Correlation of VHI-10 to Voice Laboratory Measurements Across Five Common Voice Disorders *Amanda I. Gillespie, †William Gooding, *Clark Rosen, and *Jackie Gartner-Schmidt, *yPittsburgh, Pennsylvania Abstract: Objective. To correlate change in Voice Handicap Index (VHI)-10 scores with corresponding voice laboratory measures across five voice disorders. Study Design. Retrospective study. Methods. One hundred fifty patients aged >18 years with primary diagnosis of vocal fold lesions, primary muscle tension dysphonia-1, atrophy, unilateral vocal fold paralysis (UVFP), and scar. For each group, participants with the largest change in VHI-10 between two periods (TA and TB) were selected. The dates of the VHI-10 values were linked to corresponding acoustic/aerodynamic and audio-perceptual measures. Change in voice laboratory values were analyzed for correlation with each other and with VHI-10. Results. VHI-10 scores were greater for patients with UVFP than other disorders. The only disorder-specific correlation between voice laboratory measure and VHI-10 was average phonatory airflow in speech for patients with UVFP. Average airflow in repeated phonemes was strongly correlated with average airflow in speech (r ¼ 0.75). Acoustic measures did not significantly change between time points. Conclusions. The lack of correlations between the VHI-10 change scores and voice laboratory measures may be due to differing constructs of each measure; namely, handicap versus physiological function. Presuming corroboration between these measures may be faulty. Average airflow in speech may be the most ecologically valid measure for patients with UVFP. Although aerodynamic measures changed between the time points, acoustic measures did not. Correlations to VHI-10 and change between time points may be found with other acoustic measures. Key Words: Voice–Voice laboratory–Voice handicap–Acoustic–Aerodynamic. INTRODUCTION This study represents the first in a series of studies aimed at creating evidence-based, disorder-specific, patient-centered voice laboratory evaluation protocols. At specialized voice clinics,1 regardless of an individual’s voice symptom or disorder, a standard voice laboratory evaluation includes three component parts: (1) acoustic and aerodynamic voice analyses; (2) auditory-perceptual evaluation; and (3) patient selfevaluation of voice handicap or quality of life.2 These three components of voice laboratory testing theoretically measure different aspects of vocal function but are presumed to be related. However, questions about the reliability and validity of each of the evaluation components as both diagnostic tools and outcomes measures have been raised, making it difficult for clinicians to determine which measurement is most relevant in the management of patients with voice problems.3,4 Furthermore, it is unknown how each laboratory measure correlates with change in a patient’s perception of voice handicap perceptions over time. In general, acoustic and aerodynamic voice laboratory measurements have demonstrated sufficient sensitivity to differentiate grossly normal from dysphonic voices.5–7 More specifically, these laboratory measurements have greater Accepted for publication October 30, 2013. From the *Department of Otolaryngology, University of Pittsburgh Voice Center, University of Pittsburgh Medical Center Mercy, Pittsburgh, Pennsylvania; and the yBiostatistics Facility, University of Pittsburgh Cancer Institute, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania. Address correspondence and reprint requests to Amanda I. Gillespie, Department of Otolaryngology, University of Pittsburgh Voice Center, University of Pittsburgh Medical Center Mercy, 1400 Locust Street, Suite 11-500, Building B, Pittsburgh, PA 15219. E-mail: [email protected] Journal of Voice, Vol. -, No. -, pp. 1-9 0892-1997/$36.00 Ó 2013 The Voice Foundation http://dx.doi.org/10.1016/j.jvoice.2013.10.023

