J Am Acad Audiol 8: 53-58 (1997)
Multifrequency 'l'ympanometry :
Effects of Ear Canal Volume Compensation
on Middle Ear Resonance
Wendy D. Hanks*
Bryan A. Mortensen*t
Abstract
Sweep frequency tympanograms compensating for ear canal volume at 200 and -400 daPa
were obtained on 106 ears of 53 subjects during four different sessions . Resonance frequencies were obtained and analyzed to determine normative data, resonance differences
between the compensation methods, and the interaction effect of running consecutive tympanograms. There were no gender or interaural differences for either compensation method .
Although statistically significant, the compensation method by order interaction found between
the first and second runs of tympanograms was not clinically significant. Compensation at
200 daPa was recommended when using sweep frequency tympanometry. Mean resonance
frequency difference of the two compensation methods was 400 Hz . Large intersubject variability emphasized the need for audiologists to use proper compensation norms when
examining middle ear resonance.
Key Words : Ear canal compensation, immittance, middle ear, tympanometry, multifrequency tympanometry, resonance
ultifrequency tympanometry may be
used to estimate the resonance frequency of the middle ear. Experimental data have shown that middle ear
pathologies alter tympanometric shapes at high
frequencies and shift the resonance frequency of
the middle ear transmission system . For example, increases in stiffness due to otosclerosis can
shift middle ear resonance to a higher than normal frequency, and increases in mass (or lack of
stiffness) due to ossicular discontinuity can shift
middle ear resonance to a lower than normal frequency (Colletti, 1975, 1976, 1977 ; van Camp et
al, 1983 ; Funasaka et al, 1984 ; van Camp and
Vogeleer, 1986 ; Funasaka and Kumakawa, 1988 ;
Shanks et al, 1988).
Resonance can be measured from acoustic
susceptance or phase angle values after compensating for ear canal volume (Shanks et al,
1993). To compensate for ear canal volume, the
M
"Hearing and Speech Science Laboratories, Brigham
Young University, Provo, Utah ; tcurrently affiliated with Dr.
Thomas A . Chasse, York, Maine
Reprint requests : Wendy D . Hanks, 138 TLRB, Brigham
Young University, Provo, Utah 84602
susceptance at high positive or at high negative
ear canal pressure is subtracted from all tympanometric values . A higher resonance frequency
for negative ear canal pressure is attributed to
normal tympanometric asymmetry, where there
is lower admittance under negative pressure
settings than under positive pressure settings
(Margolis and Smith, 1977 ; Vanpeperstraete et
al, 1979 ; Shanks and Lilly, 1981).
Multifrequency tympanometry can be performed with either the sweep pressure or sweep
frequency method . The sweep pressure method
holds the probe tone constant and sweeps the
pressure . The sweep frequency method holds
the pressure constant at specified intervals and
sweeps the probe tone frequencies.
Shanks et al (1993) examined the effects of
ear canal volume compensation on estimates of
middle ear resonance using sweep frequency
tympanometry. They evaluated the methods
currently used on two acoustic immittance
instruments with multifrequency capabilities .
The Grason-Stadler Model 1733 middle ear
analyzer Version 2 (GSI 33) plots the tympanometric peak to tail differences in acoustic
susceptance, as a function of probe tone frequency. The Virtual Model 310 digital impedance
Journal of the American Academy of Audiology/ Volume 8, Number 1, February 1997
instrument (Virtual 310) plots peak compensated phase angle as a function of probe tone frequency. Shanks et al concluded that the methods
described above give comparable results. The
mean middle ear resonance estimate was 817 Hz
for measurements compensated at 200 daPa
and was greater than 1052 Hz for compensation
at -350 daPa . Twelve of their 26 subjects reached
resonance above their cutoff' frequency of 1243
Hz when ear canal compensation was at -350
daPa ; therefore, an exact mean could not be
reported .
Margolis and Goycoolea (1993) provided
normative multifrequency data as obtained on
the Virtual 310 for both sweep pressure and
sweep frequency tympanometry. For the sweep
frequency method, they reported a mean resonance of 1135 Hz for tympanograms compensated at 200 daPa and 1315 Hz for compensation
at negative ear canal pressure (approximately
-500 daPa). Margolis and Goycoolea reported
that two or fewer subjects of 28 had resonance
frequencies above their cutoff' frequency of 2000
Hz for each method examined . Similar data
obtained on the GSI 33 have not been reported .
Reported differences of mean resonance frequencies compensated at positive and negative
ear canal pressures ranged from 140 to 300 Hz
(Margolis and Goycoolea, 1993 ; Shanks et al,
1993). It is not known if these resonance differences are consistent across subjects .
