The Laryngoscope C 2014 The American Laryngological, V
Rhinological and Otological Society, Inc.
Longitudinal Variations of Laryngeal Overpressure and Voice-Related Quality of Life in Spasmodic Dysphonia Jeffrey C. Yeung, MD; Kevin Fung, MD, FRCSC; Eric Davis, BHSc; Sunita K. Rai, BHSc; Adam M. B. Day, PhD; Agnieszka Dzioba, MSc; Catherine Bornbaum, PhD; Philip C. Doyle, PhD Objectives/Hypothesis: Adductor spasmodic dysphonia (AdSD) is a voice disorder characterized by variable symptom severity and voice disability. Those with the disorder experience a wide spectrum of symptom severity over time, resulting in varied degrees of perceived voice disability. This study investigated the longitudinal variability of AdSD, with a focus on auditory–perceptual judgments of a dimension termed laryngeal overpressure (LO) and patient self-assessments of voice-related quality of life (V-RQOL). Study Design: Longitudinal, correlational study. Methods: Ten adults with AdSD were followed over three time periods. At each, both voice samples and self-ratings of V-RQOL were gathered prior to their scheduled Botox injection. Voice recordings subsequently were perceptually evaluated by eight listeners for LO using a visual analog scale. Results: LO ratings for all-voiced and Rainbow Passage sentence stimuli were found to be highly correlated. However, only the LO ratings obtained from judgments of AV stimuli were found to correlate moderately with self-ratings of voice disability for both the physical functioning and social-emotional subscores, as well as the total V-RQOL score. Based on perceptual judgments, LO appears to provide a reliable means of quantifying the severity of voice abnormalities in AdSD. Conclusions: Variability in self-ratings of the V-RQOL suggest that perceived disability related to AdSD should be actively monitored. Further, auditory–perceptual judgments may provide an accurate index of the potential impact of the disorder on the speaker. Similarly, LO was supported as a simple clinical measure that serves as a reliable index of voice change over time. Key Words: laryngology, voice/dysphonia, spasmodic dysphonia, laryngeal overpressure, voice-related quality of life. Level of Evidence: 4. Laryngoscope, 125:661–666, 2015
INTRODUCTION Adductor spasmodic dysphonia (AdSD) is a rare focal laryngeal dystonia that is characterized by abnormal, intermittent hyperadduction of the vocal folds during phonation.1 This produces a voice that is typically perceived by the listener as exhibiting a strained, strangled quality, and at times coexists with phonatory breaks.2 AdSD can be a debilitating condition that may significantly impact communication for the individual in all areas of functioning.3,4 The current mainstay of treat-
From the Department of Otolaryngology–Head and Neck Surgery (J.C.Y.), University of Ottawa, Ottawa; the Department of Medicine (S.K.R.), McMaster University, Hamilton; the Department of Otolaryngology–Head and Neck Surgery, Schulich School of Medicine and Dentistry (K.F.), School of Communication Sciences and Disorders (E.D., A.M.B.D., A.D., C.B., P.C.D.), and the Health and Rehabilitation Sciences (P.C.D.), Western University, London, Ontario, Canada. Editor’s Note: This Manuscript was accepted for publication September 8, 2014. Editor’s Note: This Manuscript was accepted for publication on September 8, 2014. The authors have no funding, financial relationships, or conflicts of interest to disclose. Send correspondence to Philip C. Doyle, PhD, Voice Production and Perception Laboratory and The Laboratory for Well-Being and Quality of Life Oncology, Elborn College, Room 2200, Western University, London, Ontario, Canada, N6G 1H1. E-mail: [email protected] DOI: 10.1002/lary.24953
Laryngoscope 125: March 2015
ment for AdSD is symptom management using Botulinum Toxin A (Botox) injections in the thyroarytenoid/ lateral cricoarytenoid muscle complex.1,5,6 Because treatment is not curative and variability exists in the symptomology associated with AdSD,7 a valid metric for longitudinal evaluation of this disorder is warranted. If this data is also reflective of self-assessments of the voice disability experienced, its value becomes much greater, and the ability to better define outcomes may also be enhanced.8 Efforts that facilitate a more complete description of characteristics of AdSD in each patient may serve two purposes: 1) It would serve to identify patterns of change that can be addressed more systematically specific to Botox treatment intervals. 2) Patient-specific information on the clinical variability of their disease would improve patient education about the disorder and its course over time. The development of a simple metric for clinical documentation has been challenging. While a variety of clinical voice-evaluation methods potentially exist,6,9,10 no current approach to clinically defining AdSD is without limitation. Auditory–perceptual measures (i.e., listener judgments) are potentially advantageous due to their general ease of use, time efficiency, and potential to capture elements of one’s vocal quality that cannot be easily quantified by acoustic analyses.6 However, such measures are also reliant on the correct use of Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
661
auditory–perceptual scaling and associated methodology, as well as the expertise of the listener and the inherent potential for inconsistency in both intra- and interrater reliability.11,12 Despite these limitations, the advantages of auditory–perceptual measures have resulted in their increased use in the clinical setting.13 Similarly, selfperception of the patient’s voice disorder provides valuable information regarding the disability that is experienced.3,4,14 If auditory–perceptual evaluations and selfreported measures of AdSD can be combined in a reliable fashion, this would enable comparison between a patient’s voice characteristics and the speaker’s selfperceived voice disability. In the case of AdSD, the variability of the disorder may at times be so substantial that seeking to index the disorder with a simple measure of severity may be less than accurate.15 Thus, the ability to identify a descriptive dimension that holds the greatest potential to comprehensively index the characteristics of AdSD would be of benefit to clinicians who wish to conduct long-term treatment monitoring.14,16,17 One such listener-based descriptor termed laryngeal overpressure (LO) has been previously reported to accurately and consistently document clinical variability in AdSD.18,19 As a descriptive auditory–perceptual dimension, LO refers to the amount of muscular force believed to be required to produce the voice. At a basic conceptual level, we believe that LO as an auditory–perceptual dimension is harmonious with the characteristics of AdSD.18 More explicitly, LO as an auditory–perceptual measure has the direct potential to link listener perception with the variable characteristics of laryngeal adductory control and force that are commonly associated with AdSD. Because adductory force occurs on a physiologic continuum ranging from underpressure (breathy) to overpressure (hyperclosure), it is perceptually salient to a listener. Thus, we believe that the face validity of LO as a potential descriptive index or outcome measure for AdSD over time is clear. This investigation sought to correlate auditory–perceptual ratings of LO by listener over time, as well as identify the potential relationship between LO ratings and patients’ self-ratings of vocal functioning. We hypothesized that listener ratings of LO would correlate with patients self-perceptions of voice disability, as quantified by a validated voice-related quality of life measure.
MATERIALS AND METHODS
provide a standard voice recording that would later be submitted to auditory–perceptual analysis. Both the voice recording and the self-assessment of voice were gathered prior to a Botox injection.
Patient Self-Assessments Patients’ self-ratings of vocal functioning were documented using the 10-item V-RQOL instrument.20 Briefly, each V-RQOL question was rated on a 5-point Likert style scale. Response options for all questions ranged from a 1 representing “Not a problem at all” to a 5, which represented a problem that was judged “As bad as it can be.” Two separate subscores from the V-RQOL were then generated (social/emotional and physical functioning domains), along with a total V-RQOL. This resulted in three voice-related “quality of life” scores ranging from 0 to 100. Based on the scoring algorithm and related transformation,20 lower scores represented better functioning, whereas higher scores represented increasing levels of disability specific to voice use.20
Voice Recordings Each participant was asked to record two sentences: one that contained all voiced (AV) sounds11,19 (i.e., “Early one morning a man and a woman were ambling along a one-mile lane running near Rainy Island Avenue”) and a second sentence from the Rainbow Passage21 (RP) that contained both voiced and voiceless sounds (i.e., “The rainbow is a division of white light into many beautiful colors.”). The AV stimulus is constructed with the assumption that the vocal folds will be adducted throughout production of the entire sentence. In contrast, the RP sentence includes both adductory and abductory events. All recordings were gathered in a counterbalanced fashion and adjusted for input volume using a commercially available digital-recording system (Sona-Speech, KayPENTAX, Montvale, NJ) at a sampling rate of 44KHz installed on a laptop computer (Toshiba Satellite P100, Tokyo, Japan). The methods used to record these passages have been previously described.19 All recordings were obtained by the same researcher in a setting free of background noise.
