- Email: [email protected]
The Effect of a Vocal Loading Test on Cough and Phonation in Patients With Chronic Cough
ARTICLE IN PRESS The Effect of a Vocal Loading Test on Cough and Phonation in Patients With Chronic Cough *,†,‡Anne E. Vertigan, *,†,‡Sarah M. Kapela, §Ingolf Franke, and *,†,‡Peter G. Gibson, *‡Newcastle, and †Australia, and §Forchheim, Germany
Summary: Objective/Hypothesis. Talking is a significant trigger for cough in patients with chronic cough; however, the stimulus required to trigger cough has not been quantified. The aim of this study was to examine the effect of a vocal loading task on phonation and cough behavior in patients with chronic cough and identify change following therapy. Study Design. This is a prospective observational study. Methods. This study involved 33 patients with chronic cough. Participants were assessed with the lingWAVES Vocal Loading Test protocol before and after intervention for chronic cough. Results. At baseline, almost 40% of patients had impaired vocal function and were unable to complete the vocal loading test. This improved following therapy, with 94% of patients being able to complete the test at follow-up. There was difficulty maintaining phonation, with 60% of the task unvoiced at baseline. This improved following therapy. The vocal loading test triggered coughing in 58% of patients; however, this improved following intervention. Acoustic measures during the vocal loading test did not change following therapy. Conclusion. Phonation is an important trigger for cough. Patients with chronic cough demonstrated impaired performance on tests of vocal loading. Most parameters improved following therapy. Key Words: Vocal loading–Dysphonia–Voice assessment–Chronic cough–Speech pathology.
INTRODUCTION Chronic cough is a common medical condition and the main reason for seeking healthcare in the ambulatory setting.1 It affects between 8% and 12% of the adult population.2 Cough persists despite medical management in 46% of patients and is termed chronic refractory cough.3 Talking is a significant trigger for coughing in patients with chronic cough,4,5 which impacts morbidity and quality of life. In a prospective study of patients with chronic refractory cough, talking triggered cough in 71% and talking on the telephone triggered cough in 56%. In patients with cough due to asthma, talking triggers cough in 22%.6 Voice symptoms are common in patients with chronic cough, and 40% have clinically significant abnormalities detected on auditory perceptual voice analysis.7 Although dysphonia is common in patients with chronic cough, the link between dysphonia and cough is unknown. Although talking is a significant trigger for cough, the duration and intensity of phonation required to trigger coughing are unknown. Understanding the impact of vocal loading on cough and dysphonia in patients with chronic cough and identifying changes after therapy have the potential to improve our understanding of the role of phonation in triggering cough. A normal healthy voice can be used for 4–8 hours without any vocal load problems.8,9 If voice problems occur during this time period, it is possible that a voice disorder exists. Under clinical conditions, it is not feasible to investigate a patient’s voice Accepted for publication March 28, 2017. Conflict of interest: AEV, SMK and PGG declare no conflicts of interest. From the *John Hunter Hospital, Newcastle, Australia; †Centre for Asthma, and Respiratory Disease, University of Newcastle, Newcastle, Australia; ‡Hunter Medical Research Institute, Newcastle, Australia; and the §WEVOSYS Medical Technology, Forchheim, Germany. Address correspondence and reprint requests to Anne E. Vertigan, John Hunter Hospital, Locked Bag 1, Hunter Region Mail Centre, Newcastle, NSW 2310, Australia. E-mail: [email protected] Journal of Voice, Vol. ■■, No. ■■, pp. ■■-■■ 0892-1997 © 2016 The Voice Foundation. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jvoice.2017.03.020
over several hours. This problem can be overcome by using vocal loading tests that are designed to simulate a vocal load in a shorter period of time. Previous studies of vocal loading have examined healthy individuals,9–11 student teachers,12 and individuals with voice disorders.9 Despite the prevalence of dysphonia in patients with chronic cough,7 and the potential for phonation to trigger cough, vocal loading has not been formally examined in these patients. The aim of this study was to examine the effect of a vocal loading task on phonation and cough behavior in patients with chronic cough and identify change following therapy. We hypothesized that the vocal loading task would trigger cough and objectively identify phonatory problems in chronic cough. Participants Participants included 33 patients with chronic refractory cough who were undergoing treatment as part of a randomized trial of speech pathology and pregabalin treatments.13 These participants had cough that had persisted after medical treatment based on the anatomic diagnostic protocol.14 Demographic information for the participants is provided in Table 1. Inclusion criteria were age between 18 and 80 years and unexplained cough (ie, no associated diagnoses) or refractory cough (ie, cough that persisted after treatment of associated diagnoses of asthma, rhinitis, gastroesophageal reflux disease including proton pump inhibitors, H2 agonists and lifestyle modifications, and angiotensin converting enzyme inhibitor use). Exclusion criteria included cough productive of purulent sputum, current smoking, pregnancy or breast-feeding, other active respiratory disease, respiratory tract infection during the month prior to randomization, significant psychiatric or neurologic disorder, laryngeal lesions, and previous speech pathology intervention for cough or dysphonia within the last 12 months. Previous medical treatment for cough was assessed and patients who had not completed
ARTICLE IN PRESS 2 TABLE 1. Participant Characteristics Age (years), mean (SD) 62.6 (11.9) Gender (% female) 66 Duration of cough, months, mean (SD) 135 (120) Smoking Never smoked 19 (58) Ex-smoker 14 (42) Previous treatments for cough, number (%) Nonprescription medication 9 (27) Oral corticosteroids 6 (18) Antibiotics 8 (24) Codeine/opiates 5 (15) Inhaled medication 24 (73) Proton pump inhibitors/H2 antagonists 26 (79) Nasal corticosteroids 17 (52) Antihistamines 8 (24) Lifestyle modification for reflux 7 (21) Gabapentin 4 (12) CPAP 11 (3) FEV1 (% predicted), mean (SD) 85.5 (16.5) FVC (% predicted), mean (SD) 83.5 (15.6) Leicester Cough Questionnaire Score, 11.9 (3.3) mean (SD) 24-Hour ambulatory cough frequency, 16 (10.4–34.7) coughs per hour, median (IQR) Abbreviations: CPAP, continuous positive airway pressure; FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; IQR, interquartile range; SD, standard deviation.
their treatment were excluded from the study until they had appropriate investigations and treatment. Baseline voice assessment data for all participants are reported in Table 2. This study was approved by the Hunter New England Research Ethics Committee (Reference number 11/11/16/3.04). All participants provided written informed consent to participate in the study. Eleven participants (33%) were male and 22 (66%) were female. The average age was 62 years old (standard deviation [SD] = 12). Twenty-five participants (63%) reported voice symptoms such as voice loss or hoarseness, in addition to chronic cough at baseline.
Journal of Voice, Vol. ■■, No. ■■, 2017
Procedure Vocal loading was assessed using the lingWAVES Vocal Loading Test. The Vocal Loading Test is a module on the lingWAVES voice analysis program. lingWAVES is a voice and speech assessment system consisting of software (running on Windows computers) and a standardized recording hardware, USB connector, and a precalibrated and certified sound level meter. The hardware provides microphone and sound level data in real time to the software application. lingWAVES fulfills the recommendations of the Union of the European Phoniatrics for voice range profile recordings (30-cm distance between microphone/sound pressure level [SPL] meter and client’s mouth, and measurement in dB[A]). lingWAVES is a certified European Union medical product. The lingWAVES Vocal Loading Test controls vocal load and endurance of the speaking voice by controlling the duration and intensity of phonation. In addition to measuring traditional parameters such as SPL and fundamental frequency, it measures correlates of audible changes in voice quality. These features include roughness, which correlates with irregularity, and breathiness, which correlates with noise of the vocal fold activity. These signs are measured and analyzed with the lingWAVES Vocal Loading Test in real time. The vocal loading test requires participants to alternately repeat the phonemes /ɛ/ and / æ/ for approximately 1 second each for 16 minutes. The target intensity was set at two different predetermined levels, 70 and 75 decibels sound pressure level (dB SPL), which are the default settings in the lingWAVES Vocal Loading Test program. These intensity levels are referred to as low (70 dB SPL) and high (75 dB SPL) conditions. The test lasted for 16 minutes and ended before the target 16 minutes if the participant became short of breath or distressed. The target intensity level alternated between low and high conditions every minute. Continuous vocalization has been recommended to produce the highest level of vocal load during a short period (16 minutes).15 The 70 dB SPL level is related to normal slightly increased speaking, for example in a classroom, whereas the 75 dB SPL level is slightly louder, for example talking in a classroom over background noise. Voice samples were recorded using a WEVOSYS SPL meter microphone positioned 30 cm from the participant’s mouth. The lingWAVES Vocal Loading Test provides continuous real-time visual feedback comparing actual and target intensity on a com-
TABLE 2. Baseline Voice Assessment Data
Maximum phonation time (seconds) Jitter (%) Shimmer (%) Average fundamental frequency (Hz) Male Female Consensus Auditory Perceptual Evaluation-Voice Voice Handicap Index Abbreviations: CI, confidence interval; SD, standard deviation.
