International Journal of Pediatric Otorhinolaryngology
29 (1994) 235-248
The effects of noise in transient EOAE newborn
hearing screening”f
John T. Jacobson*a, Claire A. Jacobsonb
‘Department of Otolaryngology - Head and Neck Surgery, Division of Audiology.
Eastern Virginia Medical School, 825 Fairfax Ave, Suite No. 510.
Norfolk, VA 23501, USA
“Department of Audiology, Children’s Hospital of the King’s Daughters, Norfolk, VA, USA
(Received 27 July 1993; revision received 30 September 1993; accepted I October 1993)
The use of transient evoked otoacoustic emissions (TEOAEs) has been advocated as the
first stage entry level technique for universal newborn hearing screening. To date, the majority
of TEOAE infant testing has been conducted under controlled noise conditions; i.e., acousti-
cally treated sound suites. As a result, previously reported TEOAE evaluations may not
realistically represent test outcomes in actual hospital screening settings. The purpose of this
study was to compare the results of TEOAEs with auditory brainstem response (ABR) hear-
ing screening in a hospital environment where noise conditions do not meet the same ambient
noise specifications as those found in sound rooms. A total of 119 stable newborns (67 high
risk, 52 normal) ranging in post-conceptual age (PCA) from 33 to 41 weeks received both the
ABR and TEOAE screening protocols. Testing was conducted at crib side in either the well
baby nursery or the neonatal special care unit (NSCU). Newborn ABR screening failed 8
(3.8%) of 224 ears, whereas TEOAE testing failed 85 (38.4%) and could not test another 22
(9.8%) ears. That is, only 117 (52.2%) of the 224 ears passed the TEOAE test. Using the ABR
as the reference test the specificity and sensitivity for TEOAE was 52% and 50%. respectively.
Noise levels measured by the probe microphone within the ear canal exceeded those levels (30
dBA SPL) recommended for TEOAE newborn hearing screening. Results of this study sug-
gest that under realistic hearing screening test conditions, TEOAE results may be influenced
by the level of noise in the testing environment. Whereas significant advances have been at-
tained in TEOAE measurement during the past decade, clinical evidence supports the need
for continued research aimed at solving problems before this technique can be used efficiently
for newborn screening.
* Corresponding author. Tel.: (804) 446 5934.
tPortions of this article were presented at the SENTAC Meeting, December, 1991
Ol65-5876/94/$07.00 0 1994 Elsevier Science Ireland Ltd. All rights reserved.
SSDI Ol65-5876(93)01002-2
236 J. T. J acobson, C. A. J acobson / Int. J. Pediatr. Ororhinolaryngol. 29 (1994) 235-248
Key words: Otoacoustic emissions; Auditory brainstem response; Hearing loss; Infants;
Newborns; Noise levels; Hearing screening
1. Introduction
Evoked otoacoustic emissions (EOAE) are considered a by-product of sensory
outer hair cell transduction. They represent frequency dispersion in response to
transient stimuli and are detected as measurable cochlear echoes (release of acoustic
energy) in the external auditory canal [8,16,24,53]. EOAEs are preneural in origin
and are directly dependent on outer hair cell integrity. Furthermore, auditory de-
ficits tend to abolish the recording of an evoked emission [36,39,41,42]. Initially
reported by Kemp [ 151, EOAEs have been introduced clinically as an objective tech-
nique for evaluating cochlear function and sensitivity, and to a lesser degree, the
status of the middle ear apparatus.
There is growing evidence that transient evoked otoacoustic emissions (TEOAEs)
are present in newborns, infants and children with normal hearing and that TEOAEs
are sensitive to the presence of even mild hearing loss [2,5-7,9,10,14,17,
l&30,35,38,43,44,46,54]. Through initial success with healthy term infants, the use
of TEOAEs has also been extended to the high risk infant population
[4,21,38,47,50,54]. The consensus of clinical study has advocated the use of emis-
sions as a screening measure which has resulted in recommendations by the National
Institutes of Health (NIH) [34] that this technique be adopted as the first level of
screening in universal infant hearing testing. The NIH contends ‘...that EOAE shows
best promise as a rapid, cost-effective means of quickly discharging all babies with
normal auditory systems’. They also state that ‘... the sensitivity of EOAE in the de-
tection of congenital hearing impairment is very high, but newborn EOAE testing
tends to have more false-positives when compared to auditory brainstem response
(ABR), especially during the first 48 h of life’. Because of their concern for EOAE
specificity, the NIH encourages the use of ABR as a ‘second-stage’ screening tech-
nique for all babies who fail the initial EOAE screen. Thus, the NIH draft recom-
mendation states that ‘ . ..the preferred model for universal screening begins with an
initial screen by EOAE’ and all babies that fail should be rescreened by ABR’.
