Tracking occupational voice state with a visual analog scale: voice quality, vocal fatigue and effort

Abstract

Background:

Due to elevated vocal health risk in industries such as call centers, there is a need to have accessible and quick self-report tools for voice symptoms. This study aimed to determine if the concurrent and construct validity of three visual analog scales (VASs) of voice quality and symptoms could be used as a screening tool in call center agents.

Methods:

A cross-sectional study was carried out in three call center companies. The Voice Handicap Index-10 (VHI-10) and a vocal hygiene and symptoms survey were administered to 66 call center workers. Further, acoustic parameters including harmonics-to-noise ratio (HNR), smoothed cepstral peak prominence (CPPs), L1-L0 slope, and Alpha ratio were collected. Finally, workers completed three VASs capturing self-perception of vocal effort (VAS-1), voice quality (VAS-2), and vocal fatigue (VAS-3). Linear regression models with bootstrapping evaluated the possible relationship between the three VASs measurements, self-perceived vocal symptoms, and acoustic parameters.

Results:

VAS-1 scores were associated with HNR and voice breaks, VAS-2 with voice breaks, and VAS-3 with Alpha ratio. Using the area under a receiver operating characteristic curve (AUC), the highest AUC for detecting an altered VHI-10 questionnaire score was observed for the three VASs. Also, the highest AUC for detecting altered CPPs was reached for the VAS-1.

Conclusions:

VAS as a self-report instrument of vocal symptoms is related to psychosocial voice impairment and alterations of acoustic voice parameters in call center workers. Such instruments could be easily implemented to identify voice complaints in these populations.

Short summary:

In this study, three visual analog scales were shown to have a good association with acoustic metrics and psychosocial voice impairment in call center workers, allowing them to be incorporated in occupational settings for the early detection of vocal symptoms (260)

INTRODUCTION

Many workers, such as call center agents, use their voice as a primary occupational tool of the trade both in remote work and in-house call center environments. These workers can be referred to as occupational voice users.(1–3) Occupational voice users experience up to 4 times greater risk for voice pathologies than non-occupational voice users due to a high voice use at work,(2,4) one of the most reported being vocal fatigue. (5,6) In particular, call center workers to have a 33–68% (2,7) prevalence of work-related voice disorders (WRVD). In addition, absenteeism and disability due to WRVD have been associated with high health care costs.(8)

Occupational voice problems have a multifactorial etiology due to personal factors, vocal demand, and vocal demand response making early screening and diagnosis challenging to address in this group of voice users.(6,9,10) However, the detection of voice complaints is key to making decisions related to early vocal health intervention and eventual modifications of working conditions.

Self-assessment, also known as patient-reported outcome measures (PROM), are valuable because they capture the patient’s feedback and experience (e.g., using visual analog scales (VASs)).(11,12) A validated, voice-centered PROM could predict the risk of having or the severity of impact a voice disorder an also, it would allow an early approach, timely treatment, and a better prognosis.(3,13)

Based on prior research, it is hypothesized that PROMs for voice characteristics (e.g., voice quality) are correlated with objective measures (e.g., acoustic analysis), constituting a form of concurrent validation. Finally, we hypothesize that vocal effort, vocal quality, and vocal fatigue can be quantified using the efficiency of a VAS, and such quantification relates to the risk of greater vocal disability. The aim of this study was to determine if the concurrent and construct validity of three VASs for voice symptoms could be used (separately or in a combination of scores) as a screening tool for voice problems in call center workers.

MATERIALS AND METHODS

Participants

This cross-sectional analytical study was implemented at three different call center locations in Santiago, Chile. Participants were recruited through a non-probability sampling method with the following inclusion criteria: (1) employed by the company for more than two months, (2) worked daytime shifts to avoid differences in vocal workload and physical fatigue with other workers, and (3) no previous diagnosis (self-reported) of voice disorders due to non-occupational causes. The human subject participation protocol was approved by the Institutional Review Board of the Pontificia Universidad Católica de Chile. Consent to participate and all data collection procedures were conducted at the participant’s place of work during working hours in a single session.