Mean (SD)
Range
95% CI
12.2 (6.7) 1.2 (2.0) 12.4 (10.0)
1.8–28.6 0.07–7.7 8.8–15.9
9.8–14.5 0.5–1.9 38.8–54.9
0–88 1–112
114–134 118–216 22.5–35.6 16.4–33.8
124 (13.6) 202 (34) 29.0 (20.5) 25.1 (27.2)
ARTICLE IN PRESS Anne E. Vertigan, et al
Effect of Vocal Loading Test on Cough and Phonation
3
FIGURE 1. Visual feedback obtained for one participant during the vocal loading test. The target sound pressure levels for the high and low decibel conditions for each minute of the vocal loading test are shown in red. The actual sound pressure level is shown in blue. The fundamental frequency is shown in green. The target sound pressure level is reached most of the time; however, the fundamental frequency, irregularity, and noise increased throughout the test. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
puter screen for the duration of the test (Figure 1). This enabled participants to adjust their intensity if it was not at the target level. The test did not begin if the participant had been experiencing a coughing episode. The Vocal Loading Test was administered at three time points, namely (1) baseline; (2) at the completion of treatment, that is, posttreatment 15 weeks following baseline; and (3) at followup, which was 18 weeks following baseline. All participants received five sessions of speech pathology intervention for chronic cough between baseline and posttreatment assessments. The aim of the treatment was to increase voluntary control over cough symptoms and to reduce laryngeal irritation. The treatment involved education, strategies to minimize laryngeal irritation, cough suppression exercises, and psychoeducational counseling. Specific details about the treatment program are outlined elsewhere.13,16,17 In addition to the speech pathology treatment,
participants were randomized to concurrently receive either pregabalin or placebo medication.13 Data analysis The data obtained from the vocal loading test included task completion, task accuracy, the effect of the task on cough, and the effect of the task on acoustic voice measures, and are outlined in Table 3. Task completion included the percentage of patients who could complete the test, the total time which is the average duration of the test, and the percentage of the task that was completed. Task accuracy data included the percentage of the task that was unvoiced, that is, where no phonation occurred, and the percentage time below the target SPL. Task accuracy data were analyzed for the entire sample and then separately for the low (70 dB SPL) and high (75 dB SPL) conditions.
ARTICLE IN PRESS 4
Journal of Voice, Vol. ■■, No. ■■, 2017
TABLE 3. Data Obtained During the Vocal Loading Test (VLT) at Baseline, Posttreatment, and Follow-Up Domain
Measure
Description
Task completion Percentage of patients who could complete the test Total time
The total duration of the test. The normal and maximum is 16 minutes, and any values less than 16 minutes suggest that the test could not be completed. Percentage complete Percentage of the vocal loading task completed. Task accuracy Percentage of task unvoiced The percentage of time with absent phonation. Higher values denote worse performance. Percentage below sound Percentage of task with phonation below target sound pressure level: % pressure level below sound pressure level is the percentage of time that the vocalization was below the designated sound pressure level during the test. The target is 0, and higher values denote worse performance. Effect of VLT Cough threshold The number of minutes of the vocal loading test that were completed on cough before a single cough was elicited. Normal performance was behavior considered to be >16 minutes, that is, no coughs elicited throughout the procedure. C5 The number of minutes that were completed before five coughs or more were elicited during any 1-minute interval of the procedure. This is a common measure used in the objective assessment of cough sensitivity with tussive agents such as capsaicin or citric acid. Effect of VLT on Irregularity An acoustic correlate of rough vocal quality. Values range from 0 to 3, acoustic voice and values below 1 are considered normal. measures Noise An acoustic correlate of breathy vocal quality. Values range from 0 to 3, and values below 1 are considered normal. Average fundamental An acoustic correlate of pitch. Normal values are between 90 and 120 Hz frequency for males and between 80 and 220 Hz for females.
The effect of the lingWAVES Vocal Loading Test on cough behavior was determined by measuring the cough threshold, that is, the number of minutes of the vocal loading test that was completed before a single cough was elicited, and C5 which is the number of minutes that was completed before five or more coughs were elicited. The effect of the lingWAVES Vocal Loading Test on acoustic voice measures was determined by measuring changes in fundamental frequency, irregularity, and noise. These measures were analyzed separately for the low and high decibel conditions. Fundamental frequency data were analyzed separately for males and females in the low and high conditions. Data were analyzed as mean (SD) and 95% confidence interval (CI). Changes between baseline and end-treatment, and between baseline and follow-up, were analyzed using a paired sample t test. A paired sample t test was used to examine differences between low (70 dB) and high (75 dB) decibel conditions and between males and females for fundamental frequency. Comparison of patients taking pregabalin versus placebo was conducted using an independent t test. The significance level was set at P < 0.01. RESULTS All patients received speech pathology intervention, as well as either placebo or pregabalin tablets. Data are reported for all participants as well as separately for those receiving
pregabalin (SPEECH + PREG) versus placebo (SPEECH + PLAC). Task completion Twenty (61%) patients were able to complete the entire 16 minutes of the vocal loading task at baseline compared with 23 (70%) at posttreatment (P = 0.432) and 31 (94%) at follow-up (P = 0.008). In patients receiving SPEECH + PREG, 8 (33%) were able to complete the vocal loading task at baseline, and 10 (30%) at both posttreatment and follow-up. In patients receiving SPEECH + PLAC, 12 (36%) completed the vocal loading test at baseline, 13 (39%) at posttreatment, and 15 (45%) at followup. Patients who perceived a voice problem at baseline were less likely to complete the 16-minute test compared with those who did not perceive a voice problem (40% vs 93% respectively, P < 0.001) (Figure 2). The total time completed and the percentage of task completed at baseline, end-treatment, and follow-up for all participants combined and separately for those receiving SPEECH + PREG and SPEECH + PLAC are shown in Table 4. There was no significant change in total time completed between baseline and end-treatment (P = 0.896), or between baseline and follow-up (P = 0.270). Likewise, there was no significant difference in the percentage of task completed between baseline and endtreatment (P = 0.675) and between baseline and follow-up (P = 0.567). There was no significant difference between
ARTICLE IN PRESS Anne E. Vertigan, et al
5
Effect of Vocal Loading Test on Cough and Phonation
FIGURE 2. Comparison of the percentage of patients who could complete the vocal loading test according to whether they perceived voice problems at baseline. SPEECH + PREG and SPEECH + PLAC groups at baseline (P = 0.327), posttreatment (P = 0.503), or follow-up (P = 0.388). Task accuracy Task accuracy (Table 5) was measured using the percentage of unvoiced periods (ie, percentage of time with absent phonation; Table 3) during the test and the percentage of time below the target SPL (ie, ≥70 dB SPL in the low condition and ≥75 dB SPL in the high condition). Percentage unvoiced and percentage below SPL for the entire sample and high and low conditions at baseline, end-treatment, and follow-up are reported in Table 5. Participants demonstrated difficulty sustaining phonation during the task, with over 60% of the sample unvoiced at baseline. The percentage unvoiced was similar between the high and low decibel conditions (baseline, P = 0.012; end-treatment, P = 0.970; followup, P = 0.591). There was no significant difference between SPEECH + PREG and SPEECH + PLAC at baseline (P = 0.766), posttreatment (0.936), or follow-up (P = 0.427). For the entire sample, the mean improvement between baseline and posttreatment was 10.2% unvoiced (SD = 12.3, 95% CI: 5.7–14.6,
P < 0.001), and between baseline and follow-up was 8.2% unvoiced (SD = 11.5, 95% CI: 3.9–12.5, P = 0.001). For participants in SPEECH + PREG, the mean improvement between baseline and posttreatment was 10.1 (SD = 13.1, 95% CI: −8.8 to 9.3, P = 0.955), and between baseline and follow up was 7.4 (SD = 10.5, 95% CI: −6.7 to 10.5, P = 0.682). In the low decibel condition, the percentage unvoiced improved by a mean of 11.5% (SD = 12.6, 95% CI: 6.8–16.11, P < 0.001) between baseline and posttreatment, and 8.0% (SD = 12.2, 95% CI: 3.2–12.7, P = 0.002) between baseline and follow-up. In the high decibel condition, there was a mean improvement of 8.9% (SD = 13.6, 95% CI: 4.0–13.9, P = 0.001) between baseline and posttreatment, and 6.9% (SD = 11.4, 95% CI: 2.5–11.3, P = 0.004) between baseline and follow-up. There was no significant difference between SPEECH + PREG and SPEECH + PLAC between baseline and posttreatment (P = 0.863) or between baseline and follow-up (P = 0.596). The average percentage of time below target SPL was generally low (Table 5), indicating that participants were able to reach target SPL most of the time. The percentage below target was
TABLE 4. Task Completion Including Total Time and Percentage of the Vocal Loading Task Completed
Total time (minutes)
% of Task completed
* Compared with baseline, P = 0.896. † Compared with baseline, P = 0.270. ‡ Compared with baseline, P = 0.675. § Compared with baseline, P = 0.567.