The GSI 33 compensates for ear canal volume based on the direction of pressure change
as the tympanogram is run. That is, if the tympanogram is run in the positive to negative
direction, it compensates from the positive tail,
and vice versa. In order to examine resonance
differences as a function of ear canal volume compensation using the GSI 33, two multiple frequency tympanograms must be obtained . In a
clinical setting, both tympanograms would be
recorded in one session . When these tympanograms are run consecutively, the middle
ear resonance for the second tympanogram may
change from the effects of multiple sweeps (Wilson et al, 1984) . The results may also be affected
by the interaction of running one tympanogram
from a positive to negative direction and another
from a negative to positive direction. This interaction would not occur on the Virtual 310 as
ear canal volume compensation may be performed after the tympanogram is run.
The purposes of this study were to (1) provide normative data for multifrequency tympanometry as obtained on the GSI 33, (2) examine
the differences in resonance frequency obtained
54
from sweep frequency tympanograms compensated at both positive and negative ear canal
pressures, and (3) examine the interaction effect
of running consecutive multifrequency tympanograms at both positive and negative ear
canal pressures.
METHOD
ifty-three adults (106 ears) 18 to 25 years
F of age served in this study (25 males, 28
females) . Each subject had normal hearing sensitivity ( :- 10 dB HL re : ANSI, 1989), a negative
otoscopic exam, no reported history of chronic
middle ear problems, and normal 226-Hz tympanometric peak pressures (100 to -150 daPa).
Each subject attended a 15-minute session on 4
different days, with no more than 2 weeks passing between the first and the last session.
At each session, two sets of multifrequency
data for each ear were collected with either ear
canal volume compensation at -400 daPa first,
followed by compensation at 200 daPa, or vice
versa. The probe was left in the ear between
test runs but the ear canal was vented . Initial ear
order and compensation methods were randomly
chosen for each subject. The order of the conditions was then counterbalanced among subjects .
This rotation of trials provided four estimates of
resonance from susceptance compensated at
-400 daPa for each subject and four estimates
compensated at 200 daPa . Trials 1 and 3 and trials 2 and 4 were identical, yielding test-retest
data for all of the conditions examined .
The multiple frequency procedure incorporated in the GSI 33 is based on the sweep frequency tympanometry method. The GSI 33 first
swept the probe tone from 250 to 2000 Hz in
50-Hz steps, while staying at the start pressure
(200 daPa or -400 daPa) . Susceptance and
phase measurements from each frequency were
stored in memory. An admittance tympanogram
(226 Hz) was run with a sweep pressure rate of
50 daPa/sec and a pressure range of -400 to 200
daPa . This tympanogram was needed to obtain
the tympanometric peak pressure (TPP). A second probe tone sweep was performed at TPP and
the susceptance and phase measurements again
were stored in memory.
Differences in. susceptance and phase values
between the first and second sweep of frequencies were calculated and plotted as a function of
frequency. As the GSI 33 measured susceptance
and phase in 50-Hz steps, the resonance frequency was marked by the equipment as the
measured frequency that was closest to 0 mmhos.
Effects of Compensation on Resonance /Hanks and Mortensen
Because of the step size, the calculated resonance frequency should be within 50 Hz of resonance frequencies up to 1000 Hz . For resonance
frequencies from 1050 to 2000 Hz, the variation
should not exceed 100 Hz (5% of the resonance
frequency) . Though phase measurements were
taken, they were not used in the calculation of
the resonance frequency (see Shanks et al, 1993
for a discussion) .
All equipment was calibrated before the
study began and weekly during data collection
and met ANSI standards (ANSI 1987, 1989).
RESULTS AND DISCUSSION
Resonance Frequency
A multivariate analysis of variance
(MANOVA) was performed on resonance frequency for ear, gender, compensation method,
and order (p < .05) . Results are displayed in
Table 1 . Mean resonance frequencies for ear
canal volume compensated at 200 daPa were significantly lower than the mean resonance frequencies compensated at -400 daPa . Mean,
median, mode, and 90 percent normal range
data for the two compensation methods are
shown in Table 2.
The mean resonance frequency with ear
canal volume compensated at 200 daPa is 908
Hz with a 90 percent range of 650 to 1300 Hz .