It is clear from empirical observation that valid measures of TEOAEs should be
conducted under relatively ideal test conditions [19,25,44]. These conditions would
minimally include a sleeping or quiet/restful infant (absence of motor activity) and
an acoustic environment with low ambient noise. Such acoustic control is typically
difficult to achieve in hospital areas such as the newborn intensive care unit (NICU)
or the neonatal special care unit (NSCU), which are typical sites for screening infants
at risk for hearing loss [ 1, 1 1,13,29,33]. Frequently, rooms adjacent to the NSCU are
available for neonatal testing; however, these settings are often plagued with their
own difficulties, including excessive noise levels, space and traffic problems and
improper electrical grounding. Finally, it is not realistic to expect that infants can
be routinely transported to a controlled sound environment for screening.
To date, most TEOAE infant studies [6,10,28,38,43-46,501 have been conducted
under ideal environmental settings, usually in an acoustically treated sound suite.
.I. T. Jacobson. C. A. Jucobson / ht. J. Pediufr. Olorllinolur~~ngol. 29 / 1994) 235-248 731
Unfortunately, real-world hospital environments do not meet the same noise stan-
dards and conditions. Furthermore, the well-known high levels of physiologic noise
typically associated with infants will add to the overall noise floor problem. For ex-
ample, noise interference and probe fit (a tight, but not hermetically sealed ear canal
is required) may account for as much as 50% of the unacceptable TEOAE results
demonstrated in young subjects [7,20]. Kemp and Ryan [20] have noted that obtain-
ing a good probe fit is difficult in neonates and noise may easily penetrate the tissue
forming the meatus. Additionally, patient status including any motor activity, noisy
breathing, swallowing, sucking and other myogenic activity may contribute to the
ultimate pass/fail outcome. Because ambient and ear canal noise levels may affect
the reliability of TEOAE test results, the theoretical advantage of TEOAEs as a
method for screening newborn babies at risk for hearing loss may not be realized
in actual practice. Thus, the present study was undertaken to determine if TEOAEs
infant hearing screening, when conducted in a typical NSCU or well-baby hospital
nursery, would produce test results similarly reported to those obtained in a sound
treated environment. Conventional ABR testing, currently in place as the primary
hearing screening technique, was conducted on every infant in this study and used
as a comparative test measure.
2. Method
2.1. Subjects
A total of 119 stable newborns, 67 high risk (43 male, 24 female) and 52 normal
(27 male, 25 female) comprised this patient group. Newborns ranged in post-
conceptual age (PCA) from 33 to 41 weeks for this investigation. High risk infants
in this study met at least one risk category as defined by the 1990 Joint Committee
on Infant Hearing (JCIH) risk register and/or admission to the hospital NICU. ABR
and TEOAE screening techniques were administered to every infant during the same
test session. The order of the screening procedure was arbitrary but influenced by
the availability of test equipment and appropriate staff.
Testing was conducted at crib side in either the well-baby nursery or the NSCU
when the baby was in a state of natural sleep or resting quietly. As in the majority
of hospital facilities, hearing screening is conducted in a ‘step-down’ unit (NSCU).
Screening in the NICU has not been encouraged because of the unstable physical
and neurological status of the newborn prior to discharge from this setting. The rela-
tionship between infant status at the time of testing and pass/fail outcome has been
reported recently [31]. Every effort was taken to test only when recommended condi-
tions were observed.
The status of the middle ear was not routinely evaluated prior to screening and
therefore not a consideration in the initial test protocol. There are many reasons for
this approach. First and foremost, hearing screening is an attempt to identify those
with disease from those without, not a diagnostic evaluation to identify specific dis-
orders; secondly, it is difficult at best and often problematic to visually observe the
tympanic membrane in the neonate. In most instances, a narrow and cartilaginous
external auditory canal often prevents accurate inspection. Thirdly, on the hospital
floor, sophisticated microscopic instrumentation is not available for use by a physi-
238 J. T. J acobson, C. A. J acobson /ht. J. Pediatr. Otorhinolaryngol. 29 (I 994) 235-248
cian and/or residents. Fourthly, immittance audiometry remains suspect as a reliable
measure to middle ear compliance in the neonate. Finally, the condition of the mid-
dle ear should be reflected in the pass/fail test results of the two screening measures,
i.e., abnormal middle ear status would elevate ABR thresholds and prolong wave
latencies and as earlier stated, TEOAEs are sensitive to even mild conductive
pathology and would result in absent emission reproduction.