Baseline Posttreatment Follow-up Baseline Posttreatment Follow-up
Mean (Standard Deviation)
Range (95% Confidence Interval)
12.7 (4.8) 12.8 (5.2)* 13.3 (4.5)† 77.8 (32.0) 79.9 (32.9)‡ 82.7 (9)§
1–16 (10.9–14.4) 2–16 (10.9–14.7) 2–16 (11.7–15.0) 0.06–100 (66.1–89.5) 6.8–100 (67.9–92.0) 11.3–100 (72–93.3)
ARTICLE IN PRESS 6 14.3–94.3 (55.5–69.6) 15.9–95.9 (46.1–58.8) 14.1–82.6 (47.2–58.8) 0–100 (11.7–30.8) 0–100 (12.6–33.9) 0–80 (6.9–22.7)
significantly greater in the high than the low decibel condition at each time point (baseline, P = 0.001; end-treatment, P = 0.001; and follow-up, P = 0.006), suggesting greater difficulty maintaining louder volume. There was no significant difference between SPEECH + PREG and SPEECH + PLAC at baseline (P = 0.839), posttreatment (P = 0.102), and follow-up (=0.953). For the entire sample, the mean change in percentage below SPL between baseline and end-treatment was 1.3% (SD = 17.8, 95% CI: 5.3–8.0, P = 0.585), and the mean change between baseline and follow-up was 4.2% (SD = 11.1, 95% CI: 0.1–8.4, P = 0.045). Improvement between SPEECH + PREG and SPEECH + PLAC groups was not significantly different from baseline to posttreatment (P = 0.064) or from baseline to followup (P = 0.614). In the low decibel condition, the mean change in percentage below SPL between baseline and end-treatment was 2.8% (SD = 13.7, 95% CI: −2.5 to 8.1, P = 0.210), and between baseline and follow-up was 1.5% (SD = 6.2, 95% CI: 0.9 to 3.9, P = 0.199). In the high decibel condition, the mean change between baseline and posttreatment was 2.1% (SD = 33.2, 95% CI: −11 to 15.3, P = 0.909), and between baseline and followup was 7.0% (SD = 20.6, 95% CI: −14.4 to 15.5, P = 0.101).
78.9–98.3 (56.8–67.9) 19.2–75.7 (48.0–57.8) 19.9–81.3 (47.6–59.3) 0.9–28.9 (3.9–8.3) 0.2–69.8 (4.5–14.0) 0.2–28.2 (2.4–7.2) * Compared with baseline, P < 0.001. † Compared with baseline, P = 0.001. ‡ Compared with baseline, P = 0.002. § Compared with baseline, P = 0.004. ‖ Compared with baseline, P = 0.585. ¶ Compared with baseline, P = 0.045. # Compared with baseline, P = 0.210. ** Compared with baseline, P = 0.199. †† Compared with baseline, P = 0.909. ‡‡ Compared with baseline, P = 0.101. Abbreviations: CI, confidence interval; M, mean; SD, standard deviation.
% Below target sound pressure level
% Unvoiced
Baseline Posttreatment Follow-up Baseline Posttreatment Follow-up
61.3 (15.2) 51.9 (14.0)* 52.7 (15.3)† 12.1 (10.7) 13.9 (17.7)‖ 8.6 (10.9)¶
16.8–93.9 (55.3–68.5) 17.5–76.8 (46.9–56.9) 17.0–8.08 (46.8–58.5) 0.5–35.2 (8.3–16.0) 0.1–84.5 (17.6–20.2) 0.2–47.0 (4.5–12.7)
63.3 (14.9) 52.9 (13.8)* 54.4 (15.7)‡ 6.1 (6.0) 9.3 (13.5)# 4.8 (6.4)**
61.3 (16.1) 52.9 (16.8)† 53.0 (15.6)§ 21.2 (23.7) 23.2 (30.0)†† 14.8 (21.0)‡‡
Range (95% CI)
High Decibel Condition
M (SD) Range (95% CI) M (SD)
Low Decibel Condition Entire Sample
Range (95% CI) M (SD)
TABLE 5. Task Accuracy Including Percentage Unvoiced and Percentage Below Target Sound Pressure Level at Baseline, Posttreatment, and Follow-Up
Journal of Voice, Vol. ■■, No. ■■, 2017
Effect of the vocal loading test on cough behavior The effect of the vocal loading test on cough behavior was examined by measuring the cough threshold, that is, the minimum amount of time before one or more coughs were elicited, and C5, that is, the duration of the test performed before five or more coughs were triggered. The cough threshold was reached in a substantial number of participants (19, 58%) at baseline. The number of participants reaching the cough threshold decreased to 11 (33%) at the posttreatment visit (P = 0.009) and 14 (42%) at follow-up (P = 0.110). For participants in the SPEECH + PREG group, cough threshold was reached in 7 (44%) participants at baseline and posttreatment, and 6 (38%) participants at follow up. For participants in the SPEECH + PLAC group, cough threshold was reached in 10 (59%) participants at baseline, 11 (65%) at posttreatment, and 8 (47%) at follow-up. There was no significant difference in improvement in cough threshold between SPEECH + PREG and SPEECH + PLAC groups (P > 0.01). C5 was reached in 14 (42%) of the participants at baseline, 4 (12%) at end-treatment (P < 0.001), and 3 (9%) at follow-up (P < 0.001). C5 was reached in 5 (31%) of the participants in the SPEECH + PREG group at baseline, and 1 (6%) at both posttreatment and follow up. In the SPEECH + PLAC group, 5 (29%) of the participants reached C5 at baseline, 3 (18%) at posttreatment, and 2 (12%) at follow-up. There was no significant difference in improvement in C5 between SPEECH + PREG and SPEECH + PLAC groups (P > 0.01). Effect of the vocal loading test on acoustic voice measures Acoustic voice measures included average fundamental frequency (AvF0), irregularity, and noise. The values for the low and high decibel conditions at baseline, end-treatment, and followup are shown in Table 6.
ARTICLE IN PRESS Anne E. Vertigan, et al
7
Effect of Vocal Loading Test on Cough and Phonation
TABLE 6. Effect of the Vocal Loading Test on Acoustic Voice Measures Measure Mean fundamental frequency: males (Hertz) Mean fundamental frequency: females (Hertz) Irregularity (%)
Noise (%)
Time Baseline End-treatment Follow-up Baseline End-treatment Follow-up Baseline End-treatment Follow-up Baseline End-treatment Follow-up
Low Decibel Condition
High Decibel Condition
M (SD)
95% CI
M (SD)
95% CI
120 (11) 131 (21) 128 (11) 203 (35) 200 (33) 196 (30) 1.4 (0.2) 1.3 (0.2) 1.4 (0.2) 0.6 (0.5) 0.5 (0.2) 0.5 (0.4)
111–129 117–145 119–136 187–218 185–214 182–209 1.3–1.5 1.2–1.4 1.3–1.4 0.04–0.8 0.4–0.5 0.3–0.6
130 (12) 143 (40) 139 (30) 212 (36) 208 (32) 201 (28) 1.4 (0.2) 1.3 (0.2) 1.3 (0.3) 0.6 (0.5) 0.5 (0.2) 0.4 (0.4)
120–140 116–145 116–162 196–227 194–222 188–213 1.3–1.4 1.2–1.4 1.2–1.4 0.4–0.7 0.3–0.5 0.3–0.6
Abbreviations: CI, confidence interval; M, mean; SD, standard deviation.