Perceptual Assessments of LO All identifying information was initially removed from all the voice recordings, and randomized playlists containing all recordings were created. A separate playlist was created for each of the two stimulus sentences (AV and RP), and two versions of each playlist (using a different randomization code) were also created, which resulted in four final lists (RP-A, RPB, AV-A, and AV-B). Five additional samples chosen at random (16.6%) from each list of 30 stimuli were also duplicated and included on each list, to be played a second time at the end of each playlist in order to assess for test–retest reliability of each listener.
Study Design A longitudinal, correlational study design was employed. Patients who were included in the study were previously diagnosed with AdSD by a board-certified otolaryngologist (K.F.) and spoke English as a first language. Exclusion criteria included the presence of other concurrent medical conditions, other frank neurologic deficits (including a diagnosis of multifocal dystonia), perceptual evidence of vocal tremor, or a current or previous history of hearing loss. All patients had previously undergone at least one prior treatment with Botox. At three separate visits, the participant was asked to complete two tasks: 1) a selfassessment questionnaire that addressed their voice problem (the Voice-Related Quality of Life20 [V-RQOL] measure), and 2)
Laryngoscope 125: March 2015
662
Listeners Eight normal-hearing graduate students (6 women, 2 men) served as listeners for the auditory–perceptual phase of the investigation. Listeners ranged in age from age range 23 to 27 years, and all were currently enrolled in the SpeechLanguage Pathology Program. Although none were considered to be expert listeners, all had prior formal exposure to, or experience with, the vocal characteristics of those with AdSD, as well as to auditory–perceptual methods. Listeners were also knowledgeable relative to laryngeal physiology and glottal dynamics associated with voice production. Following informed
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
TABLE I. Mean Values and SDs for Listener Ratings of Laryngeal Overpressure Across All Speakers Over Three Sessions. Mean
SD
Range
All-voiced
65.45
18.67
33.87–97.0
Rainbow Passage
63.72
17.65
31.75–93.12
SD 5 standard deviation.
consent, instructions for the listening task were presented to each listener individually. All were presented with one randomly selected RP playlist and one AV playlist. Listeners were then asked to provide a rating of LO for each sentence using a visual analogue scale. Ratings were documented using a separate 100-mm visual analog scale for each sample; no numerical reference points were provided. Each scale was constructed so that the midpoint of the scale (i.e., 50 mm from the ends of the scale) represented “normal” laryngeal pressure, whereas the far left of the scale represented minimal laryngeal (under) pressure and the far right of the scale represented maximal laryngeal (over) pressure. Listeners were explicitly asked to listen to each voice sample presented. They were instructed to mark the scale at any point along its continuum that they believed best represented the overall degree of laryngeal closure required by the speaker during the production of each stimuli. Thus, scaled values could ultimately be converted to a numeric score that ranged from 1 to 100. Once all samples had been rated, the distance (in mm) from the left of the scale was manually measured by an independent and blinded assistant, and these data were then compiled for each speaker across the group of 10 listeners.
collected across a potential range of symptoms/severity. Thirty V-RQOL surveys (3 measures 3 10 speakers) and 60 voice recordings (10 speakers 3 2 sentences 3 3 time periods) were obtained from the participant group as part of this study. However, at the end of the entire data collection period, one V-RQOL from a single speaker was found to only have the first two questions completed. Hence, that single V-RQOL was discarded prior to data analysis, which resulted in 29 completed measures. Additionally, one other speaker was noted to have consistently not answered one question, namely, “I have trouble doing my job or practicing my profession.” Because this nonresponse was due to the speaker’s employment status (the speaker had been retired for a number of years), and because consultation with the participant at termination of the study indicated that the voice problem would have likely not posed a problem specific to this question, we sought to mitigate that omission by scoring it as a “1” (indicating that a problem was not experienced). Our rationale was two-fold: First, it did not unduly penalize the participant; and second, omission of that score would have required a modification in the standard scoring algorithm due to the missing response. Finally, because variability over time was an important measure in the context of this investigation, all individual speaker data were retained. A summary of both the listeners’ auditory–perceptual ratings of LO and the V-RQOL data over the three time periods of data acquisition are presented in Tables I and II, respectively.
Statistical Analysis Initial analyses of the ratings of LO by the listeners between the AV and RP stimuli were performed using t tests for independent samples. Next, the analysis of potential differences between the speakers’ self-evaluation of voice using the VRQOL were performed; again, potential differences were evaluated using a t test for independent samples. Following that, evaluation of the degree of relationship between the three VRQOL scores (physical functioning [PF], social-emotional [SE], and total) and auditory–perceptual assessments of LO were made using a Pearson’s product moment correlation. Intra- and interreliability of the listener ratings were assessed using intraclass correlation (ICC). Therefore, the primary outcome measures evolved from the correlation coefficients between the VRQOL subscores (PF, SE, and total scores) and the composite listener ratings of LO for each of the recorded voice samples obtained. Secondary statistical measures included interrater reliability of the listener ratings as well as test–retest reliability. All statistical analyses were conducted using commercially available software (PASW Statistics 18, Minneapolis, MN).
Auditory–Perceptual Ratings of LO
RESULTS
V-RQOL Self-Ratings
Ten randomly selected patients (3 men and 7 women) diagnosed with AdSD participated in this study. Women ranged in age from 39 to 73 years (M 5 63;8) and men ranged in age from 55 to 57 years (M 5 56;0). Those who consented to serve as participant speakers were followed over the course of their next three regular treatment or follow-up appointments. The period ranged from 16 to 22 months. This extended time period was seen to be of advantage in that it allowed data to be
Mean data and SDs for V-RQOL subscores and the total score are presented in Table III. Specific to the VRQOL data, both SE and PF subscores were found to be significantly correlated to one another (r 5.832; P < 0.01). Both SE and PF subscores were also found to be strongly and significantly correlated to the V-RQOL total score (r 5.913 and r 5.953; P < 0.01, respectively). Thus, self-perceived disability as a consequence of AdSD did vary across the two areas of assessment, namely,
Laryngoscope 125: March 2015
Mean data and standard deviations (SD) for the listener ratings of LO are presented in Table I. Results of a t test reveal no significant differences between ratings of the sentences. Additionally, listener ratings for the two sentences, AV and RP, demonstrated a strong, statistically significant correlation (r 50.774; P < 0.01). These ratings were also shown to be highly reliable with the ICCs for the listener ratings ranging from 0.773 to 0.872. These data suggest that listener judgments were reliable across the two passages. Furthermore, the AV and RP sentence stimuli provide a valid means of using the LO scaled measure to index the voice characteristics of AdSD. We have also provided a summary of the individual data for each speaker over time in Table II. These data clearly reveal that variability is substantial across this group of 10 participant speakers.
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
663
TABLE II. Auditory–Perceptual Scaled Ratings of Laryngeal Overpressure for Individual Speakers Over Three Data Collection Sessions. Speaker (session no.)