2.2. Instrumentation and stimuli - auditory brainstem response
Both conventional (Navigator, Bio-Logics Systems Corp.) and automated
(ALGO- 1, Natus Medical, Inc.) ABR instrumentation were employed. The ALGO- 1
is a battery-operated microprocessor dedicated solely for newborn ABR screening.
This device uses a statistical model for objective response detection and incorporates
a number of unique design features that include: (1) a disposable circumaural foam
cushion with an adhesive back that is sealed around the infant’s ear resulting in a
noise reduction in excess of 14 dB SPL at 2000 Hz; (2) an artifact-rejection system
that is designed to control for increased ambient noise such as that experienced in
the NICU; (3) to improve the myogenic rejection capabilities of the system, a series
of parallel-filters were designed to allow for separate filtering characteristics for sig-
nal processing and for myogenic-artifact detection; and (4) in order to extract signal
response (wave V of the ABR) embedded in EEG activity, a template-matching de-
tection algorithm is used [48,49]. A detailed account of the operation and principles
of the ALGO- are described elsewhere [22,23,37].
The ALGO- employs a 35 dB alternating click stimulus presented monaurally at
a rate of 37 pulses/s. Unfiltered clicks offer an acceptable auditory signal for infant
ABR testing by providing optimum stimuli for eliciting synchronized neural dis-
charge patterns. A presentation rate of 37 pulses/s offers a reasonable compromise
between wave morphology and test duration. That is, at 37 pulses/s it will take about
45 s to complete a total of 2000 repetitions (given the absence of any artifact rejec-
tion) without sacrificing wave peak amplitude and shape.
Conventional ABR instrumentation was programmed to produce stimulus par-
ameters as similar as possible to the automated device given inherent differences be-
tween the two systems. Stimuli were transduced by an Etymotic ER-3A insert
earphone with modified impedance tip adapters inserted directly into the ear canal.
Intensity was expressed in decibels normal hearing level (nHL) relative to a group
of normal hearing young adults. The peak equivalent sound pressure level for 0 dB
nHL was 31 dB.
Silver-silver chloride cup electrodes were attached to the forehead, midline and
just anterior to the fontanelle (non-inverting/positive), the ipsilateral (inverting/nega-
tive) and contralateral (ground) earlobe. Once attached, electrodes remained in posi-
tion for both tests. Interelectrode impedance was maintained at 55 KQ for both
systems throughout testing. It should be noted that the automated device will not
‘average’ if electrode impedance exceeds 12 KQ. The electrical activity recorded
between the forehead and ipsilateral earlobe was amplified, filtered and digitized
prior to averaging.
J. T. J acobson. C.A. J acobson / Int. J. Pediarr. Otorhinolaryngol. 29 ( 1994) 235-248 239
2.3. Instrumentation and stimuli - transient EOAEs
TEOAEs were obtained with a commercially available IL088 Oto-Dynamic
Analyser, using the default protocol developed by Kemp et al. [18,19]. The IL088
system includes a signal processor, probe and software compatible with DOS format.
This instrument measures transient evoked otoacoustic emissions (TEOAEs) apply-
ing an 80 ps rectangular pulse stimulus with a frequency response of approximately
600-6000 Hz [9,40]. The default click rate (SO/s - an effective rate of about 12.5/s)
and intensity level (80 dB pe SPL) were chosen; however, intensity was observed to
vary slightly and is likely influenced by ear canal volume and probe tit [14,35,42],
ear canal resonance [26], physical characteristics of the EAC and middle ear im-
pedance [32].
The differential nonlinear test paradigm inherent to the default protocol was used
throughout; i.e., stimulus artifact was cancelled using a sequence of clicks grouped
into sets of 4 pulses that were subaveraged using a 20 ms time window. Three com-
pressional pulses of equal amplitude are followed by one rarefaction pulse 9.5 dB
greater in amplitude. After each block of stimuli, the responses in the ear canal are
summed, cancelling the linear components and leaving only the nonlinear cochlear
generated responses, i.e., TEOAEs.
