Do Obesity and Weight Loss Affect Vocal Function?

Abstract

Obesity may be associated with increased tissue bulk in the laryngeal airway, neck, and chest wall, and as such may affect vocal function. Eight obese and eight nonobese adults participated in this study; the obese participants underwent bariatric surgical procedures. This mixed-design study included cross-sectional analysis for group differences and longitudinal analysis for multidimensional changes in vocal function from four assessments collected over 6 months. No significant differences were detected between groups from the preoperative assessment. Further, no changes were detected over time for acoustic parameters, maximum phonation time, laryngeal airway resistance, and airflow during a sustained vowel for either group. Only minor differences were detected for strain, pitch, and loudness perceptions of voice over time, but not between groups. Phonation threshold pressure (PTP), at comfortable and high pitches (30% and 80% of the F0 range) changed significantly over time, but not between groups. Examination of individual data revealed a trend for PTP at 30% F0 to decrease as body mass index decreased. PTP may be informative for assessing vocal function in clients who present with obesity and voice symptoms.

Full text

Do Obesity and
Weight Loss Affect Vocal
T-1 .\
,h
unctroni
Nancy Pearl
Sotomon, Ph.D., CCC-SLP,1 Leah B. Helou, M.A., CCC-SLP,1'2
Katie Dietrich-Burns, M.S.,
CCC-SLP,1
and Alexander Stojadinovic, M.D., FACSI
ABSTRACT
Obesiry malr
fs associated with increased tissue
bulk in the
laryngeal airway, neck, and chest wall, and as such may affbct vocal
function. Eight obese and eight nonobese adults
participated in this
stud,v; the obese participants
underwent bariatric surgical
procedures.
This mixed-design
study included cross-sectional
analysis for group
differences
and longitudinal analysis for multidimensional
changes in
vocal function from four assessments collected
over 6 months. No
significant
differences were detected
between
groups
from the preop-
erative
assessment. Further, no changes were detected over time for
acoustic
parameters,
maximum phonation time, Iaryngeal
airway re-
sistance,
and airflow during a sustained vowel for either group. Only
minor differences
were detected for strair.r,
pitch, and loudness
percep-
tions
of voice
over time, but not between
groups. Phonation
threshold
pressure
(PTP), at comfortable
and
high pitches
(300/o
and 800/o
of the
F0 range) changed
significantly over time, but not between
groups.
Examination
of individual data revealed
a trend for PTP at 30% F0 to
decrease
as body rnass index decreased.
PTP may be informative for
assessing
vocal
function in clients who present
witl-r
obesity and
voice
s)/mptoms.
KEYWORDS: Obesrty, voice,
perceptual,
acoustic, aerodynamic
Learning Outcomes: As a result of this activity, the reader
wjll be able
to (1)
describe
potentral
effects
of obesity
on
vocal function;
(2)
describe several methods
for assessing
vocal
function
and identify
one measure that may be
sensitive
to changes
related
to weight loss;
and {3)
explain
why studies by different
researchers
can result rn
drasticaliy drfferent findings
and discuss
potential
reasons for these drsparities.
tDe;rnrtment of Surgerv,
Walter Reed Armv Medical
Ce nter, Washington,
DC; 'Departtnent of Comntunica-
tion Science and Disorders, Universitl' of Pittsburgh,
Pittsburgli, Pennsl'lvania.
Address fbr correspondence
and reprint requests:
Nancv Pearl
Solomon, Ph.D., Arniv Audiologi, ft gps..h
Center,
Walter Reed Army Medical Center, Building 2,
Roonr
6a53D, Washington,
DC 20307
(e-n-rail:
Nancv.P.
Solomon@us. arnry.m
il).
Bridges berween Speech
Science
and the Clinic: A Tribute
to Thor.nas
J. Hi-ron; Guest Editor, Jeannette
D. Hoit,
Ph,D,, CCC-SLP.
Scrnin
Speech Lang2011l32:31--12. Copuighr r 2011
bv Thieme Medical Publishers,
Inc., 333 Seventh Avenue,
Neu,York, NY 10001, USA. Tel: +1(212)
584-4662.
DOI : hnp://dx.doi.orgl10.
1 055/s -0031 -127 197 3.
issN
0734-0478.
31