1 (1)
Average Listener Rating: AV Sentence
Minimum–Maximum Listener Ratings: AV Sentence
Average Listener Rating: RP Sentence
Minimum–Maximum Listener Ratings: RP Sentence
91.12
83–96
89.87
74–100
84.5
73–92
82.75
72–94
(3) 2 (1)
75.37 77.37
54–86 26–97
93.12 84.25
85–100 66–99
(2)
45.12
7–62
78.87
60–98
(3) 3 (1)
97.0 60.25
89–100 45–74
61.87 53.75
14–85 49–80
(2)
54.12
35–71
49.75
35–66
(3) 4 (1)
42.0 38.62
1–69 1–56
51.87 40.5
31–66 1–52
(2)
50.75
3–66
40.62
2–51
(3) 5 (1)
39.25 78.0
2–57 50–88
47.5 70.12
25–58 38–89
(2)
77.12
62–98
60.75
37–82
(3) 6 (1)
44.75 74.87
24–52 16–100
44.87 73.87
8–54 58–98
81.0
65–99
76.25
64–89
70.25 73.75
57–85 37–94
84.0 67.75
69–96 37–95
(2)
(2) (3) 7 (1) (2)
79.0
72–87
64.25
30–85
(3) 8 (1)
82.75 70.75
68–100 49–83
74.87 72.0
30–95 57–85
(2)
69.25
62–80
58.75
25–76
(3) 9 (1)
85.37 77.62
71–96 67–100
83.25 72.5
60–100 54–95
(2)
82.75
61–100
80.12
60–94
(3) 10 (1)
43.5 41.0
15–61 11–80
45.87 37.37
9–58 7–82
(2)
33.87
1–75
38.62
8–91
(3)
42.5
7–82
31.75
1–77
Data depicted represent the average scaled scores across the listeners (n 5 8), as well as the minimum and maximum ratings in any sessions for both the all voiced and RP sentences. AV 5 all-voiced; RP 5 Rainbow Passage.
those areas explored through the SE and PF domain. Table IV presents summary data on the V-RQOL scores (SE, PF, and total) for individual participant speakers over the three points of data collection. Finally, in order to determine if any potential relationship existed between the speakers self-ratings of voice quality and the listener ratings of the speech stimuli obtained, we performed correlational analysis between LO ratings for both sentences (AV and RP) and the two V-RQOL subscores (SE and PF) and the total score. Based on this analysis, both the SE and PF subscores and the total score were found to be moderately and significantly but negatively correlated with the AV sentence (r 5 2.403, P < 0.05; r 5 2.459, P < 0.05; and r 5 2.457, P < 0.05), respectively. This negative correlation was expected. As V-RQOL scores increase, one would expect the perceived overpressure and associated impact on voice control to decrease. In contrast, however, Laryngoscope 125: March 2015
664
no significant correlations were noted between any VRQOL score and the RP sentence.
DISCUSSION This study sought to build on previous work that focused on the use of the auditory–perceptual measure of LO as a simple and reliable means of documenting the voice quality of individuals diagnosed with AdSD. If an association between this listener-based metric and participants’ self-assessments of V-RQOL was to emerge, it would greatly strengthen the utility and validity of LO as a clinical method of voice description for those with AdSD. The present data provide evidence that the evaluation of LO in this speaker group did correlate with self-generated assessments of their own voice disability (as determined by the V-RQOL scores) in the context of the AV sentence. Interestingly, relative to the Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
TABLE III. Mean Values and SDs for the Physical Functioning, SocialEmotional, and Total V-RQOL Scores Across All Speakers Over Three Sessions. Social-Emotional
Physical Functioning
Total VRQOL
Mean
66.37
53.59
57.67
SD
25.3
27.06
26.35
SD 5 standard deviation; V-RQOL 5 voice-related quality of life.
auditory–perceptual scores between the AV and RP sentence stimuli, no significant differences were revealed. This suggests that the perceptual salience of LO as a clinical descriptor may be of more generalized value (i.e., it may not be stimuli-specific). Therefore, the ability to accurately detect changes in LO as per the scaled measure of interest is potentially enhanced. However, it is important to note that when sentence stimuli are considered in the context of self-ratings, only the AV stimuli exhibited a stronger relationship to the V-RQOL scores. Our group previously demonstrated strong intraand interrater reliability associated with listeners’ auditory–perceptual judgments of LO, and similar findings were demonstrated herein.19 Our previous work also identified correlation between judgments of LO and objective acoustic measurements. The present study adds to this foundation by confirming the reliability of this measure and determining its association to participants’ self-assessments of V-RQOL. The strained voice quality and effort required for voice production is readily apparent to the listener and disruptive to efficient communication in AdSD patients. Naturally, these types of deficits can be of significant detriment to patients’ communicative, and more directly, V-RQOL.3,14,16 Further, the inability to maintain a level of control one’s voice may contribute substantially to the voice disability experienced. Kaptein et al.4 reported that AdSD patients have low perceived control over their symptoms, which can have significant psychological consequences.3,17 As a result, gathering both LO and selfratings may serve as important indices of clinical change in those with AdSD. In the present study, we identified a moderate, negative correlation between listener ratings of LO for the AV sentence stimuli and all scores generated from the VRQOL. Previous studies have suggested a weak correlation between a patient’s V-RQOL score and clinicians’ perceptual judgment of voice impairment.22 In contrast, however, Hummel et al.23 compared objective measures of voice impairment using the Dysphonia Severity Index (DSI) to the patient-reported score on V-RQOL, but no correlation between these scores was found. Thus, the correlation between listener ratings and the V-RQOL found in the present study may be a function of the stimuli used as part of the auditory–perceptual phase of the study.11,12,24 This finding may provide evidence that ratings of LO, a feature that we believe has considerable face validity in the context of AdSD, is well-matched to a speaker’s selfperception of the voice disability that they experience with Laryngoscope 125: March 2015
this disorder. However, our data also suggest that the speaker’s self-assessment of the disability they experience does not uniformly correspond to a listener’s auditory–perceptual assessment of the voice. That is, a significant relationship to self-ratings was only found for the AV stimuli; thus, the ability to link auditory–perceptual measures with the V-RQOL may offer a unique contribution to clinical description of AdSD. Ultimately, we believe that these two measures assess two distinct clinical concerns: 1) the degree of voice abnormality (i.e., the extent of overpressure due to the disorder), and 2) the disability that truly results at both “physical” and social-emotional levels of functioning. Consequently, comprehensive clinical services that broadly address the disability experienced by those with AdSD appear to be justified. Although the V-RQOL may be an excellent predictor of voice status in some individuals,14 an identical change
TABLE IV. V-RQOL Scores (Social-Emotional, Physical, and Total) for Individual Speakers Over Three Data Collection Sessions. Speaker (Session)
1 (1) (2) (3) 2 (1) (2)
Social-Emotional Domain
Physical Domain
Total V-RQOL Score
52.5
50
54.17
31.25
29.17
30
62.5 62.5
54.17 66.67
57.5 65
100
100
100
(3) 3 (1)
81.25 81.25
29.17 79.17
50 80
(2)
93.75
91.67
92.5
87.5 87.5
83.33 79.17
85 82.5
(3) 4 (1) (2)
100
91.67
95
(3) 5 (1)
93.75 87.5
83.33 58.33
87.5 70
(2)
68.75
45.83
55
(3) 6 (1)
93.75 31.25
70.83 37.5
80 35
(2)
43.75
25
32.5
(3) 7 (1)
37.5 56.25
41.67 41.67
40 17.5
(2)
31.25
16.67
22.5
(3) 8 (1)
6.25 87.5
0 87.5
2.5 87.5
(2)
93.75
87.5
90
(3) 9 (1)
* 81.25
* 33.33
* 52.5
(2)
56.25
25
37.5
(3) 10 (1)
68.75 37.5
29.17 29.17
45 32.5
(2)
56.25
37.5
45
(3)
56.25
45.83
50
Data depicted represent the scores generated as per V-RQOL scoring algorithm. *Data discarded due to incomplete questionnaire. V-RQOL 5 voice-related quality of life.
