Content uploaded by Judi A. Lapsley Miller
Author content
All content in this area was uploaded by Judi A. Lapsley Miller on Mar 02, 2015
Content may be subject to copyright.
Evaluating otoacoustic emission shifts due to middle-ear pressure
with tympanometry and wideband acoustic immittance
Abstract
Otoacoustic-emission measurements, used to assess inner-ear status, can be aected by negative middle-ear pressure.
Negative pressure can aect sound propagation through the middle ear into the inner ear, which can change the eective
stimulus level. Negative pressure can also aect sound propagation from the inner ear back through the middle ear and
into the ear canal. us, otoacoustic-emission measurements may be doubly impinged by negative middle-ear pressure,
potentially leading to false-positive signicant otoacoustic-emission shis. We tested wideband acoustic immittance,
226 Hz tympanometry, and transient-evoked otoacoustic emissions in 51 people before and aer inducing a middle-ear
change with the Toynbee manoeuver. As expected, measurements made aer the Toynbee maneuver were associated
with increased tympanometric peak pressures, increased middle-ear reectance, and decreased otoacoustic-emission
levels. Signicant emission shis were seen even when tympanometric-peak-pressure shis remained within clinically
normal limits. Both tympanometry and wideband acoustic immittance predicted otoacoustic-emission-shi magnitude,
but tympanometry produced higher correlations. ere may, however, be practical advantages to using wideband
immittance because the same equipment and probe t can be used as for the otoacoustic-emission test. ese ndings
have implications for the use of otoacoustic-emission tests in hearing-conservation programs.
Introduction
Middle-ear pressure (as measured by tympanic peak pressure, TPP) that deviates from atmospheric pressure by more
than ±50 daPa may diminish otoacoustic emission (OAE) levels to a clinically-signicant extent (Marshall et al., 1997).
When monitoring OAEs over time, changes in OAEs could be enhanced or diminished, depending on whether the
baseline and/or the monitoring test was aected, resulting in nding false-positive signicant OAE shis or missing true
OAE shis. In our laboratory and eld research, we ensure ears present with TPP within ±50 daPa before testing OAEs.
As we transition OAEs to use in hearing-conservation programs (HCPs), we are interested in establishing whether this
criterion is sucient for a population (Marshall et al., 1997, is based on just one ear), and whether wideband acoustic
immittance (specically absorbance) could provide similar information.
Absorbance (aka Transmittance) provides the amount of acoustic power absorbed by the middle ear in decibels (re 100%
absorbance) across 0.2 to 6 kHz. 0 dB indicates that all the power was absorbed.
Previous research using WAI has shown that absorbance is decreased with increased negative middle-ear pressure
especially at 1-2 kHz (e.g., Margolis, Saly, & Keefe, 1999; Keefe & Simmons, 2003; Feeney, Grant, & Marryott, 2003).
ompson (2013) found DPOAEs were reduced from 1-3.5 kHz and less power was absorbed especially at 0.5-2 kHz
when the ear was pressurized; however, because TPP was not measured at the same time, a direct relationship among
the three measurements could not be established. Most previous research is for clinically-signicant negative middle-ear
pressure. For this application we are interested in subtle amounts of negative middle-ear pressure, as found in a normal
population (i.e., between -100 and -50 daPa is still considered clinically normal). No study has previously looked at
tympanometry, WAI, and transient-evoked OAEs (TEOAEs) all measured in the same test session.
Method
Subjects: 51 adults (19-69 years; 91 ears) with normal 226 Hz tympanograms, clean ear canals, and no noise
exposure within 24 hours, were tested in one or more sessions. Subjects with at least one stable test run were included.
Equipment: GSI 33 Middle Ear Analyzer Model 1733. Mimosa Acoustics HearID Release 5; HearID Module
3.4.1.44, MEPA 3 Module 4.3.21.11, TEOAE Module 3.5.0.0 (with custom modications).