AvF0 was significantly higher in the high than in the low condition for males at baseline (P = 0.014) but not at end-treatment (P = 0.186) or follow-up (P = 0.149). For females, AvF0 was significantly higher in the high than in the low condition at baseline (P < 0.001) and end-treatment (P < 0.001) but not at follow-up (P = 0.047). At baseline, 29 (88%) of the participants demonstrated an increase of more than two semitones between the beginning and end of the test, indicating that AvF0 increased as the test progressed. Mean irregularity values were outside the normal range at baseline, end-treatment, and follow-up. There was no significant difference in irregularity values between the low and high decibel conditions at baseline (P = 0.399), posttreatment (P = 0.440), or follow-up (P = 0.058). There was no significant difference in irregularity values at baseline (P = 0.802), posttreatment (P = 0.233), or follow-up (P = 0.962). There was also no significant change in irregularity values between baseline and end-treatment or between baseline and follow-up (P > 0.01). Mean noise values were within the normal range at all three time periods. Noise values were lower in the high than in the low decibel condition at follow-up (P < 0.001), but not at baseline (P = 0.135) or posttreatment (P = 0.051). There was no significant difference in noise values between participants in either group at baseline (P = 0.249), posttreatment (P = 0.939), or followup (P = 0.157). There was no significant change in noise values between baseline and posttreatment or between baseline and follow-up (P > 0.01). DISCUSSION This is the first study to examine vocal loading in patients with chronic cough. We found a high prevalence of impaired vocalization and that phonation provoked cough; however, with treatment, these parameters significantly improved. Specifically, there was a high prevalence of abnormal results in the vocal loading task, including total time, percentage task complete, percentage unvoiced, percentage below SPL, irregularity, and C5. Of these, all but irregularity and percentage below SPL im-
proved following therapy. These results are summarized in Figure 3, and the values that improved following therapy are shown in Figure 3. Cough can be triggered by different classes of stimuli. This study objectively assessed phonation, a mechanical stimulus, as a trigger for cough. In contrast, cough challenges in other studies have employed chemical stimuli such as capsaicin or citric acid.18 It is possible that these different stimuli, that is, phonation and capsaicin, activate different afferent nerve fibers during cough challenge. The two main types of afferent nerve fibers are the C fibers (also known as chemosensors) and cough receptors (also known as mechanosensors). C fibers are unmyelinated19 and are sensitive to chemical stimuli. In contrast, cough receptors are myelinated5 and relatively less responsive to chemical stimuli.19 Phonation was a frequent trigger for cough in this study, which is consistent with previous studies involving patient self-report.4,6 However, the vocal loading test did not trigger coughing in all individuals in the current study. This result is in contrast to the effect of capsaicin on the same participants, whereby 32 (97%) of the participants coughed in response to capsaicin at baseline.13 There was a significant reduction in the extent to which the vocal loading task triggered coughing at posttreatment and followup. This result suggests that participants were becoming increasingly able to suppress their cough despite exposure to the same trigger (ie, phonation), or that their cough sensitivity to a mechanical stimulus was reduced following therapy. The ability to complete the vocal loading test was impaired at baseline but improved following therapy. This suggests that the therapy program improved vocal stamina. The participants who perceived coexisting voice symptoms at baseline had more difficulty completing the vocal loading test than those without coexisting voice symptoms. There was no significant difference in any outcome measure between participants in either group, confirming that pregabalin has little influence over vocal function. The percentage of time with absent phonation (ie, percentage unvoiced) was high at baseline due to difficulty sustaining
ARTICLE IN PRESS 8
Journal of Voice, Vol. ■■, No. ■■, 2017
FIGURE 3. Summary of normal versus abnormal results at baseline. Of the results that were abnormal at baseline, cough threshold, percentage participants complete, C5, and percentage unvoiced improved after therapy.
phonation. This result could be a sign of voice problems or a less powerful voice. The percentage unvoiced improved following intervention. Other measures including the percentage below target SPL and acoustic correlates of vocal quality did not change following therapy. The aim of the speech pathology treatment used in the current study was to teach cough suppression, and although it addressed some aspects of laryngeal function it was not a formal voice therapy program. It would be interesting to see whether these latter measures will improve more following a therapy program that directly targets phonation, such as resonant voice therapy or vocal function exercises. The high prevalence of abnormal results in this study suggests that the speech pathology assessment of the patient with chronic cough should incorporate assessment of phonation regardless of whether the patient presents with voice symptoms. Whether this needs to include a formal assessment of vocal loading is unclear. Formal vocal loading tests are time-consuming and require specialist equipment and therefore may not be feasible to conduct routinely in the clinical setting. However, they may be useful for specific patients with chronic cough, including those who are professional voice users. In addition to formal objective measures, the vocal loading test enables observation of changes in constriction, laryngeal muscle tension, and respiratory patterns. These features are a frequent precursor to cough, can often change throughout the course of the test, and may be indicative of the difficulties patients are experiencing in daily life. Anecdotally, most participants reported that the test was a
significant challenge and the most difficult portion of their participation in the entire study (Figure 4).13 Several methodological issues and limitations need to be considered when interpreting these results. The inclusion of a nontreatment control group would strengthen the interpretation of the results and are recommended in further studies of vocal loading in this population. Several types of vocal loading tests are available, and the choice of test used in the current study may have an influence on the results. Previous studies have used reading tasks with a target intensity of between 75 and 85 dB SPL.9,10,12,20 The advantage of the lingWAVES Vocal Loading Test is that the protocol is standardized, which enables consistent administration between assessments and between patients. The length of time between the three assessments was sufficient to negate a practice or fatigue effect. Previous studies have found that changing the target intensity during the test, which also occurred in the current study, resulted in more difficulty than when the target vocal load was kept constant throughout the test.11 This study examined the effect of prolonged phonation but did not examine other phonatory issues that could provoke cough. This study used a single baseline measure. It did not employ formal auditory perceptual voice ratings immediately pre and post each vocal loading session. Further studies could examine the utility of using shorter vocal loading times. Most participants who coughed during the test (81%) coughed between 1 and 7 minutes of the test, suggesting that the duration of the test could be shortened while still yielding clinically useful results.
ARTICLE IN PRESS Anne E. Vertigan, et al
Effect of Vocal Loading Test on Cough and Phonation
9
FIGURE 4. Summary of baseline, posttreatment, and follow-up results for the outcome measures with significant improvement.
CONCLUSION Objective measurement of the phonatory stimulus in chronic cough showed that in most patients with chronic refractory cough, the performance on this test of vocal loading was impaired at baseline and most parameters improved following therapy. The study demonstrates that phonation is an important trigger for cough and is the first to quantify the duration and intensity of phonation required to trigger cough. Acknowledgments The authors would like to acknowledge Carol Bishop from Multimedia Speech Pathology for training in the use of the Vocal Loading Test and Mary Aldrich for assistance with the figures. REFERENCES 1. Schappert S, Rechtsteiner E. Ambulatory medical care utilization estimates for 2007. In Vital and Health Stat, N.C.f.H. Statistics, editor, 2011. 2. Song WJ, Chang YS, Faruqi S, et al. The global epidemiology of chronic cough in adults: a systematic review and meta-analysis. Eur Respir J. 2015;45:1479–1481. 3. Haque R, Usmani O, Barnes P. Chronic idiopathic cough: a discrete clinical entity? Chest. 2005;127:1710–1713. 4. Vertigan A, Gibson P. Chronic refractory cough as a laryngeal sensory neuropathy: evidence from a reinterpretation of cough triggers. J Voice. 2011;25:596–601. 5. Vertigan AE, Theodoros DG, Gibson PG, et al. Voice and upper airway symptoms in people with chronic cough and paradoxical vocal fold movement. J Voice. 2007;21:361–383.
6. Matsumoto H, Tabuena RP, Niimi A, et al. Cough triggers and their pathophysiology in patients with prolonged or chronic cough. Allergol Int. 2012;61:132. 7. Vertigan AE, Theodoros DG, Winkworth AL, et al. Perceptual voice characteristics in chronic cough and paradoxical vocal fold movement. Folia Phoniatr Logop. 2007;59:256–267. 8. WEVOSYS. lingWAVES Vocal Load Test, 2014. Accessed June 9, 2016. https://wevosys.com/products/lingwaves/lingwaves_vocal_loading_test .html. 9. Hanschmann H, Gaipl C, Berger R. Preliminary results of a computer-assisted vocal load test with 10-min test duration. Eur Arch Otorhinolaryngol. 2011;268:309–313. 10. Whitling S, Rydell R, Lyberg Ahlander V. Design of a clinical vocal loading test with long-time measurement of voice. J Voice. 2015;29:10. 11. Echternach M, Richter B, Traser L, et al. Change of vocal capacity due to different vocal loading tests. Laryngorhinootologie. 2013;92:34–40. 12. Echternach M, Nusseck M, Dippold S, et al. Fundamental frequency, sound pressure level and vocal dose of a vocal loading test in comparison to a real teaching situation. Eur Arch Otorhinolaryngol. 2014;271:3263–3268. 13. Vertigan AE, Kapela SL, Ryan NM, et al. Pregabalin and speech pathology combination therapy for refractory chronic cough: a randomized controlled trial. Chest. 2016;149:639–648. 14. Irwin R, Boulet LP, Cloutier MM, et al. Managing cough as a defense mechanism and as a symptom: a consensus report for the American College of Chest Physicians. Chest. 1998;114:133S–181S. 15. Jilek C, Marienhagen J, Hacki T. Vocal stability in functional dysphonic versus healthy voices at different times of voice loading. J Voice. 2004;18:443–453. 16. Vertigan A, Theodoros DG, Winkworth A, et al. Chronic cough: a tutorial for speech language pathologists. J Med Speech Lang Pathol. 2007;15:189– 206.
ARTICLE IN PRESS 10 17. Vertigan A, Theodoros DG, Gibson PG, et al. Efficacy of speech pathology management for chronic cough: a randomised placebo controlled trial of treatment efficacy. Thorax. 2006;61:1065–1069. 18. Dicpinigaitis P. Short- and long-term reproducibility of capsaicin cough challenge testing. Pulm Pharmacol Ther. 2003;16:61–65.
Journal of Voice, Vol. ■■, No. ■■, 2017 19. Bessac B, Jordt S-E. Breathtaking TRP channels: trpa1 and TRPV1 in airway chemosensation and reflex control. Physiology (Bethesda). 2008;23:360–370. 20. Niebudek-Bogusz E, Kotyło P, Politan´ski P, et al. Acoustic analysis with vocal loading test in occupational voice disorders: outcomes before and after therapy. Int J Occup Med Environ Health. 2008;21:301–308.