32 SEMINARS IN SPEECH AND
LANGUAGE/VOLUME
32.
NUMBER 1 2011
\-/besiry is a growing health problem in
many parts of the world and has been ear-
marked as a health issue
of national
orioriw in
the United States.l Excessive body
fat is ,rro-
ciated with multiple health problems
such as
diabetes,
heart disease, hypertension,
and
stroke.l People who struggle with obesity
may eventually opt for such surgical
interven-
tions as
gastric
bypass,
gastric sleeve
resection,
and adjustable
gastric banding. These proce-
dures aim to increase
physical sensations of
satiery by decreasing gastric
volume
and slow-
ing ingestion,
which in turn significantly
min-
imizes dietary consumption and facilitates
weight loss.
Obesiry and its relation to upper-airway
patency
is a topic ofhigh interest in the study of
obstructive sleep apnea syndrome (OSAS).
Various imaging techniques have revealed re-
duced diameter
of the airway in the regions of
the hypopharyrx, oropharyrx, and nasophar-
ynx as well as in the nasal cavities.2'3 Th"
narrowed upper airway results fiom thick lat-
eral pharyngeal
wails, medialized tonsillar
pil-
lars, and a posteriorly positioned and thick
velum''": however. it is not
clear whether these
anatomic
aberrations are due to fat deposits2's'6
or other
unspecified mechanisms. Nonetheless,
increased
bulk of the neck and throat is the
basis
for physician
recommendations for people
with OSAS to lose weight. Weight loss has
been
associated with a reduction in pharyngeal
adipose tissue as well as the number
of apneic
and hypopneic episodes.'
Given the potential
influence
of obesity on
the size and configuration of upper airway
strucrures,
it follows that other structures in-
volved
in voice production may be affected
by
body mass. Tissue bulk in the laryngeal airway
could vary with changes in overall body weight,
although supp_ort for this hypothesis is cur-
rently lacking.' Further, obesity could impact
vocal
function due to diminished
lung function;
increased
adiposity around
the ribs, abdomen,
and viscera can affect luns volume subdivi
sions.s
Rubin et ale conte.ried that large and
abrupt changes in body weight can
be respon-
sible for a variety of voice
problems and that
professional voice users should
only engage in
carefully managed and gradual weight loss
programs.
To our knowledge,
only one
previous study
has directly investigated the effects of obesiry
on vocal qualiry and function. Da Cunha et a110
studied the
voices
of 45
obe_se
adults
(body
mass
index [BMI] >35 kg/m') and 45 normal-
weight or overweight adults (BMI < 30 kg/
m') using auditory-perceptual, visual-percep-
tual, and acoustic methods. Twice aS
many of
the obese participants displayed laryngoscopic
abnormalities compared with their control
group, although about three-quarters of the
obese participants presented with normal-ap-
pearing larynges. Auditory-perceptual evalua-
tion revealed that the voices of obese
participants were more likely to be perceived
irs evidencing vocal
strangulation
("stress
in the
voice")
or vocal fry at the ends of utterances.
Hoarseness
(defined as "irregular vocal fold
vibration" and used synonymously
with rough-
zess), murmuring (defined
as
"escape
frelease]
of sonorous transglottic
airl' and used synony-
mously
with the term breathinars in that study),
and instability (defined
as
"fluctuation of the
vocal qu'ality") were also noted to be more
apparent in the obese
group.
Acoustic
analysis
revealed
significantly greater jitter, shimmer,
and noise in the obese
group. Most striking was
the difference
in maximum phonation time
(MPT)-the ability to sustain
a vowel as long
as possible was only half the duration for the
obese group than the control group (7 to 8
seconds versus l-4 seconds). Follow-up
analyses
were conducted
to exclude the potentially con-
founding effect of gastric reflux by selecting a
paired group of participants
without laryngeal
signs ofreflux; the significantly
different
vocal
characteristics
between obese and nonobese
participants remained.
The present investigation
examined
the
role of obesity and
weight loss on vocal
function
using auditory-perceptual,
acoustic,
and aero-
mechanic
indicators. Vocal function was docu-
mented
over a 6-month period
in participants
who had substantially reduced
body mass after
undergoing bariatric surgical
procedures and in
participants who were not obese and did not
reduce their body mass significantly over time.
It was anticipated that some changes in vocal
quality or function would be seen over time as a
function of weight loss. In particular,
we ex-
pected acoustic measures
of vocal instabiliry,

DO OBESITY AND WEIGHT
LOSS
AFFECT VOCAL FUNCTIONT/SOLOMON
ET AL 33
perceptual
ratings of dysphonia,
laryngeal air-
way resistance, and phonation threshold pres-
sure to be greater in obese participants as
compared with nonobese participants. Further,
these measures were expected to normalize as
body mass decreased significantly over time.
METHODS
Participants
Data from eight obese and eight nonobese
adults scheduled for surgery
that did not in-
volve the head or neck were selected for this
study from a larger prospective longitudinal
trial of voice changes after thyroidectomy. A11
individuals selected
for the current study were
scheduled
for laparoscopic
procedures
under
general anesthesia with endotracheal
intuba-
tion, either for bariatric procedures
(gastric
banding, gastric sleeve resection,
or gastric
bypass) or for other abdominal procedures
(e.g.,
cholecystectomy, hernia repair). All par-
ticipants consented as control subjects for the
larger study, and the use oftheir records for the
present purpose was subsequently approved,
in
accordance
with the rules
and
regulations
ofthe
Walter Reed Army Medical Center Institu-
tional Review Board.
According to the World Health Organi-
zation" reporting standards, BMI of 25 to 30
kg/m' is classified as overweight, and BMI
> 30 kg/m" is obese.
Obesity is further de-
lineated
as class I (BMI :30.1 to 34.9), class II
(BMI:35.0 to 40.0), and class III (BMI
> 40). For the obese group,
eight participants
were selected
whose BMI was
greater
than 35
and who were scheduled
for bariatric surgery.
For the nonobese gpup, eight participants
with BMI < 30 kglm' were selected to match
the obese
participants
as closely as
possible
for
sex
and age. Table 1 lists individual and sum-
mary
information
for these
participants, sorted
within group
for BMI.
Procedures
Data collection took place before (usually
within 1 week of) scheduled surgery, and
was
repeated
-p weeks,
3 months, and 6 months
Table 1 Demographic Data
for Participants
Grouped, and Ranked within Group,
According
to Body Mass lndex
Obese Nonobese
ID Sex Age BMI lD Sex Age BMI
AM
BF
CF
DF
EF
FF
GF
HM
Mean
SD
48 37.3 | F
64 39.2 J F
54 39.4 K F
58 42.8 L M
59 461 M M
40 489 N M
41 49.8 0 F
61 52.2 P F
53.1 44.5
9.1 5.6
26 21 .5
58 23.5
74 26.1
54 26.2
33 26.3
54 26.8
61 29.3
38 30.0
49.8 26.21
16.0 2.80
BMl, in kg/m'; obese, BMI >35; nonobese,
BMI <30,
BMl, body mass index; SD, standard deviation.
after surgery. Visual-perceptual, acoustic, audi-
tory-perceptual, and aeromechanic
measure-
ments
were made during each data collection
session.
Visual-Perceptual
Laryngeal imaging was performed using a fi-
ber-optic endoscopic laryngoscope
(Pentax
FNL-10RP3; Kay Pentax, Lincoln Park, NJ).
The fiber-optic endoscope was passed
trans-
nasally, with the participant
seated, and posi-
tioned so that its tip was approximately below
the rim of the epiglottis, allowing as clear a
view of the vocal folds as possible.
Tasks
included sustained vowels at various pitch and
loudness levels,
pitch glides, syllables, and sen-
tences.
Images
were recorded under straight
and stroboscopic
light sources. Ratings in-
volved 4-point scales for a variety
of observa-
tions, including supraglottic constriction
(anteroposterior
and mediolateral), vocal fold
mobility and symmetry,
vocal fold lengthening,
mucosal wave, and
glottic
closure pattern. Top-
ical anesthesia was sprayed in the nasal
cavities
before
this procedure as
requested to optimize
participant
comfort; otherwise,
nasal
passages
were dilated with a decongestant spray.
Acoustic
Acoustic recordir-rgs were
made in a double-
walled sound-attenuating, electrically shielded