sensitivity for identifying some voice disorders (eg, vocal fold lesions, vocal fold paralysis) than others (eg, primary muscle tension dysphonia [MTD]8,9). These disorder-specific findings provide evidence for the argument that all voice laboratory measures may not be sensitive and appropriate for use in evaluating all voice disorders. Voice problems are typically non-life threatening but do cause a substantial impact on an individual’s quality of life.10 How a patient feels about his/her voice is one of the determining factors in treatment seeking, compliance, and discharge.11,12 How a patient feels about his/her voice has also shown to positively correlate with auditory-perceptual evaluations of voice.13,14 The Voice Handicap Index (VHI) is a widely used patientperceptual questionnaire that assesses the handicapping effects of a voice problem on an individual’s life.15,16 The VHI-10 is a 10-question adaption of the original VHI and has demonstrated similar sensitivity in assessing voice handicap.17 A trend exists in the literature of using patient-perceptual measurements such as the VHI or VHI-10 as the gold standard for assessing voice change.18–20 In light of this trend, numerous studies have attempted to correlate VHI score with objective voice measurements. These studies have identified good correlations between some acoustic and aerodynamic measures and the three subscales of the VHI (functional, emotional, and physical) for functional voice disorders, vocal fold paralysis, and vocal fold lesions.21 However, no correlation was found between the total VHI and acoustic or aerodynamic measures.21,22 These correlations—or lack of correlations—between the VHI and objective voice laboratory measures are interesting in two ways. First, each of the reviewed studies correlated voice laboratory measures with the subscales of the VHI. Any findings based on subscales are questionable due to research negating the existence of the subscales as three distinct

2 factors.17 Therefore, the correlation between various voice laboratory measurements and a subscale of the VHI may not stand as valid correlations. Second, and most importantly, these studies may provide evidence that a ‘‘one size fits all’’ approach to voice analysis is not ideal.23 Stated differently, it may be that for certain voice disorders, the VHI-10 and objective laboratory measures do correlate and for other disorders, they do not. To the best of this author group’s knowledge, no study has correlated change in quality of life and voice laboratory measures for specific voice disorders. Using the VHI-10 as an established measure of voice handicap 17 to investigate corresponding change values in auditory perceptual, acoustic, and aerodynamic measures across five common voice disorders may help improve understanding of the relevancy of certain voice laboratory measures to the handicapping effects of specific voice disorders. In addition, the authors know of no-voice laboratory protocol for acoustic and aerodynamic assessment that recommends specific measures for specific voice disorders. The present study aimed to correlate VHI-10 change scores with corresponding voice laboratory changes across five common voice disorders: vocal fold lesions (lesions), primary MTD-1, vocal fold atrophy (atrophy), unilateral vocal fold paralysis (UVFP), and vocal fold scar (scar). The hypothesis was that some voice laboratory measures would reflect corresponding perceptions of vocal handicap for certain disorders, but that due to the physiological variability of each voice disorder, all laboratory measures would not correlate with VHI-10 for all disorders. MATERIALS AND METHODS All study procedures were approved by the University of Pittsburgh Institutional Review Board (IRB #PRO12050438). Participants The investigation involved a retrospective study of patients presenting to the University of Pittsburgh Voice Center (UPVC). Informed consent was obtained from all patients before the data were entered into a clinical research database. The database was queried by a research coordinator, blinded to the experimental hypotheses, for patient records from 2009 to 2011 for the following inclusion criteria: aged >18 years, primary diagnosis of primary MTD (MTD-1), benign midmembranous vocal fold lesion(s) (lesions), UVFP, vocal fold atrophy (atrophy), or vocal fold scar (scar). Diagnosis was made by a multidisciplinary team consisting of a fellowship-trained laryngologist and a voice-specialized speech-language pathologist (SLP). Because many patients carry more than one diagnosis (eg, UVFP and atrophy), great care was taken in isolating participants with a single voice disorder diagnosis. Each participant with a mutually exclusive diagnosis was cross-referenced with the electronic medical record to display longitudinal serial VHI-10 values for all the participant’s subsequent clinical visits. For each disorder group, the participants with the largest change in VHI-10 between two separate periods were selected for inclusion, until 30 participants were identified in each disorder group. Because the study was not designed to