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
665
in symptom severity may not objectively represent a commensurate change in perceived V-RQOL. A similar lack of correlation exists for other auditory–perceptual scales and validated quality of life measures for AdSD, which also suggests that patients’ internal standards likely differ from those of experienced listeners.12,22 Finally, although the present study was able to generate important longitudinal data for both LO and selfassessments of voice disability secondary to AdSD, we were unable to make any assessment of participant performance with respect to one another because time point A for participant 1 may not have been equivalent to time point A for participant 2, and participant 3, etc. Nevertheless, the present data do support the idea that regular assessments of those with AdSD at both levels addressed in the current study may be of benefit in measuring the variability of voice symptoms in patients. Data of this type may then serve to identify patterns of vocal change that may influence decisions on the ideal time for Botox treatment. A potential limitation of the present study may be found in the listeners who provided auditory–perceptual ratings of speech samples or the stimuli employed. We purposely selected this knowledgeable yet “nonexpert” group of listeners because the listeners tend to produce highly correlated ratings of LO, more so than their inexperienced counterparts, as suggested by our previous work.19 Thus, those who have no experience (na€ıve) or who have considerable experience with this population (experts) may provide very different ratings of LO. Thus, the effect of listener experience would likely bring with it a greater level of reliability, particularly in the context of the definition of LO that we have developed and employed in this phase of our project. As a potential counter to such limitations, we believe that the individuals’ data acquired over time for both LO ratings and the V-RQOL provide a rich and important body of information specific to the variability of AdSD over time. In this respect, group means for either measure ultimately do not adequately represent individual voice deficits or a speaker’s self-perception of their own voice-related disability in those with AdSD. As such, we believe that the present data provide support for clinical applications of the LO scaling method as a means of indexing adductory deficits associated with AdSD.
CONCLUSION This study identified an association between the listener ratings of LO for a stimulus sentence comprised of all-voiced sound and both subscores and the total V-RQOL score. Thus, variations in a participant’s self-perceived VRQOL did correlate with listener assessments of LO. The existence of a strong association between participantbased and listener-based perceptual assessments, along with the previously documented association with objective acoustic measures, provides a basis for the use of the LO scale as a simple, easy-to-use perceptual metric, both in clinical and research settings. The findings of this study also support the importance of ongoing participant selfassessments of perceived voice disability as a consequence of AdSD. Lastly, the combined use of auditory–perceptual
Laryngoscope 125: March 2015
666
judgments and self-ratings of voice would appear to provide a more complete perspective of voice disability that may not be achieved through use of single measures.
BIBLIOGRAPHY 1. Cannito MP, Doiuchi M, Murry T, Woodson GE. Perceptual structure of adductor spasmodic dysphonia and its acoustic correlates. J Voice 2012; 26:818.e5–818.e 13. doi:10.1016/j.jvoice.2012.05.005. 2. Blitzer A, Brin MF. Treatment of spasmodic dysphonia (laryngeal dystonia) with local injections of Botulinum toxin. J Voice 1992;6:365–369. doi:10.1016/S0892-1997(05)80035-9. 3. Baylor CR, Yorkston KM, Eadie TL. The consequences of spasmodic dysphonia on communication-related quality of life: a qualitative study of the insider’s experiences. J Commun Disord 2005;38:395–419. doi: 10.1016/j.jcomdis.2005.03.003. 4. Kaptein AA, Hughes BM, Scharloo M, Hondebrink N, Langeveld TPM. Psychological aspects of adductor spasmodic dysphonia: a prospective population controlled questionnaire study. Clin Otolaryngol 2010;35:31– 38. doi:10.1111/j.1749-4486.2009.02070.x. 5. Blitzer A, Brin MF, Stewart CF. Botulinum toxin management of spasmodic dysphonia (laryngeal dystonia): a 12-year experience in more than 900 patients. Laryngoscope 1998;108:1435–1441. doi:10.1097/ 00005537-199810000-00003. 6. Zwirner P, Murry T, Swenson M, Woodson GE. Effects of botulinum toxin therapy in patients with adductor spasmodic dysphonia: acoustic, aerodynamic, and videoendoscopic findings. Laryngoscope 1992;102:400–406. doi:10.1288/00005537-199204000-00006. 7. Stewart CF, Allen EL, Tureen P, Diamond BE, Blitzer A, Brin MF. Adductor spasmodic dysphonia: standard evaluation of symptoms and severity. J Voice 1997;11:95–103. doi:10.1016/S0892-1997(97)80029-X. 8. Fisher K V, Scherer RC, Guo CG, Owen AS. Longitudinal phonatory characteristics after botulinum toxin type A injection. J Speech, Lang Hear Res 1996;39:968–980. doi:10.1044/jshr.3905.968. 9. Sapienza CM, Crary MA, Gorham M. Laryngeal aerodynamic aspects of women with adductor spasmodic dysphonia. Arch Otolaryngol Head Neck Surg 1996;122:385–388. doi:10.1001/archotol.1996.01890160027005. 10. Shipp T, Izdebski K, Schutte HK, Morrissey P. Subglottal air pressure in spastic dysphonia speech. Folia Phoniatr Logop 1988;40:105–110. doi: 10.1159/000265895. 11. Roy N, Gouse M, Mauszycki SC, Merrill RM, Smith ME. Task specificity in adductor spasmodic dysphonia versus muscle tension dysphonia. Laryngoscope 2005;115:311–316. doi:10.1097/01.mlg.0000154739.48314.ee. 12. Eadie TL, Kapsner M, Rosenzweig J, Waugh P, Hillel A, Merati A. The role of experience on judgments of dysphonia. J Voice 2010;24:564–573. doi:10.1016/j.jvoice.2008.12.005. 13. Kreiman J, Gerratt BR. Validity of rating scale measures of voice quality. J Acoust Soc Am 1998;104:1598–1608. doi:10.1121/1.424372. 14. Rubin AD, Wodchis WP, Spak C, Kileny PR, Hogikyan ND. Longitudinal effects of Botox injections on voice-related quality of life (V-RQOL) for patients with adductory spasmodic dysphonia: part II. Arch Otolaryngol Head Neck Surg 2004;130:415–420. doi:10.1001/archotol.130.4.415. 15. Eadie TL, Doyle PC. Direct magnitude estimation and interval scaling of pleasantness and severity in dysphonic and normal speakers. J Acoust Soc Am 2002;112:3014–3021. doi:10.1121/1.1518983. 16. Hogikyan ND, Wodchis WP, Spak C, Kileny PR. Longitudinal effects of botulinum toxin injections on voice-related quality of life (V-RQOL) for patients with adductory spasmodic dysphonia. J Voice 2001;15:576–586. doi:10.1016/S0892-1997(01)00060-1. 17. Paniello RC, Barlow J, Serna JS. Longitudinal follow-up of adductor spasmodic dysphonia patients after botulinum toxin injection: quality of life results. Laryngoscope 2008;118:564–568. doi:10.1097/MLG. 0b013e31815e8be0. 18. Izdebski K. Overpressure and breathiness in spastic dysphonia: an acoustic (LTAS) and perceptual study. Acta Otolaryngol 1984;97:373–378. Available at: http://informahealthcare.com/doi/abs/10.3109/00016488409131003. Accessed February 20, 2014. 19. Yeung J, Fung K, Bornbaum CC, et al. Clinical approach to monitoring variability associated with adductor spasmodic dysphonia. J Otolaryngol Head Neck Surg 2011;40:343–349. 20. Hogikyan ND, Sethuraman G. Validation of an instrument to measure voice-related quality of life (V-RQOL). J Voice 1999;13:557–569. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10622521. 21. Fairbanks G. Voice and Articulation Drillbook. 2nd ed. NewYork, NY: Harper and Row; 1960. 22. Braden MN, Johns MM, Klein AM, Delgaudio JM, Gilman M, Hapner ER. Assessing the effectiveness of botulinum toxin injections for adductor spasmodic dysphonia: clinician and patient perception. J Voice 2010;24: 242–249. doi:10.1016/j.jvoice.2008.08.003. 23. Hummel C, Scharf M, Schuetzenberger A, Graessel E, Rosanowski F. Objective voice parameters and self-perceived handicap in dysphonia. Folia Phoniatr Logop 2010;62:303–307. doi:10.1159/000287715. 24. Zwirner P, Murry T, Woodson GE. Perceptual-acoustic relationships in spasmodic dysphonia. J Voice 1993;7:165–171. doi:10.1016/S08921997(05)80347-9.