34 SEMINARS
IN
SPEECH
AND LANGUAGE/VOLUME
32, NUMBER 1 2011
booth, which were also used for subsequent
listening of speech
samples
for perceptual
rat-
ings. Speech
was recorded
with a
head-mounted
microphone
(AKG C .120;
AKG;Vienna, Aus-
tria) positioned
4 to 5 cm from the
lips coupled
with the KayPentax Computerized
Speech Lab-
oratory
(CSL 4500,
Lincoln Park, N). Partic-
ipants sustained
the vowels /i/ and /a/ for -3
seconds
each at a comfortable pitch and loud-
ness, and read the si-r
sentences
from the Con-
sensus Auditorv-Perceptual
Evaluation
of Voice
(CAPE-V12). Using the KayPentax Voice
Range Profile @326) and its supplied
table-
mounted microphone
(Shure
SM48)
positioned
15
cm from the lips, participants provided
their
highest and lowest
voice fundamental frequen-
cies
(F0) rs well as minimum and mtximum
vocal intensities over
several trials. F0 (in Hertz
and semitones) and sound pressure
levei (in
decibels)
ranges were determined from this
assessment.
Maximum phonation
time (MPT)
was elicited over three trials. Each trial was
measnred
from the acoustic wavefbrm and the
longest of these triils was
recorded.
The middle two-second
portion extracted
from each sustained /a/ was
analyzed ircousti-
cally. Variables of interest
fbr this report rre
jitter (relative
average perturbation), shimmer
(ampiitude
perturbation
quotient),
and noise-
to-hirrmonics ratio, 'a11
of which were obtained
using the KayPenta-x Multidimensional Voice
Program
(Model 5105). Using
values obtained
from these analyses,
in addition to the max-
imum MPT value, the Dysphonia Severity
index (DSI)13
was calculated. The DSI is a
weighted algorithm that is
considered
to reflect
vocil function and, to a lesser degree,
qualiry.
Its score range, from worst to optimal voice,
is - 5 to 5. A change greater than 2.5 is
considered significant, and a change of3 yields
maximum sensitivity
and specificity.
la
Auditory-Perceptual
Speech recordings were root mean square
(RMS) normalized
via a custom program in
Matlab (R2007a)
prior to rating. Three experi-
enced
speech-language pathologists
rated the
vowel
(/i/ and /al) and sentence samples of the
CAPE-V according to its recommended
pro-
cedures. Brieflv. the CAPE-V consists of vis-
ual-analog scales
(100-mm lines, labeled with
broad categories
for mild, moderate; and severe)
for each
parameter:
severity, roughness, breath-
iness,
strain, pitch, and loudness." A custom-
ized electronic
version of the CAPE-V was
designed in Matlab (R2007a)
and utilized for
auditory-perceprual
ratings; these specialize_d
procedures
have been described
previously."
Raters listened over headphones under blinded
irnd
randomized
within-subiect
conditions.
Aeromechanic
Aeromechanic
assessment
was
conducted
with
either the KayPentax
Phonatory Aerodynamic
System
(PAS
Model6600) and its accompany-
ing software
or the Glottal Enterprises Aero-
dynamic System (Syracuse,
NY) (MS-100)
coupled with DATAQdata acquisition
hard-
ware and \ nNDAq advanced data irnalysis
software.
Only one system was used within
each
participant
for repeated assessments over
time. Both systems involve the use of a face
mask that is firmly placed
over
the participant's
mouth and nose; the pneumotachometer is
embedded in (Glottal Enterprises) or coupled
with (KayPentax)
the mask. Differential pres-
sure
transducers detect airflow and air Dressure.
Cllibration wns
conducted daily, and baseline
values were verified during each recording
session.
Participants sustained the vowel /al for -3
seconds at 300/o ofthe F0 range
(calculated
as a
percentage
of the total F0 range, in semitones,
and selected to represent a comfortlble pitch),
for the determination of average airflow during
a 2-second midportion of phonation. Second,
participants repeated the syllable /pil seven
times slowly
(1.5
syllables/s),
and in a sustained
irnd connected
(legato)
fashion
at 300/o and .rt
800/o of the ma-ximum F0 range. For determi-
nation of laryngeal airway resistance
(Rro*),
participants produced the syilables at a com-
fortable loudness
level. For determination of
phonation threshold
pressure
(PTP), partici-
pants
repeated
the syllable as quietly as
possible
without whispering.
Analysis of R1"* followed the method of
Smitheran and Hlxon.lu In bri.f, peak oral
pressure dtrring the closed
phase
of /p/ is used
as zln estimate of tracheal
(subglottal)
pressure,