Visual analog scales (VAS)

Participants were asked to rate their voice with 3 self-perception VAS instruments (Appendix 1): vocal effort (VAS-1, 200mm bipolar scale), voice quality (VAS-2, scale 200mm bipolar), and vocal fatigue (VAS-3, 100mm unipolar scale). The information obtained through each VAS was based on the literature related to vocal quality symptoms in call center agents.(5,13) Participants were briefed on the meaning of each concept and how to perform the self-assessment task. The scoring system and size were based on a previous study in which three VASs were tested to measure vocal endurance during a working day(14), and the conceptualization of the constructs taken from Hunter’s framework.(9) Concerning the VAS size, the VAS-1 and VAS-2 are bipolar scales with open-ended questions, expecting greater response variability. Therefore, the two scales with a greater extension could detect the possible greater variability. On the other hand, the VAS-3, being a unipolar response, smaller size was expected to be sufficient to detect vocal fatigue response variability. Finally, having instruments of different sizes could avoid extreme response bias. This type of bias could generate outliers that could alter the reliability of the instruments. Liu & Zumbo (15) observed that outliers artificially inflate the estimates of coefficient alpha obtained from visual analog scales. Also, extreme response bias can lead to biased estimates of the construct validity parameters such as factor loadings, residual variance, and factor means and variance. (16)

Acoustic voice assessment

To obtain acoustic voice parameters, participants were seated and asked to perform 2 phonatory tasks: (1) a steady vowel /a:/ at comfortable loudness and pitch, and (2) a connected speech (a phonetically balanced passage). Voice samples were recorded in a quiet, acoustically isolated, low-reverberation room (30 dBA). Recordings were made to an SD storage card (44kHz, 16-bit, wav) using a portable digital recorder (Tascam, DR40) that was placed at 45º and 20 cm from the participant’s mouth. Voice acoustic metrics were estimated from the samples using the Praat v.6.1.16 software.(17)

The following parameters were obtained from the steady vowel: Harmonic to Noise Ratio (HNR), Smoothed Cepstral Peak Prominence (CPPs-vowel), jitter, and shimmer. In addition, the Praat Voice Report from the steady vowel was used to obtain the number of voice breaks, which was defined as the number of distances between consecutive pulses that are longer.(18) From connected speech recordings, the parameters collected were L1-L0 slope, Alpha ratio, and Smoothed Cepstral Peak Prominence (CPPs-speech).

Self-perception of voice and vocal behaviors

Participants were asked to complete the Spanish version of the Vocal Handicap Index (VHI-10).(19,20) In addition, a second voice survey was utilized, the Vocal Hygiene Behaviors and Symptoms Questionnaire (VHS). The VHS is a reliable 21-item tool used to explore the vocal hygiene habits and vocal symptoms of call center operators.(5) Section I consists of 6 questions capturing vocal hygiene behaviors and environmental factors. Section II consists of 15 questions capturing voice symptoms. Responses to the questions were on a Likert-type form (0 to 4 points) with four response options (i.e., never, almost never, sometimes, and always). All scales were applied in their original language and presented in English for publication since VHI-10 was originally created in English.(21)

Statistical analyses

Concurrent validity was evaluated by determining the association between the three VASs scores and acoustic voice parameters, voice handicap (VHI-10), vocal hygiene behaviors, and self-perceived voice symptoms (VHS). Univariate linear regression models were built to assess the association between each of the three VASs and the dependent variables of interest: acoustic voice parameters, voice handicap, vocal hygiene behaviors, and self-perceived voice symptoms. If the outcome variable was non-normal distributed, the standard error was estimated through bootstrapping (10,000 replications). The 95% confidence intervals were calculated using the bias-corrected and accelerated method. Also, the association between each of the self-perceived voice symptoms and VAS scores was explored through ordinal regression models, estimating odds ratios (OR).