These results are in close agreement with other
studies (Hanks and Rose, 1993 ; Shanks et al,
Table 1 Resonance Frequency MANOVA
Results for Ear, Gender, Compensation Method
(Pressure), and Presentation Order (Order)
Source
Gender
Gender*Ear
Gender*Pressure
Gender*Pressure*Ear
Gender*Order
Gender*Ear*Order
Gender*Pressure*Order
Gender"Pressure"Ear"Order
of
F Value p Value
1,847
1, 847
1,847
1, 847
1,847
1,847
1,847
1, 847
0 .59
0 .07
0 .82
0 .08
0 .29
0 .48
2 .59
1 .72
Ear"Order
1,847
1, 847
Pressure
Pressure"Ear
1, 847
1, 847
Pressure"Order
Pressure"Ear*Order
1, 847
Order
1,847
Subject*Ear*Order (Gender) 51, 847
0 .02
101 .88
0 .16
12 .29
0 .40
1 .12
1 .08
Ear
tp < .05.
1,847
0.12
0 .4449
0 .7951
0 .3694
0 .7829
0 .5932
0 .4901
0 .1084
0 .1908
0.7264
0 .8880
0 .00011
0 .6905
0 .00051
0 .5259
0 .2953
0 .3351
1993). Margolis and Goycoolea (1993) obtained
a wider 90 percent range (800-2000 Hz) using
a slightly different sweep frequency tympanometry method than the present study. Differences between the results may be due, in
part, to intersubject variability and to the different step sizes that were used to measure resonance . The current study used 50-Hz steps,
Shanks et al used 113-Hz steps, and Margolis
and Goycoolea used 1/6-octave steps. For example, a middle ear resonance around 925 Hz could
be classified as 900 or 950 Hz in the current
study, 1000 Hz by Margolis and Goycoolea, and
1017 Hz by Shanks et al .
Mean resonance frequency with ear canal
volume compensated at -400 daPa for the current study is 1318 Hz, the same as that found
by Margolis and Goycoolea (1993) . The 90 percent range is 900 to 1750 Hz . Ninety percent
ranges obtained with ear canal volume compensated at negative pressure settings from
other studies are more variable than those compensated at positive pressure settings (Funasaka
et al, 1984 ; Margolis and Goycoolea, 1993 ;
Shanks et al, 1993). All of the 90 percent ranges,
however, are skewed toward the higher frequencies . A higher resonance frequency for negative ear canal pressure is attributed to normal
tympanometric asymmetry, where there is lower
admittance under negative pressure settings
than under positive pressure settings (Margolis
and Smith, 1977 ; Vanpeperstraete et al, 1979 ;
Shanks and Lilly, 1981).
Differences in the results could be due to several factors. First, ear canal pressures used to
compensate for ear canal volume were different,
ranging from -200 to -500 daPa . The estimate
of ear canal volume would be larger for tympanograms compensated at -200 daPa than at
-500 daPa . This would result in lower estimates
of resonance frequency at -200 daPa than at
-500 daPa . Second, as discussed above, the frequency step size used to obtain resonance frequencies was different among the studies and
would affect the estimate of the resonance frequency. Third, there is a great deal of variability in the asymmetry of tympanograms among
subjects . The estimate of ear canal volume would
change based on the amount of asymmetry, thus
changing the resonance frequency estimate .
Test-Retest Reliability
Table 3 displays the Pearson r correlations
for resonance frequency as a function of compensation method by presentation order. Results
Journal of the American Academy of Audiology/Volume 8, Number 1, February 1997
Table 2 Means (Standard Deviations), Medians, Modes, and Normal Ranges for
Resonance Frequency as a Function of Compensation Method by Presentation Order
Frequency (Hz)
Method (daPa)
Mean (SD)
Median
Mode
Normal Range
200 first
200 second
200 combined
-400 first
-400 second
-400 combined
908 (188)
924 (188)
916 (188)
1318 (308)
1286(317)
1302 (312)
900
900
900
1300
1250
1250
950
800
950
1000
950
1000
650-1300
650-1300
650-1300
900-1750
850-1850
900-1800
indicated that resonance frequencies obtained
with ear canal volume compensated at either 200
daPa or -400 daPa were very reliable . The reliability coefficients obtained in the present study
are identical to or higher than those reported by
Margolis and Goycoolea (1993) for sweep frequency tympanometry.
Table 4 presents means, medians, and 90
percent ranges for the differences in resonance
frequency as a function of ear canal volume
compensation . All of the descriptive data are
identical between the first and second repetitions, indicating very small differences in resonance frequency between trials .
Resonance Differences between
Compensation Methods
The mean difference in ear canal resonance
frequency for the two compensation methods is
approximately 400 Hz (Table 4) . The majority of
this difference may be accounted for by ear canal
volume compensation, as has been discussed
earlier (Shanks et al, 1993). The remaining difference may be accounted for by the effects of
sweep direction and multiple sweeps (Vanpeperstraete et al, 1979 ; Osguthorpe and Lam,
1981 ; Wilson et al, 1984 ; Shanks and Wilson,
1986).