DO OBESITY AND WEIGHT LOSS AFFECT VOCAL FUNCTION?/SOLOIVON ET
At 35
provided
that airflow ceases
through the nose
and lips (confirmed
by airflow values
< L0 mL/
s). Oral pressure during the midportion of the
vowel /i/ is used as an estimate of pharyngeal
pressure;
this value is subtracted
from the
average ofthe two adjacent pressure peak
values
to determine translaryngeal pressure.
Airflow
averaged across the vocalic
midportion of the
syllables
is used as translaryngeal
flow. R1"," is
calculated as translaryngeal pressure/translar-
yngeal flow (in cm HzOlVs). Nine syllable
tokens
were
averaged for each result
(rypically
three syllables from each ofthree trials).
.
Ana\sis of PTP followed the theory de-
scribed
by Titze"''o using methods described
in previous
publications.le This involves meas-
uring the pressure
peaks
during /p/ while the
participant
is phonating as quietly as
possible.
Altliough airflow is not measured during tire
rninimal-loudness task, it is monitored to con-
firm that no air leakage is present during the
ciosed
pirase
oflp/; this validates the use oforal
pressure
to estimate tracheal
pressure.
Ifneces-
sary, participants were reminded to close their
lips around
the oral-pressure
tube and/or nose
clips were secured
on the partibipant's
nares
within tl.re face mask. Nine valid pressure
peaks, whenever possible
(but no fewer than
si-r
peaks),
were selected and averaged
for eacl.r
result.
Statistical Analyses
Cross-sectional
analysis was based on preoper-
ative data; independent-sarnples I tests coln-
pared the obese and nonobese groups.
Longitudinal analysis
relied on repeated-meas-
ures
analysis of variance
(RM-ANOVA), with
group as a befween-subjects factor and session
(four) as the within-subjects factor. In the
event of violations of the assumotion of soher-
iciry for RI\{-ANOVA, signifiiance leuei, re-
lied bn the Greenhouse-Geisser
correction.
Cross-sectional analyses utilized data
from all
16 participants;
longitudinal analyses
included
data from sk obese
and six nonobese partic-
ipants who completed all four data-collection
sessions.
Because
of the exploratory
nature
of
this studl', e was set at 0.05 for all analyses
without Bonferroni adjustments
for multiple
comparisons;
results
should be
interpreted cau-
tiously because of the increased risk of qpe I
statistical error.
RESULTS
Visual-Perceptual
Ratings based on laryngeal imaging with flex-
ible fiberscopic nasal endoscopy
were com-
pleted successfully for all nonobese
participants and five ofthe eight obese
partic-
ipants during the first data
collection
session.
The three obese participants for whom all
ratings
could not be completed were affected
b1'
partially obstructed
views of the vocal folds
by supraglottic constriction or tissue redun-
danry, or the inability to assess mucosal wave
because of inadequate iliumination of the vocal
folds, vocal fold ederna, or thick mucus. One of
these
subjects
(participant
H listed in Table 1)
exhibited severe
mediolateral
and anteroposte-
rior supraglottic
constriction
(rated
3 on a scale
from 0 to 3) and ventricular
phonation
during
most sustained vowel tasks preoperatively.
Over time, as his BMI decreased, supraglottic
constriction reduced, and ventricular
phonation
was
no longer obsen'ed. One additional
obese
and three nonobese participants exhibited
marked (ratings of 2 or 3) anteroposterior
constriction during the first session,
which
persisted
during subsequent sessions wher.r lar-
yngoscopic
data were available. No other par-
ticipants were rated as having excessive
mediolateral cor.rstriction
fbr any session. Lar-
yngostroboscopic
findings from the preopera-
tive examination were within normal limits,
although four participants in each group ex-
hibited
mildly reduced mucosal wave on at least
one side
and
one participant
in each group
had
a mild deviation of the vocal
fold edge
(slight
bowing or irregularity).
Acoustic
Summan'statistics
(mean
and standard devia-
tion: Table 2) revealed
that the acoustic meas-
ures for each group at the preoperative
data-
collection
session were all
within normal
limits.
Results of inferential statistical analyses, also
listed in Table 2, failed to reveal significant
differences between groups for any acoustic

36 SEMINARS
IN SPEECH
AND LANGUAGE/VOLUME
32, NUMBER
1 2011
Table
2 Summary
Statistics
and Group
Comparisons
for Acoustic Data
from the Initial
(Preoperative)
Data-Collection
Session
Obese
(n:8) Nonobese
(n:81
Measure Mean SD
SD Mean
RAP
APO
(7o)
NHR
FO renna /QTl
SPL
range
(dB)
DSI
MPT
(s)
0.72
1,76
0, 13
31.5
50.0
0.93
1A a
059
0.64
0.02
3,46
3.25
1.85
6,86
059
1.79
0.14
34.8
50.5
0.59
21,5
0.64
0.81
0.04
4,17
6.30
1.28
3.84
- 0.439
0.076
0.672
- 1.195
1,697
0.1 99
- 0.427
0.668
0.940
0.513
0.252
0.113
0 854
0.676
BAP,
relative
ave1ge perturbation
(i.e., jitter);
APO, amplitude
perturbation quotient
(i.e.,
shimmer);
NHR, noise-to-
harmonics ratio; ST, semitone,
SPL, sound
pressure
level; DSl, Dysphonia Severity
Index; MPT, maximum
phonation
time: SD.
standard deviation.
variable. Table 3 lists RNI-ANOVA results
from the longitudinal analysis
for both groups
and four sessions.
No main effects or interac-
tions reached criterion for statistical
signifi-
cance, except ibr the interaction for noise-to-
harmonics radio.
Auditory-Perceptual
The median rating from the three listeners was
used for each auditory-perceptual parameter
from each voice sample.
Table 4 lists
summary
statistics across speakers within each
group,
as
well as inferential statistical
results for each
parameter.
The auditory-perceptual
ratings all
indicate normal voice characteristics
with no
significant diflerences between groups.
Longi-
tudinal analysis results are displayed in Table
5.
Statistically significant main effects were found
for ratings ofloudness between groups and
for
ratings
of strain and pitch across sessions. The
group-by-session
interaction was signifi cant for
strain. Table 6 provides summary statistics
for
these variables separated by group and session.
Obese
participants
were
judged to be slightly
quieter than nonobese participants, and pitch
generally decreased
slightly over the course
of
the study. Strain varied differently for the rwo
groups across time, although the pattern is not
easily decipherable. All of these average data
are
within normal limits.
Aeromechanic Measures
Table 7 lists
preoperative
summary statistics
and group comparisons, and Table 8 provides
RM-ANOVA results for airflow during the
vowel /a/ produced
at 30o/o of the voice F0
range,
and R1"* and PTP determined
from a
slow-syllable
repetition task at 300/o and 800/o
the of F0 range. Cross-sectional analysis of
the preoperative data revealed no significant
Table 3 Statistical Results for Longitudinal Analysis of Acoustic Measures
Group Session Group x Session
Variable
RAP
APO
NHR
F0 range
SPL
range
DSI
MPT
0.003
0.009
0.418
0.670
0.1 59
0.002
o.o21
0.956
0.926
0.505
0.432
0.698
0.965
0.888
0.333
2.218
1.103
1.032
2.906
o.171
0.154
0.802
0.140
0.363
0.392
0.090
0.870
0.926
2.239
1. | | I
5.307
0.1
36
0.545
0.619
0.734
0.1 04
0.1
58
0 005.
0.938
0.559
0.530
0.540
*p < 0.05.
BAP, relative
average
perturbation
(i.e., jitter);
APO, amplitude perturbation quotient
harmonics ratjo;
SPL, sound
pressure
level; DSl, Dysphonia Severity Index; MPT,
standard
deviation.
(i.e.,
shimmer); NHR, noise-to-
maximum phonation
time; SD,