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
Rhinological and Otological Society, Inc.
Longitudinal Variations of Laryngeal Overpressure and Voice-Related Quality of Life in Spasmodic Dysphonia Jeffrey C. Yeung, MD; Kevin Fung, MD, FRCSC; Eric Davis, BHSc; Sunita K. Rai, BHSc; Adam M. B. Day, PhD; Agnieszka Dzioba, MSc; Catherine Bornbaum, PhD; Philip C. Doyle, PhD Objectives/Hypothesis: Adductor spasmodic dysphonia (AdSD) is a voice disorder characterized by variable symptom severity and voice disability. Those with the disorder experience a wide spectrum of symptom severity over time, resulting in varied degrees of perceived voice disability. This study investigated the longitudinal variability of AdSD, with a focus on auditory–perceptual judgments of a dimension termed laryngeal overpressure (LO) and patient self-assessments of voice-related quality of life (V-RQOL). Study Design: Longitudinal, correlational study. Methods: Ten adults with AdSD were followed over three time periods. At each, both voice samples and self-ratings of V-RQOL were gathered prior to their scheduled Botox injection. Voice recordings subsequently were perceptually evaluated by eight listeners for LO using a visual analog scale. Results: LO ratings for all-voiced and Rainbow Passage sentence stimuli were found to be highly correlated. However, only the LO ratings obtained from judgments of AV stimuli were found to correlate moderately with self-ratings of voice disability for both the physical functioning and social-emotional subscores, as well as the total V-RQOL score. Based on perceptual judgments, LO appears to provide a reliable means of quantifying the severity of voice abnormalities in AdSD. Conclusions: Variability in self-ratings of the V-RQOL suggest that perceived disability related to AdSD should be actively monitored. Further, auditory–perceptual judgments may provide an accurate index of the potential impact of the disorder on the speaker. Similarly, LO was supported as a simple clinical measure that serves as a reliable index of voice change over time. Key Words: laryngology, voice/dysphonia, spasmodic dysphonia, laryngeal overpressure, voice-related quality of life. Level of Evidence: 4. Laryngoscope, 125:661–666, 2015
INTRODUCTION Adductor spasmodic dysphonia (AdSD) is a rare focal laryngeal dystonia that is characterized by abnormal, intermittent hyperadduction of the vocal folds during phonation.1 This produces a voice that is typically perceived by the listener as exhibiting a strained, strangled quality, and at times coexists with phonatory breaks.2 AdSD can be a debilitating condition that may significantly impact communication for the individual in all areas of functioning.3,4 The current mainstay of treat-
From the Department of Otolaryngology–Head and Neck Surgery (J.C.Y.), University of Ottawa, Ottawa; the Department of Medicine (S.K.R.), McMaster University, Hamilton; the Department of Otolaryngology–Head and Neck Surgery, Schulich School of Medicine and Dentistry (K.F.), School of Communication Sciences and Disorders (E.D., A.M.B.D., A.D., C.B., P.C.D.), and the Health and Rehabilitation Sciences (P.C.D.), Western University, London, Ontario, Canada. Editor’s Note: This Manuscript was accepted for publication September 8, 2014. Editor’s Note: This Manuscript was accepted for publication on September 8, 2014. The authors have no funding, financial relationships, or conflicts of interest to disclose. Send correspondence to Philip C. Doyle, PhD, Voice Production and Perception Laboratory and The Laboratory for Well-Being and Quality of Life Oncology, Elborn College, Room 2200, Western University, London, Ontario, Canada, N6G 1H1. E-mail: [email protected] DOI: 10.1002/lary.24953
Laryngoscope 125: March 2015
ment for AdSD is symptom management using Botulinum Toxin A (Botox) injections in the thyroarytenoid/ lateral cricoarytenoid muscle complex.1,5,6 Because treatment is not curative and variability exists in the symptomology associated with AdSD,7 a valid metric for longitudinal evaluation of this disorder is warranted. If this data is also reflective of self-assessments of the voice disability experienced, its value becomes much greater, and the ability to better define outcomes may also be enhanced.8 Efforts that facilitate a more complete description of characteristics of AdSD in each patient may serve two purposes: 1) It would serve to identify patterns of change that can be addressed more systematically specific to Botox treatment intervals. 2) Patient-specific information on the clinical variability of their disease would improve patient education about the disorder and its course over time. The development of a simple metric for clinical documentation has been challenging. While a variety of clinical voice-evaluation methods potentially exist,6,9,10 no current approach to clinically defining AdSD is without limitation. Auditory–perceptual measures (i.e., listener judgments) are potentially advantageous due to their general ease of use, time efficiency, and potential to capture elements of one’s vocal quality that cannot be easily quantified by acoustic analyses.6 However, such measures are also reliant on the correct use of Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
661
auditory–perceptual scaling and associated methodology, as well as the expertise of the listener and the inherent potential for inconsistency in both intra- and interrater reliability.11,12 Despite these limitations, the advantages of auditory–perceptual measures have resulted in their increased use in the clinical setting.13 Similarly, selfperception of the patient’s voice disorder provides valuable information regarding the disability that is experienced.3,4,14 If auditory–perceptual evaluations and selfreported measures of AdSD can be combined in a reliable fashion, this would enable comparison between a patient’s voice characteristics and the speaker’s selfperceived voice disability. In the case of AdSD, the variability of the disorder may at times be so substantial that seeking to index the disorder with a simple measure of severity may be less than accurate.15 Thus, the ability to identify a descriptive dimension that holds the greatest potential to comprehensively index the characteristics of AdSD would be of benefit to clinicians who wish to conduct long-term treatment monitoring.14,16,17 One such listener-based descriptor termed laryngeal overpressure (LO) has been previously reported to accurately and consistently document clinical variability in AdSD.18,19 As a descriptive auditory–perceptual dimension, LO refers to the amount of muscular force believed to be required to produce the voice. At a basic conceptual level, we believe that LO as an auditory–perceptual dimension is harmonious with the characteristics of AdSD.18 More explicitly, LO as an auditory–perceptual measure has the direct potential to link listener perception with the variable characteristics of laryngeal adductory control and force that are commonly associated with AdSD. Because adductory force occurs on a physiologic continuum ranging from underpressure (breathy) to overpressure (hyperclosure), it is perceptually salient to a listener. Thus, we believe that the face validity of LO as a potential descriptive index or outcome measure for AdSD over time is clear. This investigation sought to correlate auditory–perceptual ratings of LO by listener over time, as well as identify the potential relationship between LO ratings and patients’ self-ratings of vocal functioning. We hypothesized that listener ratings of LO would correlate with patients self-perceptions of voice disability, as quantified by a validated voice-related quality of life measure.
MATERIALS AND METHODS
provide a standard voice recording that would later be submitted to auditory–perceptual analysis. Both the voice recording and the self-assessment of voice were gathered prior to a Botox injection.
Patient Self-Assessments Patients’ self-ratings of vocal functioning were documented using the 10-item V-RQOL instrument.20 Briefly, each V-RQOL question was rated on a 5-point Likert style scale. Response options for all questions ranged from a 1 representing “Not a problem at all” to a 5, which represented a problem that was judged “As bad as it can be.” Two separate subscores from the V-RQOL were then generated (social/emotional and physical functioning domains), along with a total V-RQOL. This resulted in three voice-related “quality of life” scores ranging from 0 to 100. Based on the scoring algorithm and related transformation,20 lower scores represented better functioning, whereas higher scores represented increasing levels of disability specific to voice use.20
Voice Recordings Each participant was asked to record two sentences: one that contained all voiced (AV) sounds11,19 (i.e., “Early one morning a man and a woman were ambling along a one-mile lane running near Rainy Island Avenue”) and a second sentence from the Rainbow Passage21 (RP) that contained both voiced and voiceless sounds (i.e., “The rainbow is a division of white light into many beautiful colors.”). The AV stimulus is constructed with the assumption that the vocal folds will be adducted throughout production of the entire sentence. In contrast, the RP sentence includes both adductory and abductory events. All recordings were gathered in a counterbalanced fashion and adjusted for input volume using a commercially available digital-recording system (Sona-Speech, KayPENTAX, Montvale, NJ) at a sampling rate of 44KHz installed on a laptop computer (Toshiba Satellite P100, Tokyo, Japan). The methods used to record these passages have been previously described.19 All recordings were obtained by the same researcher in a setting free of background noise.