When analyzing the relationship between the VAS scores and the voice handicap, the VHI-10 questionnaire’s overall score was dichotomized. We considered a cut-off of 11 points based on previous studies in Spanish-speaking population and people from other countries. (22–25) Since Ng et al. (26) and Behlau et al. (27) reported an alternative 7.5 points cut-off when validating the Cantonese and Brazilian-Portuguese versions of the VHI-10, a sensitivity analysis (28) was performed using this alternative cut-off. Logistic regression models were built to assess the association between scores over 11 and 7.5 points, and each VAS as independent variables. The accuracy of each VAS for determining voice handicap screening was compared using receiver operator characteristic (ROC) analysis. The area under the receiver operating characteristics curve (AUC) was estimated to perform the above comparison.

The construct validity of the three VASs was obtained using exploratory factor analysis. A principal component factor analysis (PCFA) was used to perform the factor analysis of the correlation matrix. To determine the number of factors identified, we used a scree plot and Kaiseŕs criterion of eigenvalue >1.0. Finally, the internal consistency of each of the items was calculated through the Cronbach Alpha Reliability Coefficient (α). Also, correlations between each VAS and the overall score obtained from the sum of the three VASs were estimated.

RESULTS

The sample comprised 66 call center workers who fully participated in the data collection. The mean age of participants was 39.7 years, with 61 women and 5 men where sex was considered a biological factor. Fifty-four participants (70.89%) reported working 44 hours per week (average of 42.79 hours per week, SD 6.19). Table 1 shows the socio-demographic characteristics of the sample.

Table 1.

Socio-demographic and work-related characteristics reported by call center operators (N=66).

	Mean (SD)	Median (25th-75th)
Age (years)	39.71 (10.44)	38.5 (33–47)
Number of working hours	7.46 (6.30)	5 (3–9)
Months of working experience	42.79 (6.19)	45 (45–45)

SD: standard deviation.

Acoustic parameters, voice handicap, vocal hygiene, and self-perceived voice symptoms

The mean values of the acoustic voice parameters assessed by HNR, CPPs-vowel/speech, L1-L0, Alpha ratio, jitter, and shimmer were within the normal range (Table 2). However, the 75^th of HNR and shimmer parameters were above the normal threshold. Also, the medians of the CPPs-vowel and CPPs-speech parameters were below the cut-off point considered in the current study (14.45 and 9.33, respectively). While the mean VHI-10 overall score was in the normal range, the 75^th of the overall score was over the cut-off point.

Table 2.

Descriptive statistics of the acoustic voice parameters (steady vowel and speech) as well as from questionnaires/indices (Voice Handicap Index-10 and VHS Questionnaire: vocal hygiene and self-perceived voice symptoms) (N=66).

		Mean (SD)	Median (25th-75th)
Steady Vowel Parameters	HNR	17.76 (3.92)	18.41 (14.66–20.56)
	CPPs-vowel	13.97 (2.05)	14.35 (12.46–15.53)
	Jitter	0.49 (0.35)	0.41 (0.30–0.55)
	Shimmer	0.48 (0.28)	0.39 (0.30–0.64)
	Voice breaks	0.12 (0.48)	0 (0–0)
Connected Speech Parameters	Alpha ratio	−18.02 (2.14)	−17.45 (−19.50 to −16.56)
	L1-L0	−2.33 (2.31)	−2.04 (−4.04 to −0.76)
	CPPs-speech	7.90(0.81)	7.81(7.39–8.37)
Voice Handicap Index-10	VHI-10 overall score	8.65 (5.86)	7 (4–12)
VHS Questionnaire	Vocal hygiene behaviors	8.78 (2.62)	9 (7–11)
	Self-perceived symptoms	20.89 (8.10)	21 (15–27)

VHS: Vocal hygiene behaviors and self-perceived symptoms questionnaire.SD: standard deviation.

The maximum possible score in the vocal hygiene behaviors section of the VHS questionnaire is 18 points (higher score reflects harmful behaviors); in this sample, the mean score was 8.78 (SD 2.62). The 2 most common detrimental vocal hygiene behaviors were drinking more than 3 cups of coffee, tea, or soft drink (46.97%) and speaking in noisy environments (30.30%). The maximum possible score for the self-perceived symptoms of the VHS questionnaire is 45 points; in this sample, the mean score reached 20.89 (8.10) points. The 2 most common symptoms were needed to throat clear when speaking (61.0%), felt dry mouth (46.9%) and felt phlegm in the throat when speaking (42.2%).