DO
OBESITY AND WEIGHT
LOSS AFFECT
VOCAL FUNCTION?/SOLOIVON
ET AL 37
Table 4 Auditory-Perceptual
Ratings
(%)
and Group Comparisons
from the Preoperative
Data-
Collection
Session*
Obese
(n:8) Nonobese
(n:8)
Parameter Mean SD Mean SD
Overall Severity
Roughness
Breathlness
Strain
Pitchi
Loudnessl
5.58
3.39
4.64
3./ |
-2.54
-4.40
1.85
2.43
2.61
3.53
3.80
8.01
5.71
5.19
3.68
2.93
- 1.11
0.53
2.25
3.75
3.08
3.12
297
0.31
0.133
1.140
- 0.674
- 0.472
0.836
1.737
0.896
0.274
0.511
0.644
0.417
0,104
*Results are based on the median
of ratings
by three experienced
listeners
using the Consensus Auditory-Perceptual
Evaluation
of Voice.12
lRated
as negative if the parameter
was perceived
as too low; positive
if too high.
SD, standard deviation.
differences for any
aeromechanic variable.
Anal-
ysis of the longitudinal data for the 72 partic-
ipants who completed the study revealed a
significant main effect across
sessions for PTP,
both at 300/o and 8070 of the F0 range.
Although
the summary
statistics, listed
in Table 6, suggest
that the significant change in PTP over time
was related to the obese
group, the main effects
for group and the time-by-group interactions
were not statistically significant. To determine
whether
BMI might be influencing
the results,
the analysis
was rerun with BMI as a covariate
factor. The main effect for session was no longer
significant for PTP 300/o
lF\3,27)
:1..497,
?:0.238), or PTP 800/o
[F(3,27):0.427,
P:0.7361, indicating that BMI explained
some of the variance
in the data.
To examine the relationshio between
BMI and PTP 300/0, Fig. 1 plots individual
data for the
obese participants
across sessions.
BMI is plotted on the abscissa
in reverse
order
to indicate the effect of losing weight over
time. Only participant A increased
BMI be-
tween any two successive
data-collection
ses-
sions (3 months to 6 months). PTP 300/0
consistently
decreased
as BMI decreased
for
three participants
(A, E, and D who partici-
pated in three sessions),
inconsistently
de-
creased for three participants
(C, F, H), and
inconsistently
increased
for one participant
(B). One participant
(G) is not included
due
to attrition.
DISCUSSION
This study
examined whether vocal quality and
function differed between
obese and nonobese
participants,
an issue that has received
meager
attention
in the literarure. No statistically sig-
nificant
differences
emerged for perceived voice
quality, acoustic measures,
or aeromechanic
measures
between
obese and nonobese partic-
ipants at the beginning
of this study. This was
surprising because
BMI differed maximally
between the groups
at this time. Nonetheless,
when data were examined
longitudinally,
a few
auditory-perceptual and aeromechanic
differ-
ences emerqeo.
Table 5 Statistical Results for Longitudinal Analysis of Auditory Perceptual Measures
Group Session Group x Session
Variable
Overall
severity
Roughness
Breathiness
Strain
Pitch
LOUOneSS
0.051
0,894
0.016
0.062
1.090
5.416
0.821
0.367
0.903
0.808
0.321
0.042*
'1
.667
0,874
2.361
3.217
5.707
0.123
0.1 95
0.466
0.091
0.037
*
o.012*
0.890
2.606
.540
1.469
3.406
1.124
1.194
0.070
0.659
0.243
0.030*
0.343
0.324
-p
< 0.05

38 SEMINARS
IN SPEECH
AND LANGUAGE/VOLUME
32, NUMBER 1 2011
Table 6 Summary Statistics for Variables with Significant Main Effects from the Longitudinal
Analyses
Preoperative 2-wk
Postoperative 3-mo 6-mo
Postoperative Postoperative
Group Measure SD Mean SD Mean Mean SD
(n:6)
Nonobese
(n:6)
Strain
Pitch
LoUdneSS
PTP 30o/o
PTP BO%
Strain
Pitch
LOU0ness
PTP 30%
PTP B07o
4.42 3.75
- 3.42 4.06
-2.23 4.24
5 22 2.93
9.97 3.85
3.73 3.23
--0.31 2 29
0.60 0 36
3.87 0 90
8.95 2.98
3.1 0 2.01
- 3.12 4.61
- 1.52 3.25
4.28 1.74
7.49 2.70
6.98 4.35
0.63 4.88
0 67 0.21
3.65 0.96
7
83 2.11
3,95 2 74
-3BB 491
- 1.43 3 32
3.81 1 34
8,79 3.31
4.68 4.41
- 1
t5 3.66
0.57 0 27
3 37 0.78
7 56 2.26
3.25 2.84
- 5.85 5 38
- 2.18' 3.77
4 14 1.99
8 52 3.06
067 037
-5.42 6 81
2.10 3 39
358 109
690 172
PTP,
phonation
threshold
pressure;
SD, standard deviation.
Strain was the onlyvoice quality parameter
to change significantly across sessions, but it
did not differ between groups. The nonobese
prrrticipants received quite variable ratings fbr
strain acrr>ss
sessions,
r,vith particularly high
average ratings for the second session. The
explani.ttion for this elevated rating is not ob-
vious, althor,rgh it is possible that these non-
obese
participants were compensating for some
mild laryngeal sensation resulting from endo-
trache'rl intubation during their strrgical
proce-
dures. Why obese participants would not have
the same reaction is puzzling, but given the
wide range of variability for both groups, this
observation could have resulted fiom a small
number of pirrticipants.
Changes in perceived loudness between
groups irnd changes in pitch across sessions
were statistically significant br-rt quite small.
'fhe range of normal is considered to be be-
rween 070 and * 10% ofthe rating scalels'2o; all
ofthe average ratings for pitch and
loudness in
this study were within 6%0.
As we have dis-
cussed
previously,15
there are
problems
inher-
ent in these scales as well as in ()Llr
measurenent
solutions. A limitation specific
to the interpretation
of loudness
differences,
the only group difference
detected, is that the
listeners
rated voice samples
alter they had been
normalized for intensity. Therefbre, any
differ-
ences in perceived
loudness
could have resulted
from some attributes other than intensiry.
PTP differed significantly
across sessions
when produced at both comfbrtable and high
pitches
(300/o
and 800/o
of the F0 range, respec-
tively). In general, PTP 30%o and 800/o de-
creased over the course
of the str,rdy. This
change
was most prominent for PTP 300/o
in
the obese
group. Importantly, PTP 3070 fbr the
obese
gror-rp dtrring the first session was well
outside of the range reported previor-rsly lbr
normal voices
of young, healthy irdults from
Table 7 Summary Statistics and Group Comparisons for the Aeromechanic
Data from the
Initial (Preoperative)
Data-Collection
Session
Obese
(n:81 Nonobese
(n:8)
Measure Mean SD
SD
Flow
lal
R
a* 30%
Rr"* B070
PTP
30%
PTP
80%
191 .4
48.8
71.6
4.83
9.35
91.6
15.4
18.4
2.60
3.51
200.8
42.3
70.4
3.77
9.41
80.5
19.1
21 .3
0.42
3.87
0.218
-o.749
-0.1
06
-1 .107
0 031
0.830
0.466
0.917
0.281
0 976
PTP,
phonation
threshold
pressure;
SD, standard deviation; R1u",, laryngeal
airway resistance

