Keywords
vocal symptoms - COVID-19 - acoustic measures - perceptual measures
Introduction
The year 2020 marked the emergence of a global pandemic, COVID-19, affecting people's
physical, social, mental, emotional, and professional lives. COVID-19, an infectious
disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)
virus, results in the most common symptom of upper respiratory tract infection (URTI)
in most patients.[1] Most symptoms of COVID-19 are related to pathological changes in the upper and lower
respiratory systems, which predict the coexistence of voice-related difficulties in
people affected by COVID-19.[1]
[2]
The effect of COVID-19 on the respiratory system has both direct and indirect impacts
on the voice of the infected individuals. The pulmonary system is reported to be the
most affected in most cases of COVID-19 globally.[1] Patients with COVID-19 experience breathlessness as a major symptom, and it is closely
linked to reduced pulmonary vital capacity. Dyspnea can lead to lower breath support
for voice production or sustaining speech. Phlegm and dry throats were the most common
signs and symptoms reported during COVID-19 infection.[1] Another vocal sign in COVID-19 patients is dysphonia, which may also be one of the
first and most persistent signs of the onset of COVID-19. The laryngeal system may
manifest edema, erythema, and congestion during the inflammation. The discovery of
high expression of the COVID-19 receptor angiotensin-converting enzyme 2 on the vocal
folds is also linked to dysphonia in these cases.[3]
[4]
[5] There is also evidence of variations in the vocal parameters of cepstral peak prominence
(CPP), harmonics-to-noise ratio (HNR), and standard deviation F0 (F0SD) during acoustic
analysis.[6] A recent study explored the perceptual phonatory characteristics (auditory perceptual
evaluation and maximum phonation duration [MPD]) in 364 COVID-19-recovered participants
and reported that phonasthenia, reduced MPD, and dysphonia were reported more in COVID-19-recovered
participants than healthy controls.[7] The shreds of evidence help link the phonatory abnormalities in COVID-19 to the
involvement of the crucial subsystems of speech (phonatory and respiratory).[7] There have been explorations on the pulmonary and phonatory functions of COVID-19
after its onset in 2020. A recent report highlighted respiratory system–related difficulties
in COVID-19 patients at the end of hospitalization.[8] Currently, although the threat of COVID infection has reduced drastically, there
are still reports of new and emerging variants that may also have effects on the voice
production mechanism similar to the earliest variants of COVID-19. In this scenario,
it is essential to explore the effect of COVID-19 on the phonatory system as individuals
are resuming their professional and personal activities after the infection, which
may be crucial in predicting future voice-related problems, too. Additionally, India
was reported to witness all the variants of COVID-19, including alpha, beta, eta,
kappa, delta, and delta AY.[2] However, there is paucity of evidence documenting voice characteristics during and
after COVID-19 in the Indian subcontinent. To our knowledge, very limited studies
have reported vocal characteristics immediately after the recommended isolation period.
The present study aimed to examine the effect of COVID-19 infection on the perceptual
and acoustic correlates in the voice of individuals infected by COVID-19 within a
cluster immediately after the isolation period of 7 days. The clinical implication
of this study would be to have preliminary data evidence to create an awareness about
voice-related changes post-COVID-19 infection and plan an appropriate interdisciplinary
protocol to treat the vocal symptoms if deemed necessary. The objectives of the study
were to explore the perceptual changes, self-reported outcomes, and acoustic changes
in the voice of individuals infected by COVID-19 immediately after the isolation period
and compare it with the control group.
Method
The study followed a case-control research design wherein two groups (experimental
and control) of participants were recruited. Ethical clearance was obtained before
the commencement of the study from the ethical committee of the parent institute in
adherence with the Declaration of Helinski.[9]
Participants
Two groups of participants were recruited for the study: experimental and control.
The participants for the experimental group were from within a COVID-19 cluster from
a single quarantine zone in the university campus. Because of the COVID-19 restrictions,
only females were selected for the present study by purposive sampling due to ease
of recruitment and access to voice recording by the investigators. The cluster during
January 2022 was generalized as an omicron viral variant.[10] However, the laboratory reports did not specify the virus variant. Individuals with
a history of voice problems or respiratory infections that were not resolved at least
2 months before the COVID-19 infection onset were excluded from the study. The control
group consisted of age- and gender-matched individuals. The participants in the control
group did not have any vocal pathology in the past or recent upper respiratory infections
(within 6 months). Informed consent was taken from all the participants before the
study was conducted.
Procedure
The participants were recruited through purposive sampling. The participants in the
experimental group were instructed to fill out a voice-related symptom questionnaire,
following which they were subjected to a voice recording in a sound-treated room.