The responses of each block of 4 stimuli were averaged for each ear and stored
into 2 separate buffers for comparative analysis. In the default setting, each average
response was the average of a maximum of 260 click stimuli trains, a total of 1040
pulses; however, in the present study, because of the screening process, testing was
terminated if, in the judgment of the examiner, a stable TEOAE was present and the
percent reproducibility exceeded the pass criteria of 50%. The validity of the re-
sponse measures are crossed-correlated providing a percent indication of similarity
between the 2 averaged waveform responses. The noise rejection levels were shifted
on an individual ear basis to compensate for noise conditions. The reader is referred
to the IL088 User Manual (1988), and Kemp et al. [19] for a detailed description of
instrumentation and test stimulus protocol.
2.4. Failure criteria
For conventional ABR screening, a failure was defined as the absence of an iden-
tifiable and replicable wave V peak for either ear at 35 dB nHL. Automated test fail-
ure criteria were predicated on a statistical likelihood ratio model; i.e., if the data
collected did not exceed a predetermined likelihood ratio (i.e., 160) a Refer message
was displayed on the instrument suggesting additional follow-up measures. Briefly,
the Algo-l uses a template-matching detection algorithm to determine the presence
of an ABR. The design of the template is based on the morphology of a normal in-
fant ABR. Using this derived waveform, 9 data points are selected for use in the
template based on their stability with age and stimulus intensity. The infant response
is analyzed every 500 sweeps until a statistically significant signal is identified. Com-
parative ABR test results between conventional and automated instruments in the
same population have been shown to be highly correlated. For example, Peters [37]
reported a 94% agreement between the automated and conventional ABR systems
despite several skilled operators using a variety of conventional instruments and test
240 J. T. J acobson, C.A. J acobson / ht. J. Pediatr. Otorhinoloryngol. 29 (1994) 235-248
protocols. Kileny [23] found in a group of 507 ears tested from 286 high-risk infants
no false-negatives findings and only a 3.0% false-positive outcome when comparing
automated and conventional ABR. Jacobson et al. [12] found similar comparative
test results to those presented by Kileny. These authors reported an overall operating
efficiency of 95% when comparing conventional and automated testing following the
evaluation of 447 infant ears. The reliability of each ABR system and their between-
system validity has been proven; for a detailed description of instrumentation and
the screening protocol, the reader is referred to Jacobson et al. [12].
A TEOAE pass was defined as the presence of a recorded evoked emission with
greater than 50% reproducibility. The minimum 50% reproducibility level is consis-
tent with hearing sensitivity of 30 dB HL or better in the mid-frequency (300-3000
Hz) range [l&19]. In the absence of standardized protocol, the use of the 50%
criterion is also consistent with an oral recommendation from the developer of the
test equipment (D. Kemp, pers. commun.), what appears to be ‘accepted practice’
(361 and most frequently reported in the literature [19].
Ambient noise levels were measured at random intervals during the study using
a noise dosimeter (Quest Electronics, Micro-l 5) in both the NSCU and the well-baby
nursery. Mean noise levels were 63.4 dBA and 47.8 dBA SPL, respectively.
3. Results
The overall pass-fail results by screening procedure are summarized in Fig. 1. Of
the 238 possible ears to be tested, repeatable ABRs were recorded from 224. A total
of 14 ears were not tested because of either infant/audiologists schedule, or parent
non-compliance and, thus eliminated from further comparative investigation. Of the
224 ears, 8 (3.6%) failed initial ABR screening (5 at-risk and 3 normal infants from
the well-baby nursery). TEOAEs which achieved screening criteria were obtained
from 117 of 224 ears, whereas 85 ears failed. Despite discretionary shifts in the noise
1 2 3 4 5
ABWTEOAE Screening Result
Fig. 1. Pass-fail outcome for ABR and TEOAE screening. Results represent a total of 224 ears of
newborns screened for hearing loss. ‘TEOAE CNT’ represents 22 babies that could not be tested during
the allotted 45 min screening period. See the text for further explanation.
J .T. J acobson, C.A. J acobson /I N. J. Pediarr. Olorhinolaryngol. 29 (1994) 235-248 241
rejection level by the examiner, recordable emissions could not be obtained from
another 22 ears, presumably due to excessive noise levels as measured within the ear
canal, possibly secondary to probe fit and placement, which continually triggered re-
jection of the data sample. When these latter two groups are combined, a total of
107 ears did not meet TEOAE screening criteria of a reproducible emission of great-
er than 50%.