DO
OBESITY
AND WEIGHT LOSS
AFFECT VOCAL
FUNCTION?/SOLOMON
ET
AL 39
Table 8 Summary
Statistics and Group Comparisons
for Aeromechanic
Measures
Group Session Session x Group
Variable
Flow lal
R1u*
30%
R1u* B0o/o
PTP
307o
PTP B0o/o
0.247
0.706
0.315
0.754
0.336
0.785
0.421
0.587
0.406
0.575
0.705
0.347
0.446
3.823
7.278
0.556
0.792
0.649
0.038*
0.001*
0.975
1.690
0.906
0.995
1.874
0.457
0.1 90
0.421
0.389
0.155
*p
<
o.o.
PTP,
phonation
threshold
pressure,
SD, standard
deviation; R6*, laryngeal
airway resistance
the first author's laboratory
(mean:3.5, stand-
ard deviation:0.S for women: mean:3.4.
standard
deviation:0.3 for men).21 Examina-
tion of the summary
data reveals that the non-
obese group produced PTP 300/o values that
were similar across
sessions and that the
variability within the group was relatively
small. In contrast,
the obese
group exhibited
elevated PTP 3070 data, marked variability,
and decreasing
PTP 3Q7o
values
over the first
three sessions.
Thus, it appears that the
amount of tracheal pressure required to
move the vocal folds into oscillation was
greater when participants
were highly obese
and decreased to some extent as they pro-
ceeded to lose weisht.
-.'-A
...r...8
--L-c
-+-D
--+--E
- l -F
--r- H
BMI(kclmz)
Figure 1 Body mass index
(BMl)
and
phonation
threshold
pressure
(PTP)
at 30ok of the
pitch
range
plotted
at each
data-collection
time
point
for seven
obese
participants
who
partictpated
in
at
least
three
sessions.
BMI is
plotted
in reverse
order to indicate werqht
loss over time.
L2
30
55
40
455055

40 SEMINARS
IN SPEECH
AND LANGUAGE/VOLUME 32. NUMBER 1 2011
PTP at 30o/o
of the F0 range decreased
as
BMI decreased, albeit somewhat inconsis-
tently, in slx of the seven obese
participants
for whom we had longltudinal data.
One par-
ticipant in particular
(H) started the study
with
the highest BMI and an extremely
high PTP.
This is the same
participant
whose vocal folds
were difficult to visualize and who used
ven-
tricular fold phonation
during portions of the
preoperative
laryngeal examination. Abnor-
mally high PTP would be easily
expected
dur-
ing ventricular
phonation
or if the bulk of the
ventricular folds compressed the vocal folds
from above, but this observation has not been
reported
previously to our knowiedge. Partic-
ipant H's PTP decreased as his BMI decreased
drastically up to 3 months after surgery and as
his ventricular phonation disappeared.
It
should also be noted that participant
H had
the highest
preoperative PTP 800/o value of all
obese
participants
(15.7
cm H2O). An expect-
edly high degree of variability
in PTP values at
. -^1q
high voice F0'" can preclude
finding group
differences.
Phonatory airflow, R1.,*, and PTP were
expected to be sensitive to obesiry because of
potential effects of tissue bulk and associated
reduction of larynp;eal a\rway
patency.
If the
laryngeal atrway
restricts airflow through the
glottis, then it is possible that resistance
to
airflow through the larynx would be greater
than normal during phonation. Further, if
mass is
added to the vocal folds or surrounding
tissues, the vocal folds might be more dilficult
to move into oscillation, especially
at a high
F0.1e Although there is no definitive
evidence
of decretrsed
pharyngeal or laryngeal tissue
bulk in this study, indirect evidence was an-
ticipated from the visual perceptual observa-
tions. Our initial impression
when examining
the larynges of obese
participants
was that it
was more difficult to obtain a nonobstructed
view of the vocal folds during indirect lar-
yngoscopy than in nonobese individuals and
that visualization
improved over
time as obese
participants
lost weight. These informal im-
pressions
were supported only by a few par-
ticipants. More telling rnight have
been formal
assessment
of the ease
of inserting
and posi-
tioning the laryngoscope. For example, visual-
ization of participant
H's larynd
was easier
and
more complete
over the course
of his weight
loss.
A much larger sample of obese adults
studied
by da Cunha et allo indicated
signifi-
cantly more
breathiness, hoarseness,
and vo-
cal instability than nonobese control
participants. Furthermore, the authors re-
ported greater
jitter, shimmer,
and noise in
the vocal
acoustic signal.
Despite
the small
size of our hypothesis-generating
conven-
ience
sample, it is surprising
that our results
did not even tend toward supporting the
group effects reported by da Cunha et a1.10
It is difficult to compare results
directly, as da
Cunha et all0 did not fully elucidate their
methods
of data collection. Their orocedures
I'or
data collection could have
solicited sub-
maximal performance on the MPT task in
particular. The present
study used
the best of
three trials, whereas
da Cunha et al do not
report the number of triais used. This ex-
planation
is supported not only by the grossly
subnormal performance by the obese
group
(7 to 8 seconds), but also by the low average
results for their nonobese participants
(14 seconds).22 Both groirps of participants
in the present study sustained vowels for
more than 20 seconds on average, and no
participant's MPT was less than-15
seconds.
Other potentially confounding factors are
that ofclinical bias and overall
health status of
study participants.
If participants
presented
with overt signs of respiratory compromise,
examiners might have
altered their instructions
so as not to distress
or overexert
the partici-
pants. Da Cunha et a110
did not describe their
participants' medical
profiles, including
pulmo-
nary
disorders such
as chronic
obstructive
pul-
monary disease or asthma. Given the poor
MPT values
for their obese
population,
it is
reasonable
to susDect
concomitant
disorders of
a pulmonary .rut.rr..t' Our sample
included
more
participants with pulmonary disorders in
the obese than nonobese
groups, yet MPT was
not affected as expected. Furthermore, our
study excluded
persons with laryngeal pathol-
ogy, whereas
da Cunha et al reported
"altered"
laryngoscopic
findings, separate from indica-
tions of gastroesophageal reflux
disease, in 12
oftheir 45 obese
participants
and 6 oftheir 45
nonobese
participants.