Perceptual Assessments of LO All identifying information was initially removed from all the voice recordings, and randomized playlists containing all recordings were created. A separate playlist was created for each of the two stimulus sentences (AV and RP), and two versions of each playlist (using a different randomization code) were also created, which resulted in four final lists (RP-A, RPB, AV-A, and AV-B). Five additional samples chosen at random (16.6%) from each list of 30 stimuli were also duplicated and included on each list, to be played a second time at the end of each playlist in order to assess for test–retest reliability of each listener.
Study Design A longitudinal, correlational study design was employed. Patients who were included in the study were previously diagnosed with AdSD by a board-certified otolaryngologist (K.F.) and spoke English as a first language. Exclusion criteria included the presence of other concurrent medical conditions, other frank neurologic deficits (including a diagnosis of multifocal dystonia), perceptual evidence of vocal tremor, or a current or previous history of hearing loss. All patients had previously undergone at least one prior treatment with Botox. At three separate visits, the participant was asked to complete two tasks: 1) a selfassessment questionnaire that addressed their voice problem (the Voice-Related Quality of Life20 [V-RQOL] measure), and 2)
Laryngoscope 125: March 2015
662
Listeners Eight normal-hearing graduate students (6 women, 2 men) served as listeners for the auditory–perceptual phase of the investigation. Listeners ranged in age from age range 23 to 27 years, and all were currently enrolled in the SpeechLanguage Pathology Program. Although none were considered to be expert listeners, all had prior formal exposure to, or experience with, the vocal characteristics of those with AdSD, as well as to auditory–perceptual methods. Listeners were also knowledgeable relative to laryngeal physiology and glottal dynamics associated with voice production. Following informed
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
TABLE I. Mean Values and SDs for Listener Ratings of Laryngeal Overpressure Across All Speakers Over Three Sessions. Mean
SD
Range
All-voiced
65.45
18.67
33.87–97.0
Rainbow Passage
63.72
17.65
31.75–93.12
SD 5 standard deviation.
consent, instructions for the listening task were presented to each listener individually. All were presented with one randomly selected RP playlist and one AV playlist. Listeners were then asked to provide a rating of LO for each sentence using a visual analogue scale. Ratings were documented using a separate 100-mm visual analog scale for each sample; no numerical reference points were provided. Each scale was constructed so that the midpoint of the scale (i.e., 50 mm from the ends of the scale) represented “normal” laryngeal pressure, whereas the far left of the scale represented minimal laryngeal (under) pressure and the far right of the scale represented maximal laryngeal (over) pressure. Listeners were explicitly asked to listen to each voice sample presented. They were instructed to mark the scale at any point along its continuum that they believed best represented the overall degree of laryngeal closure required by the speaker during the production of each stimuli. Thus, scaled values could ultimately be converted to a numeric score that ranged from 1 to 100. Once all samples had been rated, the distance (in mm) from the left of the scale was manually measured by an independent and blinded assistant, and these data were then compiled for each speaker across the group of 10 listeners.
collected across a potential range of symptoms/severity. Thirty V-RQOL surveys (3 measures 3 10 speakers) and 60 voice recordings (10 speakers 3 2 sentences 3 3 time periods) were obtained from the participant group as part of this study. However, at the end of the entire data collection period, one V-RQOL from a single speaker was found to only have the first two questions completed. Hence, that single V-RQOL was discarded prior to data analysis, which resulted in 29 completed measures. Additionally, one other speaker was noted to have consistently not answered one question, namely, “I have trouble doing my job or practicing my profession.” Because this nonresponse was due to the speaker’s employment status (the speaker had been retired for a number of years), and because consultation with the participant at termination of the study indicated that the voice problem would have likely not posed a problem specific to this question, we sought to mitigate that omission by scoring it as a “1” (indicating that a problem was not experienced). Our rationale was two-fold: First, it did not unduly penalize the participant; and second, omission of that score would have required a modification in the standard scoring algorithm due to the missing response. Finally, because variability over time was an important measure in the context of this investigation, all individual speaker data were retained. A summary of both the listeners’ auditory–perceptual ratings of LO and the V-RQOL data over the three time periods of data acquisition are presented in Tables I and II, respectively.
Statistical Analysis Initial analyses of the ratings of LO by the listeners between the AV and RP stimuli were performed using t tests for independent samples. Next, the analysis of potential differences between the speakers’ self-evaluation of voice using the VRQOL were performed; again, potential differences were evaluated using a t test for independent samples. Following that, evaluation of the degree of relationship between the three VRQOL scores (physical functioning [PF], social-emotional [SE], and total) and auditory–perceptual assessments of LO were made using a Pearson’s product moment correlation. Intra- and interreliability of the listener ratings were assessed using intraclass correlation (ICC). Therefore, the primary outcome measures evolved from the correlation coefficients between the VRQOL subscores (PF, SE, and total scores) and the composite listener ratings of LO for each of the recorded voice samples obtained. Secondary statistical measures included interrater reliability of the listener ratings as well as test–retest reliability. All statistical analyses were conducted using commercially available software (PASW Statistics 18, Minneapolis, MN).
Auditory–Perceptual Ratings of LO
RESULTS
V-RQOL Self-Ratings
Ten randomly selected patients (3 men and 7 women) diagnosed with AdSD participated in this study. Women ranged in age from 39 to 73 years (M 5 63;8) and men ranged in age from 55 to 57 years (M 5 56;0). Those who consented to serve as participant speakers were followed over the course of their next three regular treatment or follow-up appointments. The period ranged from 16 to 22 months. This extended time period was seen to be of advantage in that it allowed data to be
Mean data and SDs for V-RQOL subscores and the total score are presented in Table III. Specific to the VRQOL data, both SE and PF subscores were found to be significantly correlated to one another (r 5.832; P < 0.01). Both SE and PF subscores were also found to be strongly and significantly correlated to the V-RQOL total score (r 5.913 and r 5.953; P < 0.01, respectively). Thus, self-perceived disability as a consequence of AdSD did vary across the two areas of assessment, namely,
Laryngoscope 125: March 2015
Mean data and standard deviations (SD) for the listener ratings of LO are presented in Table I. Results of a t test reveal no significant differences between ratings of the sentences. Additionally, listener ratings for the two sentences, AV and RP, demonstrated a strong, statistically significant correlation (r 50.774; P < 0.01). These ratings were also shown to be highly reliable with the ICCs for the listener ratings ranging from 0.773 to 0.872. These data suggest that listener judgments were reliable across the two passages. Furthermore, the AV and RP sentence stimuli provide a valid means of using the LO scaled measure to index the voice characteristics of AdSD. We have also provided a summary of the individual data for each speaker over time in Table II. These data clearly reveal that variability is substantial across this group of 10 participant speakers.
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
663
TABLE II. Auditory–Perceptual Scaled Ratings of Laryngeal Overpressure for Individual Speakers Over Three Data Collection Sessions. Speaker (session no.)