A typical newborn nonlinear TEOAE response obtained from the left ear of one
male infant is shown in Fig. 2. The TEOAEs waveforms in the lower left panel are
the overlapped emissions in the time domain from 2.5 to 20 ms. The emissions power
spectrum (analyzed by fast Fourier transform - FFT) in the top right panel il-
lustrates the separation of noise from response and respective noise/response
numerical values along the right side of the plot. A reproducible response of 97%
lhulus I II1088 OIOWtMIC MUMSER ]Bbd Response
2lrin 52secs
po; r;es
Fig. 2. A normal TEOAE waveform emission response in the time domain from 2.5 to 20 ms (center bot-
tom panel) obtained from the left ear of a 37 wk PCA male. Amplitudes of A and B waveforms are mea-
sure in mP, pressure units across time in ms. The top left panel depicts the acoustic waveform of the
pulse stimulus over a 4 ms period recorded immediately prior to the test onset. Probe stimulus is
calibrated in pascals. The right center panel labeled RESPONSE displays the frequency spectrum of the
recorded emission (white area); the foreground in black represents the noise spectrum between the 2 recor-
dings. The top right panel labeled NOISE gives the rejection values in mP, and peak dB SPL: the ‘No
Lo’ enters the number of data sweeps from stimulus blocks that were accepted and the ‘No Hi’. the num-
ber rejected. The percentage of accepted sweeps (‘1% No Lo) and the average noise level in dB SPL in the
ear canal at the time sweeps were accepted follow. The right panel labeled RESPONSE offers the emission
level in dB SPL of the ‘Echo’ which is defined as the average of the 2 summed traces whereas ‘Repro’
is the correlation in percent between the two summed traces. A - B is the difference in dB SPL between
the two summed traces. The panel STIMULUS represents the stimulus peak in dB SPL in the canal during
the test and Stability (in “5) of that measurement during the test. The lower panel automatically computes
the test time from the initial stimulus onset to the end of the test.
242 J. T. J acobson, C.A. J acobson /Inr. J. Pediatr. Otorhinolaryngol. 29 (1994) 235-248
Test Operating Characteristics
I I 1
Sensitivity Speciflcily
50% 52.3%
Fig. 3. Decision matrix analysis. TEOAE operating characteristics are based on comparative ABR test
results for the same group of 224 ears. Sensitivity (the ability of a test to correctly identify patients with
hearing loss); Specificity (the ability of a test to correctly identify those with normal hearing). TP (true-
positive), FP (false-positive), FN (false-negative), TN (true-negative).
provides excellent evidence of the presence of a TEOAE. In this example, the record-
ed peak stimulus value was 83 dB SPL with a mean noise level of 42 dB SPL.
When test results were submitted to a 2 x 2 decision analysis matrix using ABR
as the pass-fail standard, TEOAEs resulted in sensitivity and specificity values of
50% and 52%, respectively (Fig. 3), i.e., half (4/8) of the ears which failed the ABR
hearing screening successfully met emission screen criteria resulting in a 50% false-
negative rate. In comparison, 103 ear TEOAE screen failures were passed by ABR
at 35 dB (47.7% false-positive rate).
n Serlesl peak levels
q series 2 mean levels
rneen paok level: 51.7 dB SPL
mesn level: 45.3 dB SPL
0 50 100 150 266
Recorded Mean Noise Levels/Infant Ear
Fig. 4. A plot of the mean and peak noise values for 202 ears tested as calculated by the IL088
Otodynamic Analyser. The horizontal line represents the average noise value in dB SPL in the ear canal
at the time sweeps were accepted (51 dB SPL). See text for a description of suggested acceptable noise
levels for TEOAE neonatal screening.
J.T. Jacobson, C.A. Jacobson/Inr. J. Pediair. Ororhinoiaryngoi. 29 11994) 235-248
The mean and peak SPL noise levels for right and left ear scores of all infants with
recordable TEOAEs (n = 202) regardless of pass/fail status were 45.3 and 51.7 dB
SPL, respectively. Individual mean and peak levels in dB SPL for this group are
presented in Fig. 4. Of those infants with measurable TEOAEs, 57% exceeded the
mean dB SPL value.
Because duration is a critical component of a screening test protocol, the actual
time to complete each screening technique was noted. The mean duration of ABR
testing (including preparation) was 26.3 min/2 ears, with a range of lo-45 min.