Journal of Voice, Vol. -, No. -, 2013

look at treatment effects but rather aimed to investigate the correlation between voice laboratory measure change scores and VHI-10 change scores, a change in VHI-10 score was interpreted to indicate that a patient-perceived change in voice handicap had occurred. It was, therefore, presumed that the largest VHI-10 change would provide the best opportunity to observe a correlation with change in voice laboratory measures, should one exist. The dates of these two VHI-10 values were then linked to corresponding audio-perceptual, acoustic, and aerodynamic voice laboratory measures taken during the same clinic visit. Based on the estimates from the past research on acoustic and aerodynamic voice measures with an alpha of .05, a sample size of 30 participants in each voice disorder group was needed to achieve 80% statistical power to detect a significant correlation with a modest effect size (between 0.45 and 0.50). Thirty participants were identified in each of the five voice disorders for a total n ¼ 150 participants. Procedures The following voice laboratory measures were obtained from routine clinical visits. Audio-perceptual evaluation. The Consensus AuditoryPerceptual Evaluation–V (CAPE-V)24,25 was used to judge each participant’s overall perceptual voice severity. Using a head-mounted microphone positioned at an approximate 45 angle from the mouth (Shure Beta 54 WBH54; SHURE, Chicago, IL), patients read the standard CAPE-V sentences in their most comfortable pitch and loudness (MCPL). SLPs specialized in voice used a visual analog scale to rate the participant’s overall dysphonia severity. CAPE-V ratings were made live at the time of recording in a sound-treated room during the diagnostic clinic evaluation and follow-up time points. Instrumental measure I: Acoustic measures. Participants completed two tasks: (1) sustained /a/ vowel at MCPL and (2) spoke the standardized CAPE-V sentences at MCPL. The Multi-Dimensional Voice Program from the Computerized Speech Laboratory (KayPENTAX, Montvale, NJ) was used to capture acoustic data. Before analyzing the sustained vowel, visual perceptual confirmation of type I acoustic signal was performed.26 After verification of a stable fundamental frequency (F0), the phonated /a/ was selected for the analysis of noise-toharmonic ratio (NHR). This measure was chosen as a primary time-based acoustic analysis because it incorporates components of two other commonly used measurements, jitter and shimmer.27 The next acoustic measurement was the F0 of the all voiced CAPE-V sentence: We were away a year ago. Instrumental measure II: Aerodynamic measures. The voice laboratory protocol for obtaining aerodynamic measures including laryngeal resistance (Rlaw) and its component parts—average phonatory airflow (L/s) divided by estimated subglottal pressure (Psub) (cmH2O) changed within the 3-year window of data collection for this study. Therefore, all aerodynamic measures were reanalyzed for inclusion in the present study. Aerodynamic data were collected with the PAS6600 (KayPENTAX). The participant’s nose and mouth were fitted in a facemask coupled to a pneumotach with an intraoral pressure

Amanda I. Gillespie, et al

Correlation of VHI-10 to Voice Laboratory Measures

tube placed between the lips. Vocal intensity (dB SPL) was captured simultaneously by an integral KayPENTAX factory– calibrated microphone. Participants produced a string of five consonant-vowel (CV) syllables (/pa pa pa pa pa/) on one breath, at a rate of approximately 1.5 syllables/s. Two of the coauthors (A.I.G. and J.G.S.) analyzed the middle three tokens of the five CV syllable string. The morphology of pressure peaks and airflow plateaus were visually inspected for the string. Acceptable morphology was defined as minimum flow corresponding with maximum pressure.28 Minimum flow was ensured to return to zero during the measurement of peak pressure. The middle three pressure peaks generated during the /p/ of each fivesyllable /pa/ string were manually selected via cursor placement.29 The corresponding flow signals of the vowel /a/ were also manually selected via cursor. These methods reflect those that have been well investigated in the literature.29,30 Fifteen percent of the aerodynamic data did not meet eligibility criteria. Those data were then discarded and replaced, per the inclusion criteria, until 30 participants with acceptable data were identified in each of the five voice disorders. Statistical methods VHI-10 scores were tested for equality among disorders by a one-way analysis of variance. Pairwise contrasts were tested by Tukey honest significant difference test. Due to some extreme values in change scores of laboratory measures, delta (Time A  Time B) VHI-10 was tested with the Kruskal-