Summary: Objective/Hypothesis. Talking is a significant trigger for cough in patients with chronic cough; however, the stimulus required to trigger cough has not been quantified. The aim of this study was to examine the effect of a vocal loading task on phonation and cough behavior in patients with chronic cough and identify change following therapy. Study Design. This is a prospective observational study. Methods. This study involved 33 patients with chronic cough. Participants were assessed with the lingWAVES Vocal Loading Test protocol before and after intervention for chronic cough. Results. At baseline, almost 40% of patients had impaired vocal function and were unable to complete the vocal loading test. This improved following therapy, with 94% of patients being able to complete the test at follow-up. There was difficulty maintaining phonation, with 60% of the task unvoiced at baseline. This improved following therapy. The vocal loading test triggered coughing in 58% of patients; however, this improved following intervention. Acoustic measures during the vocal loading test did not change following therapy. Conclusion. Phonation is an important trigger for cough. Patients with chronic cough demonstrated impaired performance on tests of vocal loading. Most parameters improved following therapy. Key Words: Vocal loading–Dysphonia–Voice assessment–Chronic cough–Speech pathology.
INTRODUCTION Chronic cough is a common medical condition and the main reason for seeking healthcare in the ambulatory setting.1 It affects between 8% and 12% of the adult population.2 Cough persists despite medical management in 46% of patients and is termed chronic refractory cough.3 Talking is a significant trigger for coughing in patients with chronic cough,4,5 which impacts morbidity and quality of life. In a prospective study of patients with chronic refractory cough, talking triggered cough in 71% and talking on the telephone triggered cough in 56%. In patients with cough due to asthma, talking triggers cough in 22%.6 Voice symptoms are common in patients with chronic cough, and 40% have clinically significant abnormalities detected on auditory perceptual voice analysis.7 Although dysphonia is common in patients with chronic cough, the link between dysphonia and cough is unknown. Although talking is a significant trigger for cough, the duration and intensity of phonation required to trigger coughing are unknown. Understanding the impact of vocal loading on cough and dysphonia in patients with chronic cough and identifying changes after therapy have the potential to improve our understanding of the role of phonation in triggering cough. A normal healthy voice can be used for 4–8 hours without any vocal load problems.8,9 If voice problems occur during this time period, it is possible that a voice disorder exists. Under clinical conditions, it is not feasible to investigate a patient’s voice Accepted for publication March 28, 2017. Conflict of interest: AEV, SMK and PGG declare no conflicts of interest. From the *John Hunter Hospital, Newcastle, Australia; †Centre for Asthma, and Respiratory Disease, University of Newcastle, Newcastle, Australia; ‡Hunter Medical Research Institute, Newcastle, Australia; and the §WEVOSYS Medical Technology, Forchheim, Germany. Address correspondence and reprint requests to Anne E. Vertigan, John Hunter Hospital, Locked Bag 1, Hunter Region Mail Centre, Newcastle, NSW 2310, Australia. E-mail: [email protected] Journal of Voice, Vol. ■■, No. ■■, pp. ■■-■■ 0892-1997 © 2016 The Voice Foundation. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jvoice.2017.03.020
over several hours. This problem can be overcome by using vocal loading tests that are designed to simulate a vocal load in a shorter period of time. Previous studies of vocal loading have examined healthy individuals,9–11 student teachers,12 and individuals with voice disorders.9 Despite the prevalence of dysphonia in patients with chronic cough,7 and the potential for phonation to trigger cough, vocal loading has not been formally examined in these patients. The aim of this study was to examine the effect of a vocal loading task on phonation and cough behavior in patients with chronic cough and identify change following therapy. We hypothesized that the vocal loading task would trigger cough and objectively identify phonatory problems in chronic cough. Participants Participants included 33 patients with chronic refractory cough who were undergoing treatment as part of a randomized trial of speech pathology and pregabalin treatments.13 These participants had cough that had persisted after medical treatment based on the anatomic diagnostic protocol.14 Demographic information for the participants is provided in Table 1. Inclusion criteria were age between 18 and 80 years and unexplained cough (ie, no associated diagnoses) or refractory cough (ie, cough that persisted after treatment of associated diagnoses of asthma, rhinitis, gastroesophageal reflux disease including proton pump inhibitors, H2 agonists and lifestyle modifications, and angiotensin converting enzyme inhibitor use). Exclusion criteria included cough productive of purulent sputum, current smoking, pregnancy or breast-feeding, other active respiratory disease, respiratory tract infection during the month prior to randomization, significant psychiatric or neurologic disorder, laryngeal lesions, and previous speech pathology intervention for cough or dysphonia within the last 12 months. Previous medical treatment for cough was assessed and patients who had not completed
ARTICLE IN PRESS 2 TABLE 1. Participant Characteristics Age (years), mean (SD) 62.6 (11.9) Gender (% female) 66 Duration of cough, months, mean (SD) 135 (120) Smoking Never smoked 19 (58) Ex-smoker 14 (42) Previous treatments for cough, number (%) Nonprescription medication 9 (27) Oral corticosteroids 6 (18) Antibiotics 8 (24) Codeine/opiates 5 (15) Inhaled medication 24 (73) Proton pump inhibitors/H2 antagonists 26 (79) Nasal corticosteroids 17 (52) Antihistamines 8 (24) Lifestyle modification for reflux 7 (21) Gabapentin 4 (12) CPAP 11 (3) FEV1 (% predicted), mean (SD) 85.5 (16.5) FVC (% predicted), mean (SD) 83.5 (15.6) Leicester Cough Questionnaire Score, 11.9 (3.3) mean (SD) 24-Hour ambulatory cough frequency, 16 (10.4–34.7) coughs per hour, median (IQR) Abbreviations: CPAP, continuous positive airway pressure; FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity; IQR, interquartile range; SD, standard deviation.
their treatment were excluded from the study until they had appropriate investigations and treatment. Baseline voice assessment data for all participants are reported in Table 2. This study was approved by the Hunter New England Research Ethics Committee (Reference number 11/11/16/3.04). All participants provided written informed consent to participate in the study. Eleven participants (33%) were male and 22 (66%) were female. The average age was 62 years old (standard deviation [SD] = 12). Twenty-five participants (63%) reported voice symptoms such as voice loss or hoarseness, in addition to chronic cough at baseline.
Journal of Voice, Vol. ■■, No. ■■, 2017
Procedure Vocal loading was assessed using the lingWAVES Vocal Loading Test. The Vocal Loading Test is a module on the lingWAVES voice analysis program. lingWAVES is a voice and speech assessment system consisting of software (running on Windows computers) and a standardized recording hardware, USB connector, and a precalibrated and certified sound level meter. The hardware provides microphone and sound level data in real time to the software application. lingWAVES fulfills the recommendations of the Union of the European Phoniatrics for voice range profile recordings (30-cm distance between microphone/sound pressure level [SPL] meter and client’s mouth, and measurement in dB[A]). lingWAVES is a certified European Union medical product. The lingWAVES Vocal Loading Test controls vocal load and endurance of the speaking voice by controlling the duration and intensity of phonation. In addition to measuring traditional parameters such as SPL and fundamental frequency, it measures correlates of audible changes in voice quality. These features include roughness, which correlates with irregularity, and breathiness, which correlates with noise of the vocal fold activity. These signs are measured and analyzed with the lingWAVES Vocal Loading Test in real time. The vocal loading test requires participants to alternately repeat the phonemes /ɛ/ and / æ/ for approximately 1 second each for 16 minutes. The target intensity was set at two different predetermined levels, 70 and 75 decibels sound pressure level (dB SPL), which are the default settings in the lingWAVES Vocal Loading Test program. These intensity levels are referred to as low (70 dB SPL) and high (75 dB SPL) conditions. The test lasted for 16 minutes and ended before the target 16 minutes if the participant became short of breath or distressed. The target intensity level alternated between low and high conditions every minute. Continuous vocalization has been recommended to produce the highest level of vocal load during a short period (16 minutes).15 The 70 dB SPL level is related to normal slightly increased speaking, for example in a classroom, whereas the 75 dB SPL level is slightly louder, for example talking in a classroom over background noise. Voice samples were recorded using a WEVOSYS SPL meter microphone positioned 30 cm from the participant’s mouth. The lingWAVES Vocal Loading Test provides continuous real-time visual feedback comparing actual and target intensity on a com-
TABLE 2. Baseline Voice Assessment Data
Maximum phonation time (seconds) Jitter (%) Shimmer (%) Average fundamental frequency (Hz) Male Female Consensus Auditory Perceptual Evaluation-Voice Voice Handicap Index Abbreviations: CI, confidence interval; SD, standard deviation.