Although mean recording time for TEOAEs was less than 3 min/2 ears (time is
automatically recorded by the ILO88), the actual mean time to obtain a response due
to noise, myogenic artifact, relocation of probe tip, etc., was 16.6 min/2 ears with
a range from 7-45 min. Parenthetically, it should be noted that a maximum 45 min
time period was allocated for test screening. In the authors’ opinion, 45 min is far
in excess of a reasonable and cost-effective period for newborn hearing screening.
However, for this study, test time was lengthened to 45 min in order to accomplish
procedural objectives.
4. Discussion
Mass hearing screening is not only desirable but was recently recommended by the
NIH at the Consensus Development Conference Statement of Early I dentification of
Hearing Zmpairmenr in I nfants and Young Children [34]. Additionally, NIH has
recommended EOAE as the preferred model for universal hearing screening. The
basic cochlear emission test has the potential to accommodate several major screen-
ing criteria and as such, TEOAEs provide a logical tirst level entry to hearing screen-
ing. For instance, the technique is fast, noninvasive and objective. Furthermore, the
application of modern microprocessing technology has made TEOAE devices ac-
cessible for clinical investigation.
It is difficult to compare the pass-fail rates of the present study to others since
variables such as population, noise conditions, location of testing site, equipment,
stimulus presentation level, software and failure criteria generally differ. For exam-
ple, several investigators [3,4,6,7,10,12,21,28,38,50] who support the contention that
TEOAEs are a reliable and valid measure of cochlear function in the human infant
all have reported results based on TEOAE testing in a sound-treated chamber or in-
Clearly, previous findings using TEOAEs in infants under controlled noise condi-
tions report higher success rates then those demonstrated from our results. Our re-
sults suggest that noise floor levels affect pass/fail test outcome. Ambient noise levels
in typical NICU environments and infant isolettes have been previously reported
[1,11,13,29,33]. In general, ambient noise in the NICU ranged from 50 to 75 dBA
SPL. In concert with life-support systems, isolette noise tends to produce greater
noise levels than extraneous hospital environments. Recently, Lonsbury-Martin and
Martin (pers. commun., 1993) obtained in-the-canal measurements of the frequency
response to ambient noise levels from adults and neonates in a procedure room at-
tached to a normal newborn nursery. It was clear from the resulting spectral analysis
244 1. T. J acobson, C. A. J acobson / I ni. J. Pediatr. Oforhinolaryngol. 29 (1994) 235-248
that under 1000 Hz, the infant ear had essentially little attenuation of room noise
and did not approach adult values until above 3000 Hz.
An alternative to unacceptable environments is the supplementary use of un-
powered isolettes in quiet procedure rooms, a technique that has been successfully
adopted by the Rhode Island Hearing Assessment Project [51] which currently
reports the findings of TEOAE testing in over 17 000 infants. This project enlists
several technical support personnel whose responsibility it is to transfer infants from
one setting (crib) to another (isolette) and perform the test, They claim that TEOAE
testing takes about 16 min for 2 ears under these conditions. However, White et al.
[52] also reported a 30% failure rate when infants were tested within the first 24 h
of life. The failure rate decreased to 18% when infants were tested between the 3rd
and 4th day following birth. This may have serious implications for TEOAE mass
screening with current hospital emphasis on early parent/infant discharge and incen-
tives from some insurance carriers.
As stated, excessive noise levels may have a detrimental affect on TEOAE testing.
Stevens et al. (1990) have reported that the combination of ambient and physiologic
noise levels in excess of 30 dBA SPL adversely effect TEOAE recordings. In fact,
they recommend that TEOAE screening should only be conducted under controlled
noise conditions. Kemp et al. [19] have indicated that noise levels greater than 45
dB SPL may prevent data collection. In our study, a total of 22 infants could not
be tested despite repeated attempts to evoke an emission over a 45 min period. The
remaining 85 ears which evoked an emission, but failed to meet the 50% replication
criterion, were associated with a mean noise level of 47.4 dB SPL. Of the 85 failures,
67 (78.8%) had dB SPL values that exceed the recommended value Kemp et al. [19]
of 45 dB and all failures exceeded the Stevens’ recommendation.