1 (1)
Average Listener Rating: AV Sentence
Minimum–Maximum Listener Ratings: AV Sentence
Average Listener Rating: RP Sentence
Minimum–Maximum Listener Ratings: RP Sentence
91.12
83–96
89.87
74–100
84.5
73–92
82.75
72–94
(3) 2 (1)
75.37 77.37
54–86 26–97
93.12 84.25
85–100 66–99
(2)
45.12
7–62
78.87
60–98
(3) 3 (1)
97.0 60.25
89–100 45–74
61.87 53.75
14–85 49–80
(2)
54.12
35–71
49.75
35–66
(3) 4 (1)
42.0 38.62
1–69 1–56
51.87 40.5
31–66 1–52
(2)
50.75
3–66
40.62
2–51
(3) 5 (1)
39.25 78.0
2–57 50–88
47.5 70.12
25–58 38–89
(2)
77.12
62–98
60.75
37–82
(3) 6 (1)
44.75 74.87
24–52 16–100
44.87 73.87
8–54 58–98
81.0
65–99
76.25
64–89
70.25 73.75
57–85 37–94
84.0 67.75
69–96 37–95
(2)
(2) (3) 7 (1) (2)
79.0
72–87
64.25
30–85
(3) 8 (1)
82.75 70.75
68–100 49–83
74.87 72.0
30–95 57–85
(2)
69.25
62–80
58.75
25–76
(3) 9 (1)
85.37 77.62
71–96 67–100
83.25 72.5
60–100 54–95
(2)
82.75
61–100
80.12
60–94
(3) 10 (1)
43.5 41.0
15–61 11–80
45.87 37.37
9–58 7–82
(2)
33.87
1–75
38.62
8–91
(3)
42.5
7–82
31.75
1–77
Data depicted represent the average scaled scores across the listeners (n 5 8), as well as the minimum and maximum ratings in any sessions for both the all voiced and RP sentences. AV 5 all-voiced; RP 5 Rainbow Passage.
those areas explored through the SE and PF domain. Table IV presents summary data on the V-RQOL scores (SE, PF, and total) for individual participant speakers over the three points of data collection. Finally, in order to determine if any potential relationship existed between the speakers self-ratings of voice quality and the listener ratings of the speech stimuli obtained, we performed correlational analysis between LO ratings for both sentences (AV and RP) and the two V-RQOL subscores (SE and PF) and the total score. Based on this analysis, both the SE and PF subscores and the total score were found to be moderately and significantly but negatively correlated with the AV sentence (r 5 2.403, P < 0.05; r 5 2.459, P < 0.05; and r 5 2.457, P < 0.05), respectively. This negative correlation was expected. As V-RQOL scores increase, one would expect the perceived overpressure and associated impact on voice control to decrease. In contrast, however, Laryngoscope 125: March 2015
664
no significant correlations were noted between any VRQOL score and the RP sentence.
DISCUSSION This study sought to build on previous work that focused on the use of the auditory–perceptual measure of LO as a simple and reliable means of documenting the voice quality of individuals diagnosed with AdSD. If an association between this listener-based metric and participants’ self-assessments of V-RQOL was to emerge, it would greatly strengthen the utility and validity of LO as a clinical method of voice description for those with AdSD. The present data provide evidence that the evaluation of LO in this speaker group did correlate with self-generated assessments of their own voice disability (as determined by the V-RQOL scores) in the context of the AV sentence. Interestingly, relative to the Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
TABLE III. Mean Values and SDs for the Physical Functioning, SocialEmotional, and Total V-RQOL Scores Across All Speakers Over Three Sessions. Social-Emotional
Physical Functioning
Total VRQOL
Mean
66.37
53.59
57.67
SD
25.3
27.06
26.35
SD 5 standard deviation; V-RQOL 5 voice-related quality of life.
auditory–perceptual scores between the AV and RP sentence stimuli, no significant differences were revealed. This suggests that the perceptual salience of LO as a clinical descriptor may be of more generalized value (i.e., it may not be stimuli-specific). Therefore, the ability to accurately detect changes in LO as per the scaled measure of interest is potentially enhanced. However, it is important to note that when sentence stimuli are considered in the context of self-ratings, only the AV stimuli exhibited a stronger relationship to the V-RQOL scores. Our group previously demonstrated strong intraand interrater reliability associated with listeners’ auditory–perceptual judgments of LO, and similar findings were demonstrated herein.19 Our previous work also identified correlation between judgments of LO and objective acoustic measurements. The present study adds to this foundation by confirming the reliability of this measure and determining its association to participants’ self-assessments of V-RQOL. The strained voice quality and effort required for voice production is readily apparent to the listener and disruptive to efficient communication in AdSD patients. Naturally, these types of deficits can be of significant detriment to patients’ communicative, and more directly, V-RQOL.3,14,16 Further, the inability to maintain a level of control one’s voice may contribute substantially to the voice disability experienced. Kaptein et al.4 reported that AdSD patients have low perceived control over their symptoms, which can have significant psychological consequences.3,17 As a result, gathering both LO and selfratings may serve as important indices of clinical change in those with AdSD. In the present study, we identified a moderate, negative correlation between listener ratings of LO for the AV sentence stimuli and all scores generated from the VRQOL. Previous studies have suggested a weak correlation between a patient’s V-RQOL score and clinicians’ perceptual judgment of voice impairment.22 In contrast, however, Hummel et al.23 compared objective measures of voice impairment using the Dysphonia Severity Index (DSI) to the patient-reported score on V-RQOL, but no correlation between these scores was found. Thus, the correlation between listener ratings and the V-RQOL found in the present study may be a function of the stimuli used as part of the auditory–perceptual phase of the study.11,12,24 This finding may provide evidence that ratings of LO, a feature that we believe has considerable face validity in the context of AdSD, is well-matched to a speaker’s selfperception of the voice disability that they experience with Laryngoscope 125: March 2015
this disorder. However, our data also suggest that the speaker’s self-assessment of the disability they experience does not uniformly correspond to a listener’s auditory–perceptual assessment of the voice. That is, a significant relationship to self-ratings was only found for the AV stimuli; thus, the ability to link auditory–perceptual measures with the V-RQOL may offer a unique contribution to clinical description of AdSD. Ultimately, we believe that these two measures assess two distinct clinical concerns: 1) the degree of voice abnormality (i.e., the extent of overpressure due to the disorder), and 2) the disability that truly results at both “physical” and social-emotional levels of functioning. Consequently, comprehensive clinical services that broadly address the disability experienced by those with AdSD appear to be justified. Although the V-RQOL may be an excellent predictor of voice status in some individuals,14 an identical change
TABLE IV. V-RQOL Scores (Social-Emotional, Physical, and Total) for Individual Speakers Over Three Data Collection Sessions. Speaker (Session)
1 (1) (2) (3) 2 (1) (2)
Social-Emotional Domain
Physical Domain
Total V-RQOL Score
52.5
50
54.17
31.25
29.17
30
62.5 62.5
54.17 66.67
57.5 65
100
100
100
(3) 3 (1)
81.25 81.25
29.17 79.17
50 80
(2)
93.75
91.67
92.5
87.5 87.5
83.33 79.17
85 82.5
(3) 4 (1) (2)
100
91.67
95
(3) 5 (1)
93.75 87.5
83.33 58.33
87.5 70
(2)
68.75
45.83
55
(3) 6 (1)
93.75 31.25
70.83 37.5
80 35
(2)
43.75
25
32.5
(3) 7 (1)
37.5 56.25
41.67 41.67
40 17.5
(2)
31.25
16.67
22.5
(3) 8 (1)
6.25 87.5
0 87.5
2.5 87.5
(2)
93.75
87.5
90
(3) 9 (1)
* 81.25
* 33.33
* 52.5
(2)
56.25
25
37.5
(3) 10 (1)
68.75 37.5
29.17 29.17
45 32.5
(2)
56.25
37.5
45
(3)
56.25
45.83
50
Data depicted represent the scores generated as per V-RQOL scoring algorithm. *Data discarded due to incomplete questionnaire. V-RQOL 5 voice-related quality of life.