Mean (SD)
Range
95% CI
12.2 (6.7) 1.2 (2.0) 12.4 (10.0)
1.8–28.6 0.07–7.7 8.8–15.9
9.8–14.5 0.5–1.9 38.8–54.9
0–88 1–112
114–134 118–216 22.5–35.6 16.4–33.8
124 (13.6) 202 (34) 29.0 (20.5) 25.1 (27.2)
ARTICLE IN PRESS Anne E. Vertigan, et al
Effect of Vocal Loading Test on Cough and Phonation
3
FIGURE 1. Visual feedback obtained for one participant during the vocal loading test. The target sound pressure levels for the high and low decibel conditions for each minute of the vocal loading test are shown in red. The actual sound pressure level is shown in blue. The fundamental frequency is shown in green. The target sound pressure level is reached most of the time; however, the fundamental frequency, irregularity, and noise increased throughout the test. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
puter screen for the duration of the test (Figure 1). This enabled participants to adjust their intensity if it was not at the target level. The test did not begin if the participant had been experiencing a coughing episode. The Vocal Loading Test was administered at three time points, namely (1) baseline; (2) at the completion of treatment, that is, posttreatment 15 weeks following baseline; and (3) at followup, which was 18 weeks following baseline. All participants received five sessions of speech pathology intervention for chronic cough between baseline and posttreatment assessments. The aim of the treatment was to increase voluntary control over cough symptoms and to reduce laryngeal irritation. The treatment involved education, strategies to minimize laryngeal irritation, cough suppression exercises, and psychoeducational counseling. Specific details about the treatment program are outlined elsewhere.13,16,17 In addition to the speech pathology treatment,
participants were randomized to concurrently receive either pregabalin or placebo medication.13 Data analysis The data obtained from the vocal loading test included task completion, task accuracy, the effect of the task on cough, and the effect of the task on acoustic voice measures, and are outlined in Table 3. Task completion included the percentage of patients who could complete the test, the total time which is the average duration of the test, and the percentage of the task that was completed. Task accuracy data included the percentage of the task that was unvoiced, that is, where no phonation occurred, and the percentage time below the target SPL. Task accuracy data were analyzed for the entire sample and then separately for the low (70 dB SPL) and high (75 dB SPL) conditions.
ARTICLE IN PRESS 4
Journal of Voice, Vol. ■■, No. ■■, 2017
TABLE 3. Data Obtained During the Vocal Loading Test (VLT) at Baseline, Posttreatment, and Follow-Up Domain
Measure
Description
Task completion Percentage of patients who could complete the test Total time
The total duration of the test. The normal and maximum is 16 minutes, and any values less than 16 minutes suggest that the test could not be completed. Percentage complete Percentage of the vocal loading task completed. Task accuracy Percentage of task unvoiced The percentage of time with absent phonation. Higher values denote worse performance. Percentage below sound Percentage of task with phonation below target sound pressure level: % pressure level below sound pressure level is the percentage of time that the vocalization was below the designated sound pressure level during the test. The target is 0, and higher values denote worse performance. Effect of VLT Cough threshold The number of minutes of the vocal loading test that were completed on cough before a single cough was elicited. Normal performance was behavior considered to be >16 minutes, that is, no coughs elicited throughout the procedure. C5 The number of minutes that were completed before five coughs or more were elicited during any 1-minute interval of the procedure. This is a common measure used in the objective assessment of cough sensitivity with tussive agents such as capsaicin or citric acid. Effect of VLT on Irregularity An acoustic correlate of rough vocal quality. Values range from 0 to 3, acoustic voice and values below 1 are considered normal. measures Noise An acoustic correlate of breathy vocal quality. Values range from 0 to 3, and values below 1 are considered normal. Average fundamental An acoustic correlate of pitch. Normal values are between 90 and 120 Hz frequency for males and between 80 and 220 Hz for females.
The effect of the lingWAVES Vocal Loading Test on cough behavior was determined by measuring the cough threshold, that is, the number of minutes of the vocal loading test that was completed before a single cough was elicited, and C5 which is the number of minutes that was completed before five or more coughs were elicited. The effect of the lingWAVES Vocal Loading Test on acoustic voice measures was determined by measuring changes in fundamental frequency, irregularity, and noise. These measures were analyzed separately for the low and high decibel conditions. Fundamental frequency data were analyzed separately for males and females in the low and high conditions. Data were analyzed as mean (SD) and 95% confidence interval (CI). Changes between baseline and end-treatment, and between baseline and follow-up, were analyzed using a paired sample t test. A paired sample t test was used to examine differences between low (70 dB) and high (75 dB) decibel conditions and between males and females for fundamental frequency. Comparison of patients taking pregabalin versus placebo was conducted using an independent t test. The significance level was set at P < 0.01. RESULTS All patients received speech pathology intervention, as well as either placebo or pregabalin tablets. Data are reported for all participants as well as separately for those receiving
pregabalin (SPEECH + PREG) versus placebo (SPEECH + PLAC). Task completion Twenty (61%) patients were able to complete the entire 16 minutes of the vocal loading task at baseline compared with 23 (70%) at posttreatment (P = 0.432) and 31 (94%) at follow-up (P = 0.008). In patients receiving SPEECH + PREG, 8 (33%) were able to complete the vocal loading task at baseline, and 10 (30%) at both posttreatment and follow-up. In patients receiving SPEECH + PLAC, 12 (36%) completed the vocal loading test at baseline, 13 (39%) at posttreatment, and 15 (45%) at followup. Patients who perceived a voice problem at baseline were less likely to complete the 16-minute test compared with those who did not perceive a voice problem (40% vs 93% respectively, P < 0.001) (Figure 2). The total time completed and the percentage of task completed at baseline, end-treatment, and follow-up for all participants combined and separately for those receiving SPEECH + PREG and SPEECH + PLAC are shown in Table 4. There was no significant change in total time completed between baseline and end-treatment (P = 0.896), or between baseline and follow-up (P = 0.270). Likewise, there was no significant difference in the percentage of task completed between baseline and endtreatment (P = 0.675) and between baseline and follow-up (P = 0.567). There was no significant difference between
ARTICLE IN PRESS Anne E. Vertigan, et al
5
Effect of Vocal Loading Test on Cough and Phonation
FIGURE 2. Comparison of the percentage of patients who could complete the vocal loading test according to whether they perceived voice problems at baseline. SPEECH + PREG and SPEECH + PLAC groups at baseline (P = 0.327), posttreatment (P = 0.503), or follow-up (P = 0.388). Task accuracy Task accuracy (Table 5) was measured using the percentage of unvoiced periods (ie, percentage of time with absent phonation; Table 3) during the test and the percentage of time below the target SPL (ie, ≥70 dB SPL in the low condition and ≥75 dB SPL in the high condition). Percentage unvoiced and percentage below SPL for the entire sample and high and low conditions at baseline, end-treatment, and follow-up are reported in Table 5. Participants demonstrated difficulty sustaining phonation during the task, with over 60% of the sample unvoiced at baseline. The percentage unvoiced was similar between the high and low decibel conditions (baseline, P = 0.012; end-treatment, P = 0.970; followup, P = 0.591). There was no significant difference between SPEECH + PREG and SPEECH + PLAC at baseline (P = 0.766), posttreatment (0.936), or follow-up (P = 0.427). For the entire sample, the mean improvement between baseline and posttreatment was 10.2% unvoiced (SD = 12.3, 95% CI: 5.7–14.6,
P < 0.001), and between baseline and follow-up was 8.2% unvoiced (SD = 11.5, 95% CI: 3.9–12.5, P = 0.001). For participants in SPEECH + PREG, the mean improvement between baseline and posttreatment was 10.1 (SD = 13.1, 95% CI: −8.8 to 9.3, P = 0.955), and between baseline and follow up was 7.4 (SD = 10.5, 95% CI: −6.7 to 10.5, P = 0.682). In the low decibel condition, the percentage unvoiced improved by a mean of 11.5% (SD = 12.6, 95% CI: 6.8–16.11, P < 0.001) between baseline and posttreatment, and 8.0% (SD = 12.2, 95% CI: 3.2–12.7, P = 0.002) between baseline and follow-up. In the high decibel condition, there was a mean improvement of 8.9% (SD = 13.6, 95% CI: 4.0–13.9, P = 0.001) between baseline and posttreatment, and 6.9% (SD = 11.4, 95% CI: 2.5–11.3, P = 0.004) between baseline and follow-up. There was no significant difference between SPEECH + PREG and SPEECH + PLAC between baseline and posttreatment (P = 0.863) or between baseline and follow-up (P = 0.596). The average percentage of time below target SPL was generally low (Table 5), indicating that participants were able to reach target SPL most of the time. The percentage below target was
TABLE 4. Task Completion Including Total Time and Percentage of the Vocal Loading Task Completed
Total time (minutes)
% of Task completed
* Compared with baseline, P = 0.896. † Compared with baseline, P = 0.270. ‡ Compared with baseline, P = 0.675. § Compared with baseline, P = 0.567.