To the best of our knowledge, infant status was not a factor contributing to failure
rates since testing was conducted when an infant was asleep or resting quietly. Fur-
thermore, infants were tested in a stable physical condition, immediately prior to dis-
charge. In the eight ABR failures, there was no evidence of central nervous system
damage as determined by examination, imaging techniques or conventional ABR
central conduction latency values. Although several variables may have affected the
present TEOAE test results, it is our opinion that a combination of physiologic and
ambient noise levels, influenced by probe lit and leakage into the canal and through
ear-canal tissue during testing, may have been sufficient to interfere with data acqui-
The requirements of a screening test are perhaps the most demanding of all evalu-
ative procedures to satisfy. For example, acceptance of a screening test is heavily
predicated on its ability to correctly detect disease, or in this case, the presence of
auditory deficits. Thus, the sensitivity of a test must be great or infants who are in
fact hearing-impaired will be discharged under the false assumption that hearing is
within normal limits. If a test cannot substantiate an accurate outcome, it will result
in diverting resources away from the hearing-impaired rather than focusing attention
on them. Therefore, a good screening test should yield a very low percentage of false-
positive and false-negative rates, even under less than ideal conditions. A total of 4
of the 8 ABR ear failures passed TEOAE screening.
J. T. Jacobson, C.A. Jacobson /Int. 1. Prdiatr. Otorhinolarygol. 29 ( 1994 J 235-248
The operating characteristics demonstrated in this infant study address the ap-
plications of TEOAE testing in high risk screening environments where noise levels
are greater than those advocated in the literature. To our knowledge, no high volume
at-risk neonatal hearing screening program has the personnel or luxury to conduct
testing in a controlled sound booth, while still demonstrating administrative and
financial responsibility. Although an idea1 model, the concept of moving infants
away from a hospital unit for the purposes of screening is fraught with medical and
logistic complications that are well beyond the boundaries of this article.
The results of this study are not intended to refute the claims of a high prevalence
of cochlear evoked emissions in neonates or infants. Rather, we point out that the
application of TEOAEs for infant screening under less than idea1 environmental
conditions has failed to confirm previously reported high confirmation rates. As
stated earlier, the literature is uncontested in its support of measurable emissions in
infants. In the largest TEOAE infant series reported to date, Stevens et al. [43-461
screened 723 neonates, ranging in gestational age from 24 to 42 weeks (mean, 34
weeks) and tested at a mean of 37.5 weeks PCA. Recordable TEOAEs were obtained
in 80% of NICU graduates and 96% in full-term infants. Sensitivity and specificity
values of 93% and 84%, respectively, were realized. As the Stevens’ study correctly
conveys, TEOAE test outcome is consistent with numerous reported pass/fail tin-
dings in hearing screening programs using ABR as the recording instrument. The
significant difference between the present study and the Stevens collection is the test-
ing environment. All infants in the Stevens’ investigation were tested in sound suites.
AS Stevens states, ‘Background noise in the sound proof room with the equipment
running was a maximum of 28 dBA SPL and the microphone noise was 27 dBA SPL’
(Ref. 19, p. 129). Our contention is that in hospital-room settings, we were unable
to duplicate the same high success rates as reported by several investigators which
tested neonates and infants under more controlled noise conditions.
Kemp and Ryan [20] acknowledged a wide diversity in neonatal otoacoustic
responses and suggested ‘...that the term normal embraces a wide range of emissions
strengths, durations, frequency spectra etc.’ (p. 83). They advocate the need for stan-
dards which will take into account the interaction of noise and probe design on the
final test result. For instance, the pass/fail reproducibility criterion is only one possi-
ble method of analysis. Bonlils et al. [3] advocated the use of three criteria. They
include: EOAE waveform repeatability; nonlinear saturating responses to high levels
of stimulation; and specific frequencies of the response in the FFT analysis. Prieve
[39] also supports a testing strategy which could be based on reaching some criterion
correlation in active infants as well as the use of the frequency domain. We also en-
courage the establishment of a well-defined standard from which reliable com-
parisons between studies could be demonstrated.
Finally, other emissions including the use of the distortion product in the infant
population must be clinically evaluated to determine its screening application.
Whereas reports on the distortion product have shown encouraging results in adults,
246 J. T. J acobson, C.A. J acobson / 1~. J. Pediarr. Otorhinolarygol. 29 (I 994) 235-248
the clinical extension to the newborn has seen little reported attention to date
[4,27,28]. Although these recommendations are certainly not to be construed as
inclusive, they represent a starting point from which we will better serve the future
of infant hearing screening.