Test Run: A “test run” consisted of a series of 4 measurements:
1. 226 Hz Tympanometry for tympanic peak pressure (TPP) in daPa.
2. WAI using chirp stimulus for absorbance (0.2-6 kHz) in dB re max absorbance (23 Hz bins, or 1/2 & 1 octave bands).
3. TEOAE using a nonlinear 50 dB SPL (66 dB pSPL) chirp, analyzed into half-octave bands in dB SPL.
4. Repeat 226 Hz Tympanometry (for a stable test, the second TPP must be within 10 daPa of the rst TPP).
A “test-run pair” was one test-run made with the middle-ear at ambient pressure and one test-run made aer inducing a
middle-ear change with the Toynbee or Valsalva maneuver.
Procedure: Aer screening, the subject did one test run with the ear as presented. If the test run was stable, the
subject attempted to induce a middle-ear change by using a Toynbee (close airways and swallow) or Valsalva (close
airways and exhale) maneuver. e other ear was then tested, sometimes starting in a pressurized state if the subject could
maintain it from the previous test run. ey then swallowed or coughed to release the pressure. e tester repeated test
runs until they were satised stable data were obtained under both Ambient and Pressurized conditions, or until it was
obvious that the subject was not able to induce or maintain the pressurized state. Within a test run, the WAI and TEOAE
test used the same probe t, and the two tympanometry tests had a probe ret. Across test-runs, the tympanometer and
OAE probe were ret. Some subjects provided multiple test-pairs over one or more sessions if they could reliably induce
and maintain stable middle-ear pressure states.
Data sets
• “Key”: 28 ears (22 subjects) producing one within-session test-pair with (1) TPP near 0 daPa and (2) induced TPP
≤-50 daPa or ≥50 daPa. If there were multiple tests, the most stable near 0 daPa and largest TPP tests were chosen.
• “Extended”: 50 ears (34 subjects) producing 1 or more test-pairs with (1) TPP near 0 daPa and (2) aer a Toynbee or
Valsalva (which may or may not have produced a large TPP change). Comparisons were within and across-sessions.
• “BiggestOAEShi”: 50 ears from “Extended”, choosing the within-session test-pair with the largest 1 kHz TEOAE
decrease. One test-run was at Ambient pressure and the other test-run was aer a Toynbee or Valsalva; however, the
TPP was not used to select the test-run because for this dataset it is used as a dependent variable.
4. Change in TEOAE and Absorbance with increasing (absolute) TPP
5. Correlations between 1 kHz TEOAE shift and TPP/Absorbance
So far the emphasis is on detecting ears with abnormal TPP, because we know that abnormal middle-ear pressure is
related to diminished TEOAEs. But TPP is not a perfect gold standard (sometimes a change in TPP doesn’t produce an
OAE shi) - especially for ears with only mild changes.
What we really want to know is how to measure middle-ear status to best predict and detect tests where the TEOAE has
shied due to the middle-ear status.
e “BiggestOAEShi” data set was used for the following analysis. 1 kHz was chosen because it was the region of biggest
change with the least missing TEOAE data (0.7 kHz was more aected by noise).
• “Absolute pressure”: correlations between the largest 1 kHz TEOAE shi from the Ambient baseline and
TPP/Absorbance for the pressurized state (aer a Toynbee or Valsalva).
• “Pressure change”: correlations between the largest 1 kHz TEOAE shis from the Ambient baseline and the shi in
TPP/Absorbance between the Ambient and Pressurized states.
Absolute (unsigned) TPP and absolute change in TPP were used because positive and negative TPP has a grossly similar eect on the middle-ear. Individual
octave-averaged Absorbance ranges (0.5, 1, 2, 4 kHz) were used to reduce the number of variables. In addition, a multivariate absorbance predictor was obtained
from all 4 absorbance bands. A nal comparison added TPP to the multivariate predictor. A multiple regression was used to obtain the predictor.