Baseline Posttreatment Follow-up Baseline Posttreatment Follow-up
Mean (Standard Deviation)
Range (95% Confidence Interval)
12.7 (4.8) 12.8 (5.2)* 13.3 (4.5)† 77.8 (32.0) 79.9 (32.9)‡ 82.7 (9)§
1–16 (10.9–14.4) 2–16 (10.9–14.7) 2–16 (11.7–15.0) 0.06–100 (66.1–89.5) 6.8–100 (67.9–92.0) 11.3–100 (72–93.3)
ARTICLE IN PRESS 6 14.3–94.3 (55.5–69.6) 15.9–95.9 (46.1–58.8) 14.1–82.6 (47.2–58.8) 0–100 (11.7–30.8) 0–100 (12.6–33.9) 0–80 (6.9–22.7)
significantly greater in the high than the low decibel condition at each time point (baseline, P = 0.001; end-treatment, P = 0.001; and follow-up, P = 0.006), suggesting greater difficulty maintaining louder volume. There was no significant difference between SPEECH + PREG and SPEECH + PLAC at baseline (P = 0.839), posttreatment (P = 0.102), and follow-up (=0.953). For the entire sample, the mean change in percentage below SPL between baseline and end-treatment was 1.3% (SD = 17.8, 95% CI: 5.3–8.0, P = 0.585), and the mean change between baseline and follow-up was 4.2% (SD = 11.1, 95% CI: 0.1–8.4, P = 0.045). Improvement between SPEECH + PREG and SPEECH + PLAC groups was not significantly different from baseline to posttreatment (P = 0.064) or from baseline to followup (P = 0.614). In the low decibel condition, the mean change in percentage below SPL between baseline and end-treatment was 2.8% (SD = 13.7, 95% CI: −2.5 to 8.1, P = 0.210), and between baseline and follow-up was 1.5% (SD = 6.2, 95% CI: 0.9 to 3.9, P = 0.199). In the high decibel condition, the mean change between baseline and posttreatment was 2.1% (SD = 33.2, 95% CI: −11 to 15.3, P = 0.909), and between baseline and followup was 7.0% (SD = 20.6, 95% CI: −14.4 to 15.5, P = 0.101).
78.9–98.3 (56.8–67.9) 19.2–75.7 (48.0–57.8) 19.9–81.3 (47.6–59.3) 0.9–28.9 (3.9–8.3) 0.2–69.8 (4.5–14.0) 0.2–28.2 (2.4–7.2) * Compared with baseline, P < 0.001. † Compared with baseline, P = 0.001. ‡ Compared with baseline, P = 0.002. § Compared with baseline, P = 0.004. ‖ Compared with baseline, P = 0.585. ¶ Compared with baseline, P = 0.045. # Compared with baseline, P = 0.210. ** Compared with baseline, P = 0.199. †† Compared with baseline, P = 0.909. ‡‡ Compared with baseline, P = 0.101. Abbreviations: CI, confidence interval; M, mean; SD, standard deviation.
% Below target sound pressure level
% Unvoiced
Baseline Posttreatment Follow-up Baseline Posttreatment Follow-up
61.3 (15.2) 51.9 (14.0)* 52.7 (15.3)† 12.1 (10.7) 13.9 (17.7)‖ 8.6 (10.9)¶
16.8–93.9 (55.3–68.5) 17.5–76.8 (46.9–56.9) 17.0–8.08 (46.8–58.5) 0.5–35.2 (8.3–16.0) 0.1–84.5 (17.6–20.2) 0.2–47.0 (4.5–12.7)
63.3 (14.9) 52.9 (13.8)* 54.4 (15.7)‡ 6.1 (6.0) 9.3 (13.5)# 4.8 (6.4)**
61.3 (16.1) 52.9 (16.8)† 53.0 (15.6)§ 21.2 (23.7) 23.2 (30.0)†† 14.8 (21.0)‡‡
Range (95% CI)
High Decibel Condition
M (SD) Range (95% CI) M (SD)
Low Decibel Condition Entire Sample
Range (95% CI) M (SD)
TABLE 5. Task Accuracy Including Percentage Unvoiced and Percentage Below Target Sound Pressure Level at Baseline, Posttreatment, and Follow-Up
Journal of Voice, Vol. ■■, No. ■■, 2017
Effect of the vocal loading test on cough behavior The effect of the vocal loading test on cough behavior was examined by measuring the cough threshold, that is, the minimum amount of time before one or more coughs were elicited, and C5, that is, the duration of the test performed before five or more coughs were triggered. The cough threshold was reached in a substantial number of participants (19, 58%) at baseline. The number of participants reaching the cough threshold decreased to 11 (33%) at the posttreatment visit (P = 0.009) and 14 (42%) at follow-up (P = 0.110). For participants in the SPEECH + PREG group, cough threshold was reached in 7 (44%) participants at baseline and posttreatment, and 6 (38%) participants at follow up. For participants in the SPEECH + PLAC group, cough threshold was reached in 10 (59%) participants at baseline, 11 (65%) at posttreatment, and 8 (47%) at follow-up. There was no significant difference in improvement in cough threshold between SPEECH + PREG and SPEECH + PLAC groups (P > 0.01). C5 was reached in 14 (42%) of the participants at baseline, 4 (12%) at end-treatment (P < 0.001), and 3 (9%) at follow-up (P < 0.001). C5 was reached in 5 (31%) of the participants in the SPEECH + PREG group at baseline, and 1 (6%) at both posttreatment and follow up. In the SPEECH + PLAC group, 5 (29%) of the participants reached C5 at baseline, 3 (18%) at posttreatment, and 2 (12%) at follow-up. There was no significant difference in improvement in C5 between SPEECH + PREG and SPEECH + PLAC groups (P > 0.01). Effect of the vocal loading test on acoustic voice measures Acoustic voice measures included average fundamental frequency (AvF0), irregularity, and noise. The values for the low and high decibel conditions at baseline, end-treatment, and followup are shown in Table 6.
ARTICLE IN PRESS Anne E. Vertigan, et al
7
Effect of Vocal Loading Test on Cough and Phonation
TABLE 6. Effect of the Vocal Loading Test on Acoustic Voice Measures Measure Mean fundamental frequency: males (Hertz) Mean fundamental frequency: females (Hertz) Irregularity (%)
Noise (%)
Time Baseline End-treatment Follow-up Baseline End-treatment Follow-up Baseline End-treatment Follow-up Baseline End-treatment Follow-up
Low Decibel Condition
High Decibel Condition
M (SD)
95% CI
M (SD)
95% CI
120 (11) 131 (21) 128 (11) 203 (35) 200 (33) 196 (30) 1.4 (0.2) 1.3 (0.2) 1.4 (0.2) 0.6 (0.5) 0.5 (0.2) 0.5 (0.4)
111–129 117–145 119–136 187–218 185–214 182–209 1.3–1.5 1.2–1.4 1.3–1.4 0.04–0.8 0.4–0.5 0.3–0.6
130 (12) 143 (40) 139 (30) 212 (36) 208 (32) 201 (28) 1.4 (0.2) 1.3 (0.2) 1.3 (0.3) 0.6 (0.5) 0.5 (0.2) 0.4 (0.4)
120–140 116–145 116–162 196–227 194–222 188–213 1.3–1.4 1.2–1.4 1.2–1.4 0.4–0.7 0.3–0.5 0.3–0.6
Abbreviations: CI, confidence interval; M, mean; SD, standard deviation.