Figure 1 displays the distribution of resonance frequency differences obtained for the
two compensation methods (averaged across
repetitions). The resonance frequency compensation differences ranged from 0 to 1300 Hz .
Some subjects had markedly different resonance
frequencies for the two compensation methods.
This may be attributed, in part, to the method
used to calculate resonance. As the GSI 33 plots
the difference in susceptance as a function of
frequency, the plot can be either steep or flat . The
56
resonance frequency is marked at the first measured frequency that is closest to 0 mmhos.
Other points may be equally as close to 0 mmhos
along the frequency continuum .
For example, one subject had a resonance
difference of 1000 Hz . For the tympanogram
that was compensated at 200 daPa, the frequency plot was very flat and near zero from
approximately 700 Hz to 1500 Hz . The GSI 33
chose 800 Hz as the resonance frequency. For the
tympanogram compensated at -400 daPa, the
frequency plot was also very flat but above zero
until 1700 Hz, the frequency which the GSI 33
calculated as the resonance frequency. Calculated
resonance frequencies on subsequent days for the
tympanograms compensated at 200 daPa were
1050, 850, and 700 Hz . Calculated resonance frequencies on subsequent days for the tympanograms compensated at -400 daPa were
1650, 1750, and 1750 Hz . These variabilities
caused by the instrumentation were also noted
by Valvik et al (1994) .
Some subjects had similar resonance frequencies for the two compensation methods.
Their tympanograms were very symmetric, or
conversely, asymmetric, near the positive tail .
This asymmetry near the positive tail is attributable to lower admittance under positive pressure settings than under negative pressure
(Margolis and Popelka,1977) . The large variability
Table 3 Test-Retest Reliability Coefficients
(Pearson r) from All Subjects (106 Ears)
Resonance Frequency
200 daPa first
200 daPa second
-400 daPa first
-400 daPa second
r
92
95
90
92
Effects of Compensation on Resonance /Hanks and Mortensen
Table 4 Means (Standard Deviations) and 90%
Ranges for Resonance Frequency Differences
(Hz) between Compensation Methods
-400-200 first
First repetition
Second repetition
-400-200 second
First repetition
Second repetition
Mean (SD)
90% Range
404 (308)
404 (308)
50-1000
50-1000
388 (317)
388 (317)
1350
1300
N 1250
x
>, 1200
c
1150
v
m
LL
m
C
C
0
N
cr
50-950
50-950
® -400 daPa 1 st, 200 daPa 2nd
® 200 daPa 1st, -400 daPa 2nd
100
1050
1
1000
950
900
850
800
in resonance differences between compensation
methods emphasizes the need for audiologists to
use proper compensation norms when examining resonance frequencies.
Compensation Method by Presentation
Order Interaction
The interaction between compensation
method and order was significant . This interaction (Fig . 2) indicated that resonance frequency increased when the -400 daPa pressure
was first, and decreased when the -400 daPa
pressure was second . The mean difference in resonance frequency for ear canal volume compensated at 200 daPa is 16 Hz, whereas the
mean difference for ear canal volume compensated at -400 daPa is 32 Hz . These differences
are less than the 50-Hz increment used to calculate the resonance frequency and would not
affect clinical decisions made regarding the presence or absence of pathology.
200 da Pa
Compensation Method
Figure 2 Interaction of mean resonance frequencies
between compensation method and order.
Preferred Compensation Method for
Clinical Use
The following variables were examined
before recommending one compensation method
for clinical use: test-retest reliability, intersubject variability, and 90 percent range. Both compensation methods were very reliable, with
compensation at 200 daPa slightly more reliable .
Intersubject variability was lower for compensation at 200 daPa . The 90 percent range should
allow for detection of abnormally high (increased
stiffness of the middle ear) or abnormally low
(increased mass of the middle ear) resonance frequencies . The available probe frequencies on
the GSI 33 range from 250 to 2000 Hz . A90 percent range that extended to either limit would
not be useful in detecting abnormalities. Based
on the above variables, multifrequency tympanometry compensated at 200 daPa is recommended for clinical use for the GSI 33 .
20
SUMMARY AND CONCLUSIONS
16
S
® DiXereece when -400d,Pe 1 at, 200 daPS 2nd
® Difreace when 200 ~P,1st, ~ d,P, 2nd
00
200
300
400
SW
600
700
800
Resonance Fregaenq DiHereace (Hz)
900
1000
1100
1200
1300
Figure 1 Distribution of mean resonance frequency differences as a function of compensation method and order.
weep frequency tympanograms compensating for ear canal volume at 200 and -400
daPa were obtained on 106 ears of 53 subjects
during four different sessions . Resonance frequencies were obtained and analyzed in order to
determine normative data (including test-retest
reliability), resonance differences between the
compensation methods, and the interaction effect
of running consecutive tympanograms using different compensation methods.