Voice Symptom Questionnaire
The participants were instructed to complete a voice-related symptom questionnaire
on the Google Form platform. The questionnaire probed into the general and vocal symptoms
experienced by the participants during the COVID-19 infection. The questionnaire contained
two sections: section A documented the demographic details of the participants and
section B probed into the general symptoms experienced during fever, vocal symptoms
experienced, and details of hospitalization. The Voice Handicap Index (VHI-10),[11] which has 10 questions to document the participants' self-perception of the functional,
physical, and emotional changes related to their voice, was used. It has a 5-point
rating scale ranging from 0 (never) to 4 (always). A score of greater than 11 is considered
abnormal.
Perceptual and Acoustic Evaluation
A sustained phonation task (vowel /a/) recording was taken from the participants in
the control and experimental groups. The recordings were taken in a sound-treated
room with an ambient noise level of less than 25 dB using Pentax Medical Multi-Dimensional
Voice Program (MDVP) interface and software (Kay Elemetrics Corporation, Lincoln Park,
NJ, United States). The sampling rate of 44,100 Hz was kept constant for all the recordings.
The vocal recording data from the experimental group were collected on the seventh
day after testing positive for SARS-CoV-2.
The voice recordings of phonation underwent a perceptual analysis using the GRBAS
scale,[12] and the MPD was also measured. The GRBAS scale is a 4-point clinician-rated perceptual
rating scale to rate the Grade, Roughness, Breathiness, Asthenia, and Strain in voice,
where 0 refers to normal and 3 refers to severe. Three experienced speech-language
pathologists performed the perceptual evaluation, and an average of their rating was
obtained for the GRBAS scores. Cronbach's alpha revealed a high inter-rater reliability
between the three ratings of GRBAS (α = 0.88). MPD is a clinical aerodynamic test to estimate the respiratory–phonatory
coordination. The patient is instructed to take a deep breath, sustain the phonation
of /a/, /i/, or /u/ vowel, and the average of three trials is compared with normatives.
The phonation voice recordings underwent acoustic analyses using MDVP and parameters
including fundamental frequency (F0), mean intensity (I0), jitter (%), shimmer (dB), noise-to-harmonics ratio (NHR), amplitude perturbation
quotient (APQ), and pitch perturbation quotient (PPQ) were extracted.
Analysis
The data from the questionnaire, VHI-10, perceptual evaluation, and acoustic parameters
were analyzed statistically in the JASP 0.16.3.[13] The data from the voice-related symptom questionnaire underwent descriptive statistical
analysis. The acoustic and perceptual analysis data were subjected to normality testing
using the Shapiro–Wilk test. The data were non-normally distributed (p < 0.05) and underwent a nonparametric Mann–Whitney U test to explore the differences in each acoustic parameter and perceptual measure
between the experimental and control groups.
Results
The data revealed that 93% of the experimental and control groups belonged to the
age group of 18 to 30 years and 7% belonged to the age group of 30 to 40 years (experiencing
no symptoms of menopause). Nearly 76% (n = 19) of the participants in the experimental group and 84% (n = 21) in the control group were from Dakshina Kannada district of Karnataka. Out
of the total, 40% (n = 10) of the participants had a graduation degree and 56% (n = 14) had completed PUC (preuniversity course). None of the participants were professional
voice users or had a history of professional voice usage (n = 0).
Voice Symptom Questionnaire and VHI
The data from the symptom questionnaire in the experimental group revealed that 88%
of the participants were symptomatic. Cough (n = 22, 84%) was the most prominent general symptom reported, and dry throat and frequent
throat clearing were the common vocal symptoms of COVID-19. The descriptive analysis
of the other variables related to the general infection course and symptoms is given
in [Table 1] and [Fig. 1].
Table 1
Factors related to the general infection course
|
Variables
|
Yes
|
No
|
|
Symptomatic
|
88% (n = 18)
|
12% (n = 7)
|
|
Hospitalization
|
32% (n = 8)
|
68% (n = 17)
|
|
Voice problems during COVID-19 infection
|
84% (n = 21)
|
16% (n = 4)
|
Fig. 1 Reported general symptoms associated with COVID-19 infection.
It was found that the maximum number of general symptoms reported by a single participant
was 11, and the minimum was 0, with a median (interquartile range [IQR]) of 5.0 (3.00).
On the other hand, the maximum number of vocal symptoms reported by a single participant
was eight and the minimum was 0, with a mean of 2.93 (± 1.41). Nearly 72% (n = 18) of the experimental group reported voice-related problems during and after
COVID-19 infection. The most and least reported vocal symptoms during COVID-19 are
detailed in [Fig. 2]. Spearman's correlation coefficient was computed to explore the linear relationship
between the general and vocal symptoms of COVID-19. The results revealed a moderate
positive correlation between the general and vocal symptoms in the experimental group
(r = 0.706).