Visual analog scales

The vocal effort score (VAS-1) reached a median of 10.01 cm (25^th-75^th: 10.0–11.7). A series of outliers can be observed in the box plots depicted in Figure 1 (black circles). The median for self-perception voice quality (VAS-2) was 10 cm. (25^th-75^th: 7.2–12.0), while vocal fatigue (VAS-3) was 5.1 cm. (25^th-75^th: 4.5– 6.2).

Fig. 1.

Box plots for the visual analog scales (VAS) assessing vocal effort, voice quality, and vocal fatigue.

Concurrent validity with acoustic voice parameters

There was a significant association between the VAS score for vocal effort (VAS-1) and the acoustics parameters such as HNR, CPPs-vowel, CPPs-speech, and voice breaks (Table 3). Specifically, an increase in perceived vocal effort significantly decreased the HNR (β=−0.34; 95%CI −0.60 to −0.08) and CPPs-vowel (β=−0.19 95%CI −0.30 to −0.06) as well as CPPs-speech (β=−0.07; 95%CI −0.12 to −0.01) parameters. Additionally, a higher number of voice breaks was associated with a higher vocal effort (β=0.05; 95%CI 0.01–0.11). Further, worse self-perceived voice quality was associated with more voice breaks (β=0.03; 95%CI 0.001–0.09). Finally, the higher the vocal fatigue self-perception, the higher the Alpha ratio (β=−0.24; 95%CI −0.48 to −0.01).

Table 3.

Univariate linear regression analyses for the VAS (dependent variable) and the independent variables of acoustic voice parameters (steady vowel and speech) as well as from questionnaires/indices (Voice Handicap Index-10 and VHS Questionnaire: vocal hygiene and self-perceived voice symptoms), voice handicap, vocal hygiene behaviors, and self-perceived voice symptoms (N=66).^a,b

	Vocal effort VAS-1 (95% CI)	p-value	Self-perception voice quality VAS-2 (95% CI)	p-value	Vocal fatigue VAS-3 (95% CI)	p-value
Steady Vowel Parameters
HNR	−0.34 (−0.60 to −0.08)	<0.05	−0.23 (−0.49–0.03)	0.087	−0.17 (−0.58–0.24)	0.410
CPPs-vowel	−0.19 (−0.30 to −0.06)	<0.01	−0.09 (−0.23–0.04)	0.173	−0.10 (−0.33–0.13)	0.397
Jittera	0.01 (−0.01–0.02)	0.381	0.01 (−0.01–0.02)	0.113	−0.01 (−0.03–0.02)	0.641
Shimmera	0.01 (−0.03–0.03)	0.287	0.01 (−0.03–0.02)	0.672	0.01 (−0.01–0.04)	0.209
Voice breaksa	0.05 (0.01–0.11)	<0.05	0.03 (0.001–0.09)	<0.05	0.01 (−0.03–0.04)	0.644
Connected Speech Parameters
Alpha ratio	−0.12 (−0.25–0.001)	0.051	−0.11 (−0.28–0.05)	0.172	−0.24 (−0.48 to −0.01)	<0.05
L1-L0	0.05 (−0.11–0.21)	0.533	0.01 (−0.12–0.15)	0.844	−0.22 (−0.50–0.07)	0.131
CPPs-speech	−0.07 (−0.12 to −0.01)	<0.05	−0.05 (−0.01–0.01)	0.078	−0.07 (−0.15–0.01)	0.079
Voice Handicap Index-10
VHI-10 overall scorea	0.61 (0.18–1.05)	<0.01	0.64 (0.29–0.99)	<0.001	1.32 (0.55–2.06)	<0.001
VHS Questionnaire
Vocal hygiene behaviors	0.20 (0.04–0.36)	<0.05	0.20 (0.03–0.38)	<0.05	0.29 (−0.03–0.60)	0.079
Self-perceived symptoms	0.62 (0.05–1.20)	<0.05	0.74 (0.32–1.17)	<0.01	2.34 (1.60–3.08)	<0.001

^aThe standard error of linear models was estimated with bootstrapping (10,000 replications).

