue to apply the DPI to gain further insights into etiological/pathophysiological mechanisms associated with PVH but should also serve to further validate DPI as a potential clinical tool. For example, the results for the patient group in this study could be compared to a group of patients who successfully completed voice therapy without needing laryngeal surgery. The hypotheses would be that (a) the therapy-only group would exhibit lower DPI values (i.e., less severe PVH) before treatment than the group that needed surgery and (b) the therapy-only group would exhibit higher DPI values after therapy (because they still have lesions) than the postsurgical therapy group. Also, these comparisons would have clinical implications. For example, if higher DPI values are associated with patients who eventually undergo laryngeal surgery, the individual patient's value could be useful in the discussion of whether to attempt surgery before therapy. Also, patients with high DPI values after surgery may require more extensive voice therapy compared to those with lower values (e.g., longer treatment course and more voicing practice vs. shorter treatment course and primarily vocal hygiene education, respectively). Interestingly, of the 13 subjects that received postsurgical voice therapy, patients with postsurgical DPI values well below the cutoff (i.e., classified as “normal”) had a lower number of voice therapy sessions compared to those with postsurgical values around or above the threshold (i.e., classified as “patients”; two to four vs. four to eight voice therapy sessions, respectively). Finally, the DPI could have potential in helping to determine when to stop voice therapy and in predicting which patients are at high risk for lesion recurrence or require additional voice therapy. To test these hypotheses, long-term outcomes (e.g., at 6 months posttherapy) could be correlated with DPI at the end of therapy to determine the predictive power of the estimates; that is, do the DPI values at the end of therapy predict which patients show signs of recurrence of phonotrauma 6 months later?

It is possible that the diagnosis of vocal fold pathology and/or the monitoring itself could have changed the patient's or the control's typical behavior, confounding the results from any ambulatory monitoring study (Hunter, 2012). However, dramatic changes in behavior due to the presence of monitoring seem improbable as patients often require extensive voice therapy to modify their habitual behaviors (Ziegler et al., 2014). Also, subjects often reported forgetting that they were wearing the device. A subset of the patients (n = 15) received voice therapy before surgery and could have improved from this first dose of behavioral treatment, but not significantly enough to avoid surgery. This study did not take into account any presurgical voice therapy changes, which—assuming voice therapy always has some sort of positive vocal impact—could have inflated the surgical changes. The data set contains a large number of occupational singers (14 of the 27 patients), which may limit the study's external validity to patients with PVH who are not singers. In response to this, Table 2 illustrates the DPI over treatment across all of the patients and two subgroups (singers and nonsingers). There were no large effect sizes between the two groups, indicating that these two potential subgroups (in a preliminary sense) are not obviously different in the DPI or the pattern of DPI changes across treatment. Currently, we have developed an automatic singing detector (Ortiz et al., 2019) to investigate the effect of singing on differences (and lack of differences) observed between patients and matched healthy controls. Interestingly, the singing detector typically detects a relatively small amount of total voicing (< 2% phonation time; Toles et al., 2020), potentially indicating a weak influence on the data outlined here. In regard to potentially confounding factors, there is likely an interaction of the patient's individual disposition or personality with their etiological and/or compensatory voice use (Roy & Bless, 2000; Roy et al., 2000a, 2000b), and future work could investigate how personality interacts with the DPI.

Table 2.

Daily phonotrauma index (DPI) mean (standard deviation) values and change in values for all patients and two subgroups of patients (singers vs. nonsingers) before and after surgery (n = 27, 14, 13, respectively), as well as after postsurgical voice therapy (n = 13, 10, 3, respectively).

Measures	All patients	Singer subgroup	Nonsinger subgroup
DPI values (logit)
Before surgery	0.62 (0.64)	0.41 (0.57)	0.84 (0.66)
After surgery	–0.05 (0.69)	–0.17 (0.39)	0.09 (0.92)
After therapy	–0.22 (0.63)	–0.31 (0.66)	0.07 (0.52)
DPI change scores (delta logit)
After–before surgery	–0.66 (0.68)	–0.58 (0.49)	–0.75 (0.86)
After surgery–after therapy	–0.19 (0.31)	–0.14 (0.29)	–0.35 (0.39)
After therapy–before surgery	–0.69 (0.52)	–0.70 (0.59)	–0.65 (0.26)

Note. “All patients” data will not exactly match the values in the text as the in-text values are “adjusted” by a multilevel model. There were no large effect sizes between the three groups (Cohen's d ≥ 0.8).

Conclusions

The DPI (using NSAM skewness and H1–H2 standard deviation) produced the expected pattern of improved ambulatory voice use across laryngeal surgery and postsurgical voice therapy in a group of patients with PVH. Specifically, group-based logits significantly decreased toward normal after surgery and normalized only after voice therapy. The DPI can be viewed as a measure of the potential for vocal fold trauma that relies on combining the long-term (daily) distributional characteristics of two parameters that represent the magnitude of phonatory forces (NSAM) and vocal fold closure dynamics (H1–H2). Further application/validation of the DPI is needed to better understand its potential use in clinical decision making.

Funding Statement

This work was supported by the Voice Health Institute and the National Institute on Deafness and Other Communication Disorders under Grants R33 DC011588 (PI: Robert Hillman) and P50 DC015446 (PI: Robert Hillman). The article's contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.