Fig. 2 Reported voice-related symptoms associated with COVID-19 infection.
The maximum total score of the VHI-10 reported by a single participant was 18 and
the minimum was 0, with a mean of 2.53 (±7.07) in the experimental group, whereas
the mean score of VHI-10 in the control group was 0.64 (±1.18). The data were, however,
not normally distributed, with few extremely high values indicating individual variation
of the self-perception of change in voice. Although the mean scores were increased
in the experimental group, there was no statistically significant difference in VHI-10
scores between the two groups in the Mann–Whitney U test (U = 234.5, p > 0.05).
Perceptual Analysis
The mean scores of the perceptual parameters of the GRBAS scale were higher in the
experimental group for the grade, roughness, and breathiness components, indicating
perceptual deviations in voice than in the control group ([Table 2]).
Table 2
Results of descriptive statistics of perceptual analysis using GRBAS
|
Variables
|
Mean
|
SD
|
Median
|
IQR
|
|
C
|
E
|
C
|
E
|
C
|
E
|
C
|
E
|
|
G
|
0.28
|
1.00
|
0.46
|
0.28
|
0.00
|
1.00
|
1.00
|
0.00
|
|
R
|
0.24
|
0.72
|
0.44
|
0.46
|
0.00
|
1.00
|
0.00
|
1.00
|
|
B
|
0.04
|
0.76
|
0.2
|
0.52
|
0.00
|
1.00
|
0.00
|
1.00
|
|
A
|
0.00
|
0.40
|
0.00
|
0.20
|
0.00
|
0.00
|
0.00
|
0.00
|
|
S
|
0.00
|
0.04
|
0.00
|
0.20
|
0.00
|
0.00
|
0.00
|
0.00
|
Abbreviation: IQR, interquartile range; SD, standard deviation.
The results of the Mann–Whitney U test revealed statistically significant differences in the overall grade (U = 96.5, p < 0.001), breathiness (U = 162.5, p < 0.001), and roughness (U = 99.5, p < 0.001) components of the scale. However, no statistical significance was obtained
for the components of asthenia (U = 325, p = 0.65) and strain (U = 288, p = 0.46). The raincloud plots of MPD for the two groups are given in [Fig. 3]. It was clear that MPD also differed significantly between the two groups (t = 5.48, p < 0.05 for /a/; t = 3.94, p < 0.05 for /i/; t = 4.40, p < 0.05 for /u/) where it was reduced in the experimental group.
Fig. 3 Raincloud plots of maximum phonation duration (MPD) for /a/, /i/, and /u/ for the
experimental and control groups.
Acoustic Analysis
The mean, median, and standard deviation of the measured acoustic parameters for the
experimental and control groups are given in [Table 3].
Table 3
Mean and standard deviation of acoustic parameters in experimental and control groups
|
Parameters
|
Mean
|
Median
|
SD
|
IQR
|
|
C
|
E
|
C
|
E
|
C
|
E
|
C
|
E
|
|
F0
|
168.64
|
147.85
|
186.60
|
123.00
|
54.75
|
54.36
|
99.30
|
84.00
|
|
I0
|
65.09
|
55.03
|
65.20
|
52.00
|
3.52
|
7.11
|
5.00
|
6.90
|
|
Jitt
|
0.35
|
0.89
|
0.30
|
0.44
|
0.17
|
1.7
|
0.28
|
0.36
|
|
Shim
|
4.91
|
12.53
|
4.91
|
12.53
|
1.87
|
5.71
|
1.90
|
10.49
|
|
NHR
|
0.01
|
0.2
|
0.01
|
0.13
|
0.008
|
0.2
|
0.002
|
0.28
|
|
APQ
|
2.97
|
6.95
|
2.80
|
6.93
|
1.07
|
3.02
|
1.58
|
5.00
|
|
PPQ
|
0.18
|
0.51
|
0.16
|
0.25
|
0.10
|
0.99
|
0.10
|
0.20
|
Abbreviations: APQ, amplitude perturbation quotient; IQR, interquartile range; NHR,
noise-to-harmonics ratio; PPQ, pitch perturbation quotient; SD, standard deviation.
Since the data from the acoustic analysis were not normally distributed, each acoustic
parameter was compared between the experimental and control groups using the Mann–Whitney
U test in JASP 0.16.3. The results revealed statistically significant differences in
mean intensity, shimmer, NHR, APQ, and PPQ between the experimental and control groups
(U = 530, p < 0.05 for I0; U = 59, p < 0.05 for shim; U = 49.5, p < 0.05 for NHR; U = 71, p < 0.05 for APQ; and U = 196, p < 0.05 for PPQ). The effect size was large for shimmer (r
rb = 0.81) and NHR (r
rb = 0.84), medium for I0 (r
rb = 0.69) and APQ (r
rb = 0.77), and small for PPQ (r
rb = 0.37).