TPP Absorbance octave bands TPP & 4 octave
Absorbance bands
Correlation with
∆TEOAE at 1 kHz 0.226 kHz 0.5 kHz 1 kHz 2 kHz 4 kHz Multivariate Multivariate
Absolute pressure -0.62 0.21* 0.42 0.38 -0.17* 0.48* 0.63
Pressure change -0.65 0.34 0.48 0.38 -0.36 0.61 0.74
* ns
Multivariate absorbance can work as well as TPP at predicting TEOAE shi size providing there is an ambient baseline
for comparison (0.61 vs -0.65). When considering a multivariate approach for change from baseline, absorbance provides
information above and beyond TPP (0.74 vs -0.65). When considering just the pressurized state (an absolute judgement),
absorbance does not help further than TPP (0.63 vs -0.62) and on its own performs worse (max 0.48).
Lynne Marshall1, Judi Lapsley Miller1,2, & Charlotte Reed3
1. Naval Submarine Medical Research Laboratory; 2. Mimosa Acoustics, Inc; 3. Research Lab of Electronics, MIT
Figure 6: Scatterplots for data underlying the three correlations in the above table for the Pressure Change condition:
(le) TPP (daPa), (middle) Multivariate absorbance (dB), and (right) Multivariate TPP & Absorbance (dB). e gray
horizontal bar indicates the size of a signicant threshold shi at 1 kHz for an individual ear (99% CI), which is 2 dB.
Discussion
Does negative middle-ear pressure aect using OAEs in HCPs?
1. Detecting noise-induced change in OAEs – MAYBE. True noise-induced OAE changes are seen mainly at high
frequencies, whereas frequencies below 2 kHz may show false positive OAE shis due to negative middle-ear
pressure.
2. Detecting low-level TEOAEs at 4 kHz (an indication of subclinical damage) - NO
3. Assessing susceptibility to noise-induced hearing loss with a medial-olivocochlear reex (MOCR) test - YES. e
most sensitive region is <2 kHz.
4. e deviations in middle-ear pressure from ambient pressure in this study were relatively small. Larger amounts of
negative middle-ear pressure are more easily detected with absorbance.
5. As a rough estimate, from screening-day data in our previous lab studies on healthy adults (668 ears), we expect less
than 2% of ears to present with TPP ≤-50 or ≥50 daPa, and less than 4.5% with TPP between -45 to -25 daPa or 25
to 45 daPa. If the baseline test battery includes OAEs, WAI, and tympanometry, with normal middle-ear pressure
required at baseline, and follow-up testing includes OAEs and WAI without tympanometry, any OAE changes can be
checked against absorbance changes. In the population, there should be few errors in categorizing OAE shis as being
due to noise damage instead of due to middle-ear pressure changes.
Conclusions
1. Tympanometry is an imperfect gold standard - it does not accurately predict which ears will experience signicant
threshold shi for small amounts of negative and positive middle-ear pressure. However, a 50 daPa criterion is not
inappropriate although a stricter one could be used, and would be a good idea if MOCR testing is involved.
2. When predicting which TEOAE measurements were aected by middle-ear status:
a. Absolute (single measurement) values of TPP from tympanometry are better than single measurements of
Absorbance (either in individual octave bands or a multivariate combination).
b. Relative changes from an ambient baseline are equally detectable with TPP and a multivariate combination of 4
Absorbance frequency bands.
3. We saw little-to-no evidence of the high-frequency OAE enhancement with middle-ear pressure changes that were
seen by others (e.g., Margolis, et al., 1999; Marshall et al., 1997) in the group. Some individual ears showed small
changes, but not by enough to move the group average.
4. ere are practical advantages to using WAI. It uses the same equipment as the OAE test, with the same probe t.
It doesn’t require pressurizing the ear, and it adds no extra time as the WAI test can be used to calibrate the OAE
stimulus.