AvF0 was significantly higher in the high than in the low condition for males at baseline (P = 0.014) but not at end-treatment (P = 0.186) or follow-up (P = 0.149). For females, AvF0 was significantly higher in the high than in the low condition at baseline (P < 0.001) and end-treatment (P < 0.001) but not at follow-up (P = 0.047). At baseline, 29 (88%) of the participants demonstrated an increase of more than two semitones between the beginning and end of the test, indicating that AvF0 increased as the test progressed. Mean irregularity values were outside the normal range at baseline, end-treatment, and follow-up. There was no significant difference in irregularity values between the low and high decibel conditions at baseline (P = 0.399), posttreatment (P = 0.440), or follow-up (P = 0.058). There was no significant difference in irregularity values at baseline (P = 0.802), posttreatment (P = 0.233), or follow-up (P = 0.962). There was also no significant change in irregularity values between baseline and end-treatment or between baseline and follow-up (P > 0.01). Mean noise values were within the normal range at all three time periods. Noise values were lower in the high than in the low decibel condition at follow-up (P < 0.001), but not at baseline (P = 0.135) or posttreatment (P = 0.051). There was no significant difference in noise values between participants in either group at baseline (P = 0.249), posttreatment (P = 0.939), or followup (P = 0.157). There was no significant change in noise values between baseline and posttreatment or between baseline and follow-up (P > 0.01). DISCUSSION This is the first study to examine vocal loading in patients with chronic cough. We found a high prevalence of impaired vocalization and that phonation provoked cough; however, with treatment, these parameters significantly improved. Specifically, there was a high prevalence of abnormal results in the vocal loading task, including total time, percentage task complete, percentage unvoiced, percentage below SPL, irregularity, and C5. Of these, all but irregularity and percentage below SPL im-
proved following therapy. These results are summarized in Figure 3, and the values that improved following therapy are shown in Figure 3. Cough can be triggered by different classes of stimuli. This study objectively assessed phonation, a mechanical stimulus, as a trigger for cough. In contrast, cough challenges in other studies have employed chemical stimuli such as capsaicin or citric acid.18 It is possible that these different stimuli, that is, phonation and capsaicin, activate different afferent nerve fibers during cough challenge. The two main types of afferent nerve fibers are the C fibers (also known as chemosensors) and cough receptors (also known as mechanosensors). C fibers are unmyelinated19 and are sensitive to chemical stimuli. In contrast, cough receptors are myelinated5 and relatively less responsive to chemical stimuli.19 Phonation was a frequent trigger for cough in this study, which is consistent with previous studies involving patient self-report.4,6 However, the vocal loading test did not trigger coughing in all individuals in the current study. This result is in contrast to the effect of capsaicin on the same participants, whereby 32 (97%) of the participants coughed in response to capsaicin at baseline.13 There was a significant reduction in the extent to which the vocal loading task triggered coughing at posttreatment and followup. This result suggests that participants were becoming increasingly able to suppress their cough despite exposure to the same trigger (ie, phonation), or that their cough sensitivity to a mechanical stimulus was reduced following therapy. The ability to complete the vocal loading test was impaired at baseline but improved following therapy. This suggests that the therapy program improved vocal stamina. The participants who perceived coexisting voice symptoms at baseline had more difficulty completing the vocal loading test than those without coexisting voice symptoms. There was no significant difference in any outcome measure between participants in either group, confirming that pregabalin has little influence over vocal function. The percentage of time with absent phonation (ie, percentage unvoiced) was high at baseline due to difficulty sustaining
ARTICLE IN PRESS 8
Journal of Voice, Vol. ■■, No. ■■, 2017
FIGURE 3. Summary of normal versus abnormal results at baseline. Of the results that were abnormal at baseline, cough threshold, percentage participants complete, C5, and percentage unvoiced improved after therapy.
phonation. This result could be a sign of voice problems or a less powerful voice. The percentage unvoiced improved following intervention. Other measures including the percentage below target SPL and acoustic correlates of vocal quality did not change following therapy. The aim of the speech pathology treatment used in the current study was to teach cough suppression, and although it addressed some aspects of laryngeal function it was not a formal voice therapy program. It would be interesting to see whether these latter measures will improve more following a therapy program that directly targets phonation, such as resonant voice therapy or vocal function exercises. The high prevalence of abnormal results in this study suggests that the speech pathology assessment of the patient with chronic cough should incorporate assessment of phonation regardless of whether the patient presents with voice symptoms. Whether this needs to include a formal assessment of vocal loading is unclear. Formal vocal loading tests are time-consuming and require specialist equipment and therefore may not be feasible to conduct routinely in the clinical setting. However, they may be useful for specific patients with chronic cough, including those who are professional voice users. In addition to formal objective measures, the vocal loading test enables observation of changes in constriction, laryngeal muscle tension, and respiratory patterns. These features are a frequent precursor to cough, can often change throughout the course of the test, and may be indicative of the difficulties patients are experiencing in daily life. Anecdotally, most participants reported that the test was a
significant challenge and the most difficult portion of their participation in the entire study (Figure 4).13 Several methodological issues and limitations need to be considered when interpreting these results. The inclusion of a nontreatment control group would strengthen the interpretation of the results and are recommended in further studies of vocal loading in this population. Several types of vocal loading tests are available, and the choice of test used in the current study may have an influence on the results. Previous studies have used reading tasks with a target intensity of between 75 and 85 dB SPL.9,10,12,20 The advantage of the lingWAVES Vocal Loading Test is that the protocol is standardized, which enables consistent administration between assessments and between patients. The length of time between the three assessments was sufficient to negate a practice or fatigue effect. Previous studies have found that changing the target intensity during the test, which also occurred in the current study, resulted in more difficulty than when the target vocal load was kept constant throughout the test.11 This study examined the effect of prolonged phonation but did not examine other phonatory issues that could provoke cough. This study used a single baseline measure. It did not employ formal auditory perceptual voice ratings immediately pre and post each vocal loading session. Further studies could examine the utility of using shorter vocal loading times. Most participants who coughed during the test (81%) coughed between 1 and 7 minutes of the test, suggesting that the duration of the test could be shortened while still yielding clinically useful results.
ARTICLE IN PRESS Anne E. Vertigan, et al
Effect of Vocal Loading Test on Cough and Phonation
9
FIGURE 4. Summary of baseline, posttreatment, and follow-up results for the outcome measures with significant improvement.
CONCLUSION Objective measurement of the phonatory stimulus in chronic cough showed that in most patients with chronic refractory cough, the performance on this test of vocal loading was impaired at baseline and most parameters improved following therapy. The study demonstrates that phonation is an important trigger for cough and is the first to quantify the duration and intensity of phonation required to trigger cough. Acknowledgments The authors would like to acknowledge Carol Bishop from Multimedia Speech Pathology for training in the use of the Vocal Loading Test and Mary Aldrich for assistance with the figures. REFERENCES 1. Schappert S, Rechtsteiner E. Ambulatory medical care utilization estimates for 2007. In Vital and Health Stat, N.C.f.H. Statistics, editor, 2011. 2. Song WJ, Chang YS, Faruqi S, et al. The global epidemiology of chronic cough in adults: a systematic review and meta-analysis. Eur Respir J. 2015;45:1479–1481. 3. Haque R, Usmani O, Barnes P. Chronic idiopathic cough: a discrete clinical entity? Chest. 2005;127:1710–1713. 4. Vertigan A, Gibson P. Chronic refractory cough as a laryngeal sensory neuropathy: evidence from a reinterpretation of cough triggers. J Voice. 2011;25:596–601. 5. Vertigan AE, Theodoros DG, Gibson PG, et al. Voice and upper airway symptoms in people with chronic cough and paradoxical vocal fold movement. J Voice. 2007;21:361–383.
6. Matsumoto H, Tabuena RP, Niimi A, et al. Cough triggers and their pathophysiology in patients with prolonged or chronic cough. Allergol Int. 2012;61:132. 7. Vertigan AE, Theodoros DG, Winkworth AL, et al. Perceptual voice characteristics in chronic cough and paradoxical vocal fold movement. Folia Phoniatr Logop. 2007;59:256–267. 8. WEVOSYS. lingWAVES Vocal Load Test, 2014. Accessed June 9, 2016. https://wevosys.com/products/lingwaves/lingwaves_vocal_loading_test .html. 9. Hanschmann H, Gaipl C, Berger R. Preliminary results of a computer-assisted vocal load test with 10-min test duration. Eur Arch Otorhinolaryngol. 2011;268:309–313. 10. Whitling S, Rydell R, Lyberg Ahlander V. Design of a clinical vocal loading test with long-time measurement of voice. J Voice. 2015;29:10. 11. Echternach M, Richter B, Traser L, et al. Change of vocal capacity due to different vocal loading tests. Laryngorhinootologie. 2013;92:34–40. 12. Echternach M, Nusseck M, Dippold S, et al. Fundamental frequency, sound pressure level and vocal dose of a vocal loading test in comparison to a real teaching situation. Eur Arch Otorhinolaryngol. 2014;271:3263–3268. 13. Vertigan AE, Kapela SL, Ryan NM, et al. Pregabalin and speech pathology combination therapy for refractory chronic cough: a randomized controlled trial. Chest. 2016;149:639–648. 14. Irwin R, Boulet LP, Cloutier MM, et al. Managing cough as a defense mechanism and as a symptom: a consensus report for the American College of Chest Physicians. Chest. 1998;114:133S–181S. 15. Jilek C, Marienhagen J, Hacki T. Vocal stability in functional dysphonic versus healthy voices at different times of voice loading. J Voice. 2004;18:443–453. 16. Vertigan A, Theodoros DG, Winkworth A, et al. Chronic cough: a tutorial for speech language pathologists. J Med Speech Lang Pathol. 2007;15:189– 206.
ARTICLE IN PRESS 10 17. Vertigan A, Theodoros DG, Gibson PG, et al. Efficacy of speech pathology management for chronic cough: a randomised placebo controlled trial of treatment efficacy. Thorax. 2006;61:1065–1069. 18. Dicpinigaitis P. Short- and long-term reproducibility of capsaicin cough challenge testing. Pulm Pharmacol Ther. 2003;16:61–65.
Journal of Voice, Vol. ■■, No. ■■, 2017 19. Bessac B, Jordt S-E. Breathtaking TRP channels: trpa1 and TRPV1 in airway chemosensation and reflex control. Physiology (Bethesda). 2008;23:360–370. 20. Niebudek-Bogusz E, Kotyło P, Politan´ski P, et al. Acoustic analysis with vocal loading test in occupational voice disorders: outcomes before and after therapy. Int J Occup Med Environ Health. 2008;21:301–308.