DO OBESITY AND WEIGHT
LOSS AFFECT
VOCAL
FUNCTION?/SOLOMON
ET AL 41
Whatever the reasons
for the differences
between groups or studies, the relevance
of
MPT as an assessment
technique
can be ques-
tioned. Indeed, a previous exploration
of lar-
yngeal versus respiratory
contributions to MPT
revealed that the
relationship is not straightfor-
ward and that this task may not reveal
the
particular functions of either subsystem.2a
Nonetheless,
MPT has become a variable
of
renewed interest,
as it is one of four variables
included in the calculation
of DSI,13
which
also
did not differ between
groups in this study.
The primary limitation of this study is its
small sample size. A larger sample
of obese and
nonobese adults
is needed to better understand
why these results
did not agree
with those from
the relatively large
study by da Cunha et a1.10
This study's longitudinal component
is a major
strength
of its design,
but certain related
lim-
itations must be considered.
Other
corroborat-
irg or confounding factors, such as
gastroesophageal reflux disease and obstructive
sleep
apnea, can change
over time as well. In
addition, multiple preoperative baseline data
were not collected;
this could have
precluded
the influence
of learning
effects on the results.
In addition, BMI is based entirely on height
and weight and fails to consider
other anthro-
pomorphic and physiological
features (e.g.,
fr ame. size, activity
level, muscle
composition).
For the present topic, the most relevant meas-
ures of obesity may have been neck circum-
ference or airway diameter. Future studies
designed to expressly
examine the effects
of
obesity
on vocal
function should
include
more
specific measures
and indicators
of upper air-
way features.
The results
of this study do not support the
hypothesis
that phonation, as it is perceived,
acoustically
analyzed, or aerodynamically
pro-
duced, differs between obese and nonobese
adults. When results
were analyzed longitudi-
nally,' however, one interesting observation
emerged. Phonation threshold pressure pro-
duced at 3070 and 800/o of the F0 range
de-
creased across
the 6-month oeriod of data
collection in both groups.
This r..ult *"s
most apparent
at 30o/o of the F0 range in obese
individuals over their course
of weight loss
consequent to bariatric surgicai procedures.
These
results
suggest that, although the voice
may not differ perceptually
or acousticall),,
it
may be easier to produce as
overall bodl' mass
decreases.
One speculation
is that iaryngeal
tissue bulk might decrease
along
with weight
Ioss, which would allow the vocal folds to
vibrate more easily. Additional studies are
needed
with larger samples, repeated
baseline
assessments, improved
methods for evaluating
neck and laryngeal a\rway
bulk, and perhaps
longer follow-up periods. Due to liigh individ-
ual variability, longirudinal
studies are
expected
to be more informative than cross-sectional
designs to evaluate
the effects of obesiry
and
weight loss on voice.
However, based
on the
means and standard deviations for PTP 300/o
from the cross-sectional
preoperative data ir.t
this study, L8 participants
per group would
provide sufficient power (500/o)
to detect a
statistically significant
difference at a :0.05.
The topic of obesity and its irnpact on
health is timeiy and important. Clinically,
sur-
geons
need
to be aware ofneck bulk and ainvav
patency as they prepare
for required proce-
dures, such as endotracheal intubation and
incision site, and postoperative
interventions
to minimize adverse events
that comoromise
pulmonary function.
Speech-language pathol
ogists also should
be aware of possible compli-
catiorrs that m y be encountered during
endoscopic laryngeal imaging and physical
ex-
amination of the r.reck.
Further, certain ther-
apeutic procedures for dysphonia require
massa€ie
or digital manipulation of the neck."
These procedures
can be complicated bi' r'e-
dundant tissue surrounding the laryngeal
carti-
lages. Whether excessive
bodl' rnass affects
vocal function is undetermined by this or p.rre-
vious research,
but one
potentially
ir.rformative
indicator may be PTP. Thus, clinicians mav
wish to evaluate
PTP in voice clients
for whom
excessive
weight may be a contributing factor.
ACKNOWTEDGMENTS
The views expressed in this article
are those
of
the authors and do not reflect the official policl'
of the Department of Army, Department of
Defense, or U.S. Government. We gratefullv
acknowledge the Department of the Anny
Congressionai Program
Grant, Armv Regional
Anesthesia and Pain Management Initiative,