^bVariables significantly associated with the VAS in bold.

Concurrent validity with the voice handicap

The VHI-10 overall score was positively associated with the VAS assessing vocal effort, self-perception of voice quality, and vocal fatigue (Table 3). Thus, a greater vocal handicap score led to a greater self-perception of vocal effort (β=0.61; 95%CI 0.18–1.05), worsened voice quality (β=0.64; 95%CI 0.29–0.99) and increased vocal fatigue (β=1.32; 95%CI 0.55–2.06).

Concurrent validity with the hygiene behaviors and self-perceived voice symptoms

Vocal hygiene behaviors (VHS questionnaire) were associated with vocal effort (β=0.20; 95%CI 0.04–0.36) and self-perception of voice quality (β=0.20; 95%CI 0.03–0.38). Also, self-perceived symptoms were associated with all three VASs scores (Table 3). The following symptoms were related to vocal effort: feel their voice is weak (OR=1.22; 95%CI 1.03–1.43); need to make an effort to speak (OR=1.19; 95%CI 1.05–1.35); and feel their voice is unsteady and/or faltering (OR=1.15; 95%CI 1.05–1.27) (Table 4). The same symptoms were associated with the VAS of voice quality, with the addition of getting tired (OR=1.18; 95%CI 1.05–1.33) and feeling their throat hurt when speaking (OR=1.13; 95%CI 1.01–1.26). A greater number of symptoms were associated with the VAS of vocal fatigue. A strong relationship (OR) was observed for fatigue-related symptoms: feel their voice tenses up when speaking (OR=1.85; 95%CI 1.33–2.59), feel their voice as weak (OR=1.67; 95%CI 1.20–2.32), and need to make an effort to speak (OR=1.64; 95%CI 1.20–2.24) (Table 4).

Table 4.

Univariate ordinal regression analyses for the VAS (dependent variable) and VHS Questionnaire: self-perceived voice symptoms (N=66)^a.

	Vocal effort VAS-1 OR (95% CI)	p-value	Self-perception voice quality VAS-2 OR (95% CI)	p-value	Vocal fatigue VAS-3 OR (95% CI)	p-value
Do you feel your voice tenses up when speaking?	1.11 (0.99–1.26)	0.080	1.09 (0.94–1.27)	0.257	1.85 (1.33–2.59)	<0.001
Do you feel your voice gets hoarse when speaking?	1.08 (0.95–1.23)	0.249	1.07 (0.93–1.24)	0.366	1.59 (1.22–2.07)	<0.01
Do you feel of a foreign body in the throat?	0.99 (0.88–1.12)	0.916	1.03 (0.90–1.17)	0.663	1.27 (1.00–1.63)	0.052
Do you feel of phlegm in the throat when speaking?	1.03 (0.87–1.21)	0.721	1.04 (0.89–1.21)	0.659	1.44 (1.02–2.05)	<0.05
Do you feel your voice as weak (do you feel your voice varnishes?	1.22 (1.03–1.43)	<0.05	1.37 (1.18–1.59)	<0.001	1.67 (1.20–2.32)	<0.01
Do you need to make an effort to speak?	1.19 (1.05–1.35)	<0.01	1.26 (1.09–1.46)	<0.01	1.64 (1.20–2.24)	<0.01
Do you feel your voice as unsteady and/or faltering?	1.15 (1.05–1.27)	<0.01	1.24 (1.10–1.40)	<0.001	1.54 (1.14–2.08)	<0.01
Do you get tired when speaking?	1.13 (0.99–1.29)	0.079	1.18 (1.05–1.33)	<0.01	1.75 (1.28–2.39)	<0.001
Have you lost your voice or suffered from dysphonia?	1.04 (0.87–1.24)	0.650	1.01 (0.90–1.15)	0.840	1.63 (1.21–2.19)	<0.01
Do you feel that your voice changes by the end of the day?	1.07 (0.95–1.21)	0.244	1.01 (0.95–1.21)	0.251	1.40 (1.08–1.82)	<0.05
Do you feel a lack of air when speaking?	0.98 (0.86–1.13)	0.808	1.04 (0.89–1.22)	0.627	1.32 (0.99–1.78)	0.057
Do you need to throat clear when speaking?	1.10 (0.92–1.32)	0.298	1.08 (0.93–1.24)	0.316	1.57 (1.28–1.93)	<0.001
Do you suffer from itchy throat?	1.10 (0.96–1.28)	0.156	1.07 (0.95–1.21)	0.274	1.28 (1.06–1.56)	<0.05
Does your throat hurt when speaking?	1.11 (0.97–1.27)	0.133	1.13 (1.01–1.26)	<0.05	1.38 (1.07–1.79)	<0.05
Do you feel of dry mouth?	1.04 (0.88–1.23)	0.628	1.14 (0.96–1.34)	0.129	1.28 (0.93–1.74)	0.140