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
665
in symptom severity may not objectively represent a commensurate change in perceived V-RQOL. A similar lack of correlation exists for other auditory–perceptual scales and validated quality of life measures for AdSD, which also suggests that patients’ internal standards likely differ from those of experienced listeners.12,22 Finally, although the present study was able to generate important longitudinal data for both LO and selfassessments of voice disability secondary to AdSD, we were unable to make any assessment of participant performance with respect to one another because time point A for participant 1 may not have been equivalent to time point A for participant 2, and participant 3, etc. Nevertheless, the present data do support the idea that regular assessments of those with AdSD at both levels addressed in the current study may be of benefit in measuring the variability of voice symptoms in patients. Data of this type may then serve to identify patterns of vocal change that may influence decisions on the ideal time for Botox treatment. A potential limitation of the present study may be found in the listeners who provided auditory–perceptual ratings of speech samples or the stimuli employed. We purposely selected this knowledgeable yet “nonexpert” group of listeners because the listeners tend to produce highly correlated ratings of LO, more so than their inexperienced counterparts, as suggested by our previous work.19 Thus, those who have no experience (na€ıve) or who have considerable experience with this population (experts) may provide very different ratings of LO. Thus, the effect of listener experience would likely bring with it a greater level of reliability, particularly in the context of the definition of LO that we have developed and employed in this phase of our project. As a potential counter to such limitations, we believe that the individuals’ data acquired over time for both LO ratings and the V-RQOL provide a rich and important body of information specific to the variability of AdSD over time. In this respect, group means for either measure ultimately do not adequately represent individual voice deficits or a speaker’s self-perception of their own voice-related disability in those with AdSD. As such, we believe that the present data provide support for clinical applications of the LO scaling method as a means of indexing adductory deficits associated with AdSD.
CONCLUSION This study identified an association between the listener ratings of LO for a stimulus sentence comprised of all-voiced sound and both subscores and the total V-RQOL score. Thus, variations in a participant’s self-perceived VRQOL did correlate with listener assessments of LO. The existence of a strong association between participantbased and listener-based perceptual assessments, along with the previously documented association with objective acoustic measures, provides a basis for the use of the LO scale as a simple, easy-to-use perceptual metric, both in clinical and research settings. The findings of this study also support the importance of ongoing participant selfassessments of perceived voice disability as a consequence of AdSD. Lastly, the combined use of auditory–perceptual
Laryngoscope 125: March 2015
666
judgments and self-ratings of voice would appear to provide a more complete perspective of voice disability that may not be achieved through use of single measures.
BIBLIOGRAPHY 1. Cannito MP, Doiuchi M, Murry T, Woodson GE. Perceptual structure of adductor spasmodic dysphonia and its acoustic correlates. J Voice 2012; 26:818.e5–818.e 13. doi:10.1016/j.jvoice.2012.05.005. 2. Blitzer A, Brin MF. Treatment of spasmodic dysphonia (laryngeal dystonia) with local injections of Botulinum toxin. J Voice 1992;6:365–369. doi:10.1016/S0892-1997(05)80035-9. 3. Baylor CR, Yorkston KM, Eadie TL. The consequences of spasmodic dysphonia on communication-related quality of life: a qualitative study of the insider’s experiences. J Commun Disord 2005;38:395–419. doi: 10.1016/j.jcomdis.2005.03.003. 4. Kaptein AA, Hughes BM, Scharloo M, Hondebrink N, Langeveld TPM. Psychological aspects of adductor spasmodic dysphonia: a prospective population controlled questionnaire study. Clin Otolaryngol 2010;35:31– 38. doi:10.1111/j.1749-4486.2009.02070.x. 5. Blitzer A, Brin MF, Stewart CF. Botulinum toxin management of spasmodic dysphonia (laryngeal dystonia): a 12-year experience in more than 900 patients. Laryngoscope 1998;108:1435–1441. doi:10.1097/ 00005537-199810000-00003. 6. Zwirner P, Murry T, Swenson M, Woodson GE. Effects of botulinum toxin therapy in patients with adductor spasmodic dysphonia: acoustic, aerodynamic, and videoendoscopic findings. Laryngoscope 1992;102:400–406. doi:10.1288/00005537-199204000-00006. 7. Stewart CF, Allen EL, Tureen P, Diamond BE, Blitzer A, Brin MF. Adductor spasmodic dysphonia: standard evaluation of symptoms and severity. J Voice 1997;11:95–103. doi:10.1016/S0892-1997(97)80029-X. 8. Fisher K V, Scherer RC, Guo CG, Owen AS. Longitudinal phonatory characteristics after botulinum toxin type A injection. J Speech, Lang Hear Res 1996;39:968–980. doi:10.1044/jshr.3905.968. 9. Sapienza CM, Crary MA, Gorham M. Laryngeal aerodynamic aspects of women with adductor spasmodic dysphonia. Arch Otolaryngol Head Neck Surg 1996;122:385–388. doi:10.1001/archotol.1996.01890160027005. 10. Shipp T, Izdebski K, Schutte HK, Morrissey P. Subglottal air pressure in spastic dysphonia speech. Folia Phoniatr Logop 1988;40:105–110. doi: 10.1159/000265895. 11. Roy N, Gouse M, Mauszycki SC, Merrill RM, Smith ME. Task specificity in adductor spasmodic dysphonia versus muscle tension dysphonia. Laryngoscope 2005;115:311–316. doi:10.1097/01.mlg.0000154739.48314.ee. 12. Eadie TL, Kapsner M, Rosenzweig J, Waugh P, Hillel A, Merati A. The role of experience on judgments of dysphonia. J Voice 2010;24:564–573. doi:10.1016/j.jvoice.2008.12.005. 13. Kreiman J, Gerratt BR. Validity of rating scale measures of voice quality. J Acoust Soc Am 1998;104:1598–1608. doi:10.1121/1.424372. 14. Rubin AD, Wodchis WP, Spak C, Kileny PR, Hogikyan ND. Longitudinal effects of Botox injections on voice-related quality of life (V-RQOL) for patients with adductory spasmodic dysphonia: part II. Arch Otolaryngol Head Neck Surg 2004;130:415–420. doi:10.1001/archotol.130.4.415. 15. Eadie TL, Doyle PC. Direct magnitude estimation and interval scaling of pleasantness and severity in dysphonic and normal speakers. J Acoust Soc Am 2002;112:3014–3021. doi:10.1121/1.1518983. 16. Hogikyan ND, Wodchis WP, Spak C, Kileny PR. Longitudinal effects of botulinum toxin injections on voice-related quality of life (V-RQOL) for patients with adductory spasmodic dysphonia. J Voice 2001;15:576–586. doi:10.1016/S0892-1997(01)00060-1. 17. Paniello RC, Barlow J, Serna JS. Longitudinal follow-up of adductor spasmodic dysphonia patients after botulinum toxin injection: quality of life results. Laryngoscope 2008;118:564–568. doi:10.1097/MLG. 0b013e31815e8be0. 18. Izdebski K. Overpressure and breathiness in spastic dysphonia: an acoustic (LTAS) and perceptual study. Acta Otolaryngol 1984;97:373–378. Available at: http://informahealthcare.com/doi/abs/10.3109/00016488409131003. Accessed February 20, 2014. 19. Yeung J, Fung K, Bornbaum CC, et al. Clinical approach to monitoring variability associated with adductor spasmodic dysphonia. J Otolaryngol Head Neck Surg 2011;40:343–349. 20. Hogikyan ND, Sethuraman G. Validation of an instrument to measure voice-related quality of life (V-RQOL). J Voice 1999;13:557–569. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10622521. 21. Fairbanks G. Voice and Articulation Drillbook. 2nd ed. NewYork, NY: Harper and Row; 1960. 22. Braden MN, Johns MM, Klein AM, Delgaudio JM, Gilman M, Hapner ER. Assessing the effectiveness of botulinum toxin injections for adductor spasmodic dysphonia: clinician and patient perception. J Voice 2010;24: 242–249. doi:10.1016/j.jvoice.2008.08.003. 23. Hummel C, Scharf M, Schuetzenberger A, Graessel E, Rosanowski F. Objective voice parameters and self-perceived handicap in dysphonia. Folia Phoniatr Logop 2010;62:303–307. doi:10.1159/000287715. 24. Zwirner P, Murry T, Woodson GE. Perceptual-acoustic relationships in spasmodic dysphonia. J Voice 1993;7:165–171. doi:10.1016/S08921997(05)80347-9.
Yeung et al.: Longitudinal Variations of LO and V-RQOL in AdSD