42 SEMINARS
IN
SPEECH
AND
LANGUAGE/VOLUME
32,
NUMBER
1 2011
and the American Speech-Lrrnguage
Hearing
Foundation,
Clinical Research
Grant, for pro-
viding valuable
resources
that supported
this
research.
REFERENCES
1. Nguyen
DIVI,
El-Serag
HB.
The
cpidemiologlT
of
obesirv.
Gastroenterol
Clin North Am 2010;
39(1):l-7
2. Hora
F, N.ipolis
LN{,
Daltro
C, et el.
Clinicrl,
anthropometric :rnd upper irirway anatonic char-
acteristics
ofobese p;rrients
with obstructive sleep
rpnea syrrdrorne.
Respiretion 2007
;7
a$) :517
*521
3. Tagaya NI, Nakartn
S, Yrsum:r F, et al. pathoge-
netic role of increrrsed
nls:rl resistance
in obese
piltients with obstructive sleep irpneir syndrome.
Arn J Rhinol Allergy 2010;24(1):51-5.1
,1. N{artinho FL, Tangerin..r RP, Moura SNIGT,
Gregdrio LC, Tutik S, Bittencourt LR. Systenr_
;rtic head and neck physical exlminLLrion irs a
predictor of obstrtrctive sleep rpnea ir-r
class III
obese
parienrs.
Braz J Med Biol Res 2008;ll(I2):
1093-1097
5. Shelton
KE, lVoodson
H, Gay
S, Suratt
pN{.
Pharyngeal
lat in obstnrctive
sleep
apnea.
Arn Rev
Respir Dis 1993
;I 4B(2)
: 162-466
6. Horner RL, N{ohiaddin RH, Lowell DG, et al.
Sites
and sizes
of fat deposits
rr()uncl
rhe phuyn_r
in obese
patients
with obstructive
sleep
:rpnoel and
weight rnatched
controls. ELrr
Respir
J l9B9;2(7):
613-622
7. Busetto L, Calo' E, N,lrrzz:r
M, et rrl.
Upper linvay
size is related
to obesiry and bocly
far distribution
in wornen. Er.rr Arch Otorhinolarynsol 2009;
266(a):5s9-563
8. Salome
CM, King GG, Berend
N. physioloey
of obesiry
lnd c'ffects
on lung tunction.
J Appl
Physiol
2010;
108(1
):206-2
1 1
9. Rubin JS, Sataloff RTT, Korovin GS e,ls.
Diagnosis
:rnd Treatment of Voice Disorders.
San
Diego,
CA: Plural
Publishing;
2006
10. da Cunha NlB, Passerotti
GH, Weber R,
Zilberstein
B, Cecconello
I. Voice
fbrrure
charr:rc-
teristic
in rnorbid
obese
population.
Obes Surg
2009;
Septernber
18
[Epub
ahead
ofprint]
11. Physical
status:
the use lnd interpretirtion
of
.anthropontetry.
Geneva,
Switzerlrnd:
Report
of a
WHO Expert
Comrnittee;
1995
12. Wults FL, De Bodt NIS,
Molenberghs
G, et al.
The dysphonia severity index: rur objective
n)eeslrre
of vocrl qr"rrrliw
based
on a nrultipara_
meter
rrpproach.
J Speech.
Lane Hear Res 2000;
43(3):796-809
13. Hakkesteegt
l\{M, Wieringa
MH, Brocaar
NIp,
Nlulder
PGH, Feenstra
L. The interobserver
and
test-retesr
variabiliry
of the dysphonil severity
index.
Folia
Phoniatr
Logop
2008;60(2):96_90
14. Kernpster
GB, Gerratt
BR, Verdolini
Abbon K,
Barkmeier-Kraemer
J, Hillrnan RE. Consensus
auditory-perceptual
evirluation
of voice:,levelop-
menr of a stendardized
clinical
protocol.
Anr J
Speech
Lang
Pathol
2009;18
(2):L2a-fi2
15. Solomon
NP, Helou
LB, Stojrrdinovic
A. Clinical
versus
laboratory
ratings of voice using the
CAPE-V.
-[
Voice
2011;25(1):e7-e1.1
16. Smitherln
JR, Hlron TI. A clinical
merhod
fbr
esrimating
larynqeal
airway resistance
during
vorvel
production..
J Speech
Hear Disord 19g1;
16(2):138-1a6
77. 'litz.e
IR. The phvsics
of small-ampiirude
oscil-
lation
of rhe
vocal
fbltls.
J Acoust
Soc Am 19gB;
83(a):
i536-1552
18. Titze
IR. Phonation
threshold
pressure:
a urissing
link in glottrl rerodynamics.
J Acoust Soc Am
1992;9L(5):2926-2935
19. Solomon NP, Ranranrthan
P, N4.,rkashay
MJ.
Phonation
threshclld
pl.essure
xcross
the pirch
range:
prelin.rinrrry
test
of rr
rnodel.
J Voice
2007;
21
(5):5a1-550
20. Eadie
TL, Kapsner
N{, Rosenzweis
J, Waueh p,
' Hillel A, Nlerati
A. TI.re
role of experience
on
jtrdgments
of dysphonir.
J Voice
2010;24(5):564_
s73
21. Solomon
NP. Aerodynanics:
glottal
enterprises.
In: NIa E, Yiu E, eds. Hrrndbook
of Voice
Assessments.
San
Diego,
CA: phrral
publishing;
2017
22. Baken
RJ,
Orlikoif RF. Clinical
N{easurement
of
Speech
and Voice. 2nd ed. San Diego, CA:
Singular
Publishing
Group;
2000
23. Dogan M, Erluksel E, Kocak I, Celikel T,
Sehitoglu
N,IA.
Subjecive
and
objective
evrluation
of voice
quality
in parients
with asthma.
J Voice
2007;21(2):224-230
2-{. Solomon
NP, Grrrlitz SJ,
Milbrath RL. Respi_
r?ltory
and laryngeal
contributions
to tn&ximLlm
phonation
duration.
J Voice 2000;14(3):331*
340
25. Rov N. Assessment
ancl trertment of muscu-
loskeletil
tension
in hyperfunctional
voice
disor_
ders.
Int J Speech-Language
prrthol
2008;10(4):
L95-209