References
Feeney, M. P., Grant, I. L., & Marryott, L. P. (2003). Wideband energy reectance measurements in adults with middle-ear
disorders. J Speech Lang Hear Res, 46(4), 901-911.
Keefe, D. H., & Simmons, J. L. (2003). Energy transmittance predicts conductive hearing loss in older children and adults.
J Acoust Soc Am, 114(6 Pt 1), 3217-3238.
Margolis, R. H., Saly, G. L., & Keefe, D. H. (1999). Wideband reectance tympanometry in normal adults. J Acoust Soc
Am, 106(1), 265-280.
Marshall, L., Heller, L. M., & Westhusin, L. J. (1997). Eect of negative middle-ear pressure on transient-evoked
otoacoustic emissions. Ear and Hearing, 18(3), 218-226.
ompson, S. E. (2013). Impact of negative middle ear pressure on distortion product otoacoustic emissions.
Unpublished PhD thesis. CUNY.
Acknowledgements
Lynne Marshall is now at the University of Connecticut. anks to Tim Villabona for assistance with data collection
and to Joshua Hajicek for suggestions for methodology. is work was supported by the Oce of Naval Research, work
unit 50904. e views expressed in this article are those of the author and do not necessarily reect the ocial policy or
position of the Department of the Navy, Department of Defense, nor the U.S. Government. Dr. Lynne Marshall was an
employee of the U.S. Government. is work was prepared as part of her ocial duties. Title 17 U.S.C. §105 provides that
‘Copyright protection under this title is not available for any work of the United States Government.’
A PDF of this poster is available from Judi at judi@psychophysics.org or
https://www.researchgate.net/prole/Judi_Miller2/publications
Questions
1. Can Absorbance discriminate ears with middle-ear pressure (TPP) ≤-50 or ≥50 daPa?
2. Is ≤-50 daPa a sucient criterion to ensure middle-ear pressure doesn’t change OAEs?
3. Can TPP, absorbance, or change in TPP or absorbance, predict OAE shi size aer pressurization?
Results
1. Mild middle-ear pressure was induced
2. Absorbance decreased when ear was pressurized
3. Change in TEOAE and Absorbance for various TPPs in an individual ear
Figure 5: Box plots for the “Extended” data set for TEOAE and Absorbance change from Ambient baseline in dB. e absolute
(unsigned) value for TPP was grouped into ambient, slight, mild, and moderate. Signicant change in TEOAE and Absorbance
occurs below ~2 kHz once absolute TPP ≥ 50 daPa. e box is interquartile range, whiskers include data-points within
1.5xIQR, dots are outliers). e gray bands indicate the size of a statistically signicant change (99%CI) for an individual ear.
Number of ears is shown below each boxplot.
Figure 4: A cherry-picked example for one ear in the “Extended” data set. is subject was able to induce both positive
and negative TPP over a wide range, and did not have any low-level TEOAEs requiring noise-oor substitutions.
As TPP moved away from ambient pressure, TEOAEs diminished at 1.4 kHz and lower. Likewise absorbance also
diminished at 2 kHz and lower. e horizontal gray bar indicates the size of a statistically signicant change (99% CI
derived from the standard error of measurement). Twelve test runs were conducted over two sessions (six in each).
Figure 3: ROC analyses on Absorbance distributions
conrms that the best separation (largest Area under
the ROC curve) is around 0.7-1.3 kHz.
Figure 2: Absorbance distributions (10-90th percentiles) for the 28
“Key” ears. Within the same ear, absorbance decreased due to middle-
ear stiening. e 0.7-1.3 kHz region showed the least overlap.
Figure 1: (le) TPP achieved for the 28 ears in the “Key”
dataset at Ambient pressure and aer Pressurization.
(right) TPP achieved for the 50 ears in the “Extended”
dataset, including repeated test-runs in the same ears, and
the test-runs that did not achieve criterion for inclusion in
“Key”, providing they were stable.