^aVariables significantly associated with the VAS in bold.

Concurrent validity with the Voice Handicap Index

The highest AUC (0.73) for detecting an altered VHI-10 questionnaire score was observed for the overall VAS (Table 5). Also, the highest AUC (0.70) for detecting an altered CPPs-vowel was reached for the vocal effort VAS. The selected cut-off points to reflect a balance between sensitivity and specificity were 26.2 cm. and 10.1 cm., respectively. Since no AUC for CPPs-speech reached a value greater than or equal to 0.70, the cut-off point was not estimated. In general, an AUC of 0.7 to 0.8 is considered acceptable.(29) Changing the cut-off from 11 points to 7.5 did not change our previous estimates significantly. Interestingly, with a 7.5 points cut-off, the AUC obtained by VAS-3 increased (AUC=0.69), as did the sensitivity (77.4) and specificity (67.7) of the Overall VAS score (Supplementary Table 1 in Appendix 2).

Table 5.

Concurrent validity of the VAS measurements for detecting both altered Voice Handicap Index-10 score and CPPs-vowel parameter among call center operators (N=66).

	Vocal effort VAS-1	Self-perception voice quality VAS-2	Vocal fatigue VAS-3	Overall VAS
VHI-10
AUC	0.68	0.72	0.66	0.73
Cut-off	10.5	10.1	5.1	26.2
Sensitivity	63.16	73.68	63.16	73.68
Specificity	82.61	63.04	54.35	63.04
CPPs
AUC	0.70	0.61	0.57	0.66
Cut-off	10.1	9.9	5.0	25.0
Sensitivity	70.59	70.59	61.76	67.65
Specificity	64.52	45.16	45.16	51.61

Construct validity

The principal component factor analysis suggested the existence of single factor, which explained 60.3% variance of the overall VAS score. Thus, the factor analysis showed that the three VASs are a unidimensional instrument, and their scores can be combined to obtain an overall score.

Reliability

The three VASs combined obtained a Cronbach’s Alpha of 0.70. The Spearman’s rank correlation coefficient between each VAS and the global score -obtained from the sum of the three VASs- fluctuated between 0.86 (p<0.001) and 0.45 (p<0.001).

DISCUSSION

The aim of this research was to assess the concurrent and construct validity of three VASs to assess self-voice symptoms in call center workers. In the current study, the VASs that assess self-perception of vocal effort, voice quality, and vocal fatigue were associated with VHI-10 scores, vocal hygiene behaviors, and acoustic metrics in a group of call center workers. Additionally, the combined use of the three VASs scores highly correlates with the VHI-10 and the CPPs scores.

It is also shown in this study that using a VAS for voice self-assessment can be useful in occupational settings, thus assisting in the early identification of occupational voice problems.(30) Nevertheless, the use of such tools is dependent on the instrument’s validity and ease of use. This study has shown that the use of an instrument like the VAS to quantify components like presented here could meet those criteria.