Results indicated that there were no gender
or interaural differences for resonance frequency
for either compensation method . Compensation
57
Journal of the American Academy of Audiology/Volume 8, Number 1, February 1997
at 200 daPa is preferred for obtaining resonance
frequencies using sweep frequency tympanometry on the GSI 33 . Test-retest data indicated
high reliability between trials . Although statistically significant, the interaction between compensation method and order was not clinically
significant .
Resonance frequency differences obtained
from the two compensation methods are
extremely variable and emphasize the need for
audiologists to use the proper compensation
norms when examining middle ear resonance.
Acknowledgment. The authors express their gratitude
to Walt Hanks, Judy Mortensen, David McPherson,
Richard Harris, and the reviewers for their help on earlier versions of this manuscript .
Portions of this paper were presented at the 1993
American Academy of Audiology annual convention in
Phoenix, AZ .
REFERENCES
Hanks WD, Rose KJ. (1993) . Middle ear resonance and
acoustic immittance measures in children. JSpeech Hear
Res 36 :218-222 .
Margolis RH, Goycoolea H. (1993) . Multifrequency tympanometry in normal adults . Ear Hear 14 :408-413 .
Margolis RH, Popelka GR . (1977) . Interactions among
tympanometric variables . J Speech Hear Res 20:447-462 .
Margolis RH, Smith P. (1977) . Tympanometric asymmetry. J Speech Hear Res 20 :437-446 .
Osguthorpe JD, Lam C . (1981) . Methodologic aspects of
tympanometry in cats . Otolaryngol Head Neck Surg
89 :1037-1040 .
Shanks JE, Lilly DJ . (1981) . An evaluation of tympanometric estimates of ear canal volume. J Speech Hear Res
24 :557-566 .
Shanks J, Lilly D, Margolis R, Wiley T, Wilson R. (1988) .
Tutorial : tympanometry. J Speech Hear Disord 53 :
354-377.
Shanks J, Wilson R. (1986) . Effects of direction and rate
of ear-canal pressure changes on tympanometric measures. J Speech Hear Res 29 :11-19 .
American National Standards Institute. (1987) .
Specifications for Instruments to Measure Aural Acoustic
Impedance and Admittance (ANSI 53.39-1987) . New York :
Shanks JE, Wilson RH, Cambron NK. (1993) . Multiple
frequency tympanometry : effects of ear canal volume
compensation on static acoustic admittance and estimates of middle ear resonance. J Speech Hear Res
36 :178-185 .
American National Standards Institute . (1989) .
Specifications for Audiometers (ANSI S3 .6-1989) . New
York : ANSI .
Valvik B, Johnsen M, Laukh E. (1994) . Multifrequency
tympanometry : preliminary experiences with a commercially available middle-ear analyzer. Audiology
33 :245-253 .
ANSI .
Colletti V (1975) . Methodologic observations on tympanometry with regard to probe tone frequency. Acta
Otolaryngol 80 :54-60 .
Colletti V (1976) . Tympanometry from 200 to 2000 Hz
probe tone . Audiology 15 :106-119 .
Colletti V. (1977) . Multifrequency tympanometry.
Audiology 27 :278-287 .
Funasaka S, Funai H, Kumakawa K. (1984) . Sweep-frequency tympanometry : its development and diagnostic
value. Audiology 23 :366-379 .
Funasaka S, Kumakawa K. (1988) . Tympanometry using
a sweep-frequency probe tone and its clinical evaluation .
Audiology 27 :99-108.
van Camp K, Creten W Van de Heyning P, Decraemer
W Vanpeperstraete P. (1983) . Asearch for the most suitable immittance components and probe tone frequency
in tympanometry. Scand Audiol 12 :27-34 .
van Camp K, Vogeleer M. (1986) . Normative multifrequency tympanometric data on otosclerosis . Scand Audiol
15 :187-190 .
Vanpeperstraete PM, Creten WL, van Camp KJ . (1979) .
On the asymmetry of susceptance tympanograms . Scand
Audiol 8:173-179 .
Wilson RH, Shanks JE, Kaplan SK . (1984) .
Tympanometric changes at 226 Hz and 678 Hz across 10
trials and for two directions of ear canal pressure change .
J Speech Hear Res 27 :257-266 .