Discussion
COVID-19 existed as a worldwide pandemic since late 2019 and was linked to several
physiological, physical, socioemotional, and psychological complications. The present
study aimed to explore the effects of the SARS-CoV-2 viral infection on the perceptual,
acoustic, and self-reported measures of voice and compare it with a group of the noninfected
healthy control group. However, due to the small sample size and the study design,
the results may not be easily generalized to the population for the following findings.
The results of the voice symptom questionnaire in the present study showed that frequent
throat clearing was the most commonly reported vocal symptom during COVID-19 within
the cluster, followed by dryness in the throat (56.7%) and difficulty in raising loudness
level (36.7%). These symptoms align with other studies on vocal symptoms in COVID-19
cases.[6] The experimental group showed a significant moderate positive correlation between
general COVID-19 symptoms and voice-related symptoms. This may suggest that as the
severity of COVID-19 infection in an individual's body increases, the number of vocal
symptoms also increases. The most commonly reported vocal symptom in the present study
was frequent throat clearing and dryness in the throat, followed by difficulty raising
loudness and speaking for long hours. Other studies documenting the symptoms of COVID-19
in India have also reported a similar hierarchy of voice-related symptoms.[2] Since the study analyzed the vocal characteristics immediately after the isolation
period, the reminiscences of the infection in the body might have contributed to the
greater vocal symptoms reported. Moreover, most of these symptoms are related to URTI
and lower respiratory tract infections (LRTIs), which were reported as the leading
cause of voice-related changes in COVID-19 cases.[14] The mean VHI scores were higher in the experimental group than in the control group;
the change in voice was not handicapping for most participants in the experimental
group. This may be attributed to the fact that the lockdown period had limited the
extent of vocal demands and the resulting vocal loading.
The higher mean scores and statistically significant differences in overall grade,
breathiness, and roughness in the experimental group obtained on the auditory perceptual
assessment using GRBAS aligned with previous findings, which reported significantly
increased breathiness in COVID-19 patients compared to controls.[15] The MPD also differed substantially between the experimental and control groups,
which was reduced in the experimental group. This could be delineated as an effect
of respiratory insufficiency, which was not far from expectation.[15] The results are in line with a previous study that compared acoustic, perceptual,
and self-reported outcomes of COVID-19-recovered patients with normal,[14] where the results revealed lower MPD, increased VHI-10 scores, decreased voice-related
quality of life (V-RQOL), and a mild change in perceptual voice parameters in COVID-19-recovered
patients.[14]
The acoustic analysis results infer that mean intensity, shimmer, NHR, APQ, and PPQ
were higher in the experimental group compared to the control group. It can be interpreted
that amplitude-related measures have shown significant differences between the two
groups. This difference in amplitude measures could be due to the predominance of
vocal fold inflammation due to repeated cough episodes, respiratory distress, and
breathing difficulty in the participants.[6]
[16] An increased NHR was also noted in the experimental group, which can be linked to
the presence of an increased noise component in the spectrum.[17] The recurrent throat clearing and dry cough reported by the participants in the
present study might contribute to air leakage and incomplete vocal fold closure, increasing
spectral noise.[18] This has been well supported in the previous studies that concluded that the increased
spectral noise results in the voice being perceived to be breathy.[6]
Summary and Conclusion
The physiology of normal voice production is dependent on the efficient coordination
between the respiratory, phonatory, articulatory, and resonatory systems. In COVID-19,
due to the infection of these subsystems, especially the respiratory and phonatory
systems, there is an effect on the vocal mechanism that was evident in the cases of
the current study because of the reported perceptual symptoms, findings of perceptual
evaluations, and the deviations in the acoustic parameters of the experimental group
as compared to the control group However, no significant vocal handicap was reported
by the participants. The experimental group was found to have a significant reduction
in the MPD, indicating respiratory insufficiency in COVID-19 infection. The study
also revealed significant differences in the experimental group's mean intensity,
perturbation measures, and NHR than the control group. These early changes in voice
parameters are generally aggravated by vocal loading and may lead to secondary voice
disorders, like muscle tension dysphonia, in order to compensate for the imbalance
of the subsystems. Hence, using a multiparametric voice assessment protocol should
be considered to identify voice disorders in the patients post-COVID-19 infection
so that monitoring and managing the vocal symptoms become more effective while minimizing
the chances of vocal handicap.
Limitations
The current study could only generate baseline information on the immediate effects
of COVID-19 on voice in the Indian scenario due to the study design and sample size.
A longitudinal study design with a larger sample size will help document the long-term
vocal symptoms in this population, which will also give insight into developing the
protocol for treatment.