Concurrent validity

The VAS scores for the perception of effort and voice quality were associated with vocal hygiene habits. While the three self-perceived quantities are unique (vocal effort, voice quality, vocal fatigue), they are not mutually exclusive in the perception of a voice. Also, it is not surprising that the three individual VAS scores all had a concurrent relationship with VHI-10; a greater VHI score was associated with a greater self-perception of vocal effort, yet a worsening perceived voice quality and an increased sense of vocal fatigue. Therefore, it is conceivable that the ease and applicability of the three individual VASs could be used in occupational settings as a replacement or complement to other vocal self-perception scales, which are more cognitively challenging and more time-consuming. This was also observed by Naunheim et al.(13), where they found a very high correlation between a 4-item VAS and the VHI-10, which is further evidence to justify the application of a more straightforward assessment that offers similar advantages to longer and more cumbersome ones. Also, the use of VASs has been tested to assess vocal training outcomes in occupational voice users and has shown to be a sensitive instrument to changes.(14)

These singularities of the three VASs, unlike other studies that used the VHI as a PROM(31,32) hold for the association with the acoustic metrics. For example, VAS scores for the perception of vocal effort were associated with CPPs-vowel, CPPs-speech, HNR, and the number of voice breaks. It has previously been reported that HNR is a significant predictor of vocal effort by speakers,(33) and that CPP increases during effortful speech productions.(34) Further, self-perceived voice quality is related only to the number of voice breaks, and self-perceived vocal fatigue pertaining only to the Alpha ratio, which aligns with other studies that have evaluated the acoustic changes of the voice and vocal fatigue during a working day.(35) As a secondary finding, we can also highlight that the sustained vocal task was more sensitive to the correlations between acoustic quality of the voice and vocal self-perception, which brings once again the need to evaluate further the form and usefulness of the type of vocal tasks for the evaluation and self-evaluation of the voice. While the other acoustic measures, such as jitter, shimmer, and L1-L0, showed no relationship with any of the VASs, possibly because these scales are either a global perception of a construct that is not related to these types of acoustic metrics from steady vowels or the metrics are not related to the self-perceived components queried.

Construct validity

The construct validity of the VAS was obtained using exploratory factor analysis, showing that the three VASs are a unidimensional instrument. Thus, the instrument measures only 1 characteristic, and the amount of that attribute is the only factor influencing the probability of responding to an instrument item.(36) The single factor identified accounts for 74% of the responses’ variability in the VAS. A large amount of variance explained means that other factors not captured by the VASs might not significantly influence subjects’ voice self-perception.

All VASs were sensitive to the three individual concepts (i.e., fatigue, effort, voice quality). Since the three are different concepts, each may capture a different component of a person’s self-perception of voice. Therefore, it is not surprising that combining scores from the three VASs are associated with altered VHI-10 score and CPPs-vowel. Both measures relate to voice quality, but from different perspectives, the VHI relates to psychosocial effects of voice disorders,(37) and the CPPs to the overall level of dysphonia and breathiness.(38) We could then obtain information from different perspectives to focus on the same phenomenon, voice quality. Nevertheless, we must be careful with these interpretations, using these data only as a screening measure because CPPs have been shown to have only a low to moderate relationship with VHI.(39) Hence, while pointing towards the same goal (i.e., voice assessment), both measures are not equivalent; the quality of the subjects’ voice may not be related to its impact at the communicative level or to the voice diagnosis made by clinicians.(40,41)

Also, the three VASs scores were related to the VHI-10 and an altered CPPs-vowel value. Specifically, a high average of the three VASs could indicate that a worker could perceive psychosocial alterations (VHI-10 score) derived from underlying vocal complaints. Moreover, vocal effort VAS was correlated with general voice quality (CPPs).

Limitations & future studies

One of the main limitations of this study is that the number of participants selected as a convenience sample may impact its translation to a larger population. Also, while the current study focused on one type of call center, it is reasonable to assume that similar job requirements would be observed in other call center workers. Another issue is that the predictive validity is unknown, as only one sample was taken for each participant. Future studies could test the predictive character of this tool, specifically tracking change over time.

CONCLUSION

The results of this study indicate an association between more significant vocal disability, vocal hygiene, vocal acoustic metrics, and the self-perception of vocal effort, voice quality, and/or vocal fatigue. These three VASs have construct and concurrent validity, which allows their potential use as a screening tool for quickly and easily screening of voice complaints.