3

Wallis test. Change from Time A (TA) to Time B (TB) was tested with the signed rank test. The associations between nine laboratory measurements and VHI-10, as well as change scores, were examined graphically with an estimate of the Spearman rank correlation coefficient. Formal estimates of association between change in laboratory measure and delta VHI-10 were conducted with a three-step analysis of covariance (ANCOVA). Initially, an omnibus test of significance of all regression slopes was conducted. This test found an optimal functional form for regression by differentiating between a linear function and 3 degrees of freedom restricted cubic spline. Next, an interaction test was conducted to decide whether the data were consistent with equal slopes among disorders or whether different slope estimates were needed. Finally, slopes were tested to determine if they were parallel or coincident. RESULTS Inclusion/exclusion The database query identified n ¼ 4020 individuals aged older than 18 years for further evaluation of inclusion/exclusion criteria. Figure 1 depicts the path of each potential participant from identification in the database through data analysis.31 Four hundred sixty-five patients were identified with a single primary diagnosis. Patient files were reviewed in detail to confirm complete voice laboratory data sets until 30 complete data sets in each voice disorder category were identified.

FIGURE 1. Flow chart of all potential participants from identification through final data analysis.

4

Journal of Voice, Vol. -, No. -, 2013

TABLE 1. Number of Participants in Each VHI-10 Category From Time A to Time B VHI-10 Score 0–10 (normal range) 11–20 21–30 31–40

FIGURE 2. Box and whisker plot of VHI-10 by disorder. Two VHI10 scores for each of 150 subjects were measured for providing 300 total scores by disorder. VHI-10 scores differed by disorder (P ¼ 0.0011) with VHI-10 score highest for subjects presenting with UVFP.

Reliability Reproducibility of estimated Psub (cmH2O) and average phonatory airflow (L/s) for 10% of the aerodynamic data resulted in an interclass correlation coefficient of P ¼ 0.953 for Psub and P ¼ 0.998 for average phonatory airflow. Voice Handicap Index-10 The median VHI-10 score for all patients at all time points was 17, which connotes a voice handicap (>11).32 Paralysis patients, regardless of time point, had significantly greater VHI-10 scores than the other disorder groups (P < 0.05). The VHI-10 scores of the four other groups (MTD-1, atrophy, scar, and lesions) were not statistically different (P > 0.30, for all comparisons) (Figure 2). Table 1 shows the distribution of change in VHI-10 by quartile category. Twenty-nine subjects had TA VHI-10 scores within the normal range ( 0.01) Spearman correlation coefficient. Only one, overall severity, exhibited a moderately strong correlation. Using a parametric analysis, the regression of each laboratory measure on VHI-10 was examined dependent

TABLE 2. Median Change in Each Voice Laboratory Measure From Time A to Time B by Disorder Measure Psub at MCP Intensity in speech Average airflow at MCP Laryngeal resistance Average airflow in speech Intensity at MCP F0 in speech NHR Overall severity (CAPE-V)

Atrophy

Lesions

MTD-1

Paralysis

Scar

Signed Rank, P*

Kruskal-Wallis, Py

0.61 0.44 0.02 1.24 0.02 0.70 3 0.01 16

0.44 0.39 0 0.84 0.03 0.39 6 0.01 8

0.81 2.13 0.02 4.17 0.01 0.88 2.5 0 5

0.76 0.95 0.07 14.70 0.12 0.72 3.5 0.01 17

0.02 0.93 0.01 4.10 0.02 0.74 1.5 0 11

0.496 0.716 0.0002 0.0028 0.0294 0.8567 0.5386 0.0621