6. Acknowledgments
This study was supported in part by the Pennsylvania Lions Hearing Research
Foundation, No. 803. We would also like to thank Regina Hellstrom for technical
assistance in data collection.
7. References
Bess, R.H., Peek, B.F. and Chapman, J.J. (1979) Further observations on noise levels in infant in-
cubators. Pediatrics 63, 100-106.
Bonfils, P., Avan, P., Francois, M., Marie, P., Trotoux, J. and Narcy, P. (1990) Clinical significance
of otoacoustic emissions: a perspective. Ear Hear. 11, 155-158.
Bonfils, P., Dumont, A., Marie, P., Francois, M. and Narcy, P. (1990) Evoked otoacoustic emissions
in newborn hearing screening. Laryngoscope 100, 28-33.
Bonlils, P., Francois, M., Avan, P., Londero, A., Trotoux, J. and Narcy, P. (1992) Spontaneous and
evoked otoacoustic emissions in pretenn neonates. Laryngoscope 102, 182-186.
Bontils, P., Uziel, A. and Narcy, P. (1989) The properties of spontaneous and evoked acoustic emis-
sions in neonates and children: a preliminary report. Arch. Otorhinolaryngol. 216, 249-251.
Bonlils, P., Uziel, A. and Pujol, R. (1988) Screening for auditory dysfunction in infants by evoked
oto-acoustic emissions. Arch. Otolaryngol. Head Neck Surg. 114, 887-890.
Bray, P. and Kemp, D.T. (1987) An advanced cochlear echo technique suitable for infant screening.
Br. J. Audiol. 21, 191-204.
Browell, W.E. (1983) Microscopic observation of cochlear hair cell motility. Stand. Electron
Microsc. 3, 1401-1406.
Elberling, C., Parbo, J., Johnsen, N.J. and Bagi, P. (1985) Evoked acoustic emissions: clinical appli-
cation. Acta Otolaryngol. Suppl. 421, 77-85.
Dolhen, C., Hennaux, C., Chantry, P. and Hennebert, D. (1991) The occurrence of evoked oto-
acoustic emissions in a normal adult population and neonates. Stand. Audiol. 20, 203-204.
Gottfried, A.W., Wallace, L.P., Sherman, B.S., King, J., Coen, C. and Hodsman, J.E. (1981) Physi-
cal and social environment of newborn infants in special care units. Science 214, 673-675.
Jacobson, J,. Jacobson, C. and Spahr, R. (1990) Automated and conventional ABR screening
techniques in high-risk infants. J. Am. Acad. Audio]. 1, 87-95
Jacobson, J.T. and Mencher, G.T. (1981) Intensive care nursery noise and its influence on newborn
hearing screening. Int. J. Pediatr. Otorhinolaryngol. 3, 45-54.
Johnsen, N.J., Bagi, P. and Elberling, C. (1983) Evoked acoustic emissions from the human ear. III.
Findings in neonates. Stand. Audiol. 12, 17-24.
Kemp, D.T. (1978) Stimulated acoustic emissions from the human auditory system. J. Acoust. Sot.
Am. 64, 1386-1391.
Kemp, D.T. (1980) Toward a model for the origin of cochlear echoes. Hear. Res. 2, 533-548.
Kemp, D.T. (1988) Developments in cochlear mechanics and techniques for noninvasive evaluation.
Adv. Audiol. 5, 27-45.
Kemp, D.T., Bray, P.J,. Alexander, L. and Brown, A.M. (1986)Acoustic emission cochleography-
practical aspects. Stand. Audio]. Suppl., 25.
Kemp, D.T., Ryan, S. and Bray, P. (1990) A guide to the effective use of otoacoustic emissions. Ear
Hear. 1 I, 93-105.
Kemp, D.T. and Ryan, S. (1991) Otoacoustic emissions tests in neonatal screening programmes.
Acta Otolaryngol. Suppl. 482, 73-84.
J . T. J acobson. C.A. J acobson /Int. J . Pediarr. Otorhinoluryn,yol. 29 c 1994) 235-24X
Kennedy, D.T.. Crimm, L., Dees, D.C., Evans, P., Hunter, M., Lenton. S. and Thornton. R.D.
(1991) Otoacoustic emissions and auditory brainstem reponses in the newborn. Arch. Dis. Child. 66,
Kileny, P.R. (1987) Algo-l automated infant hearing screener: preliminary results. Semin. Hear. X.
Kileny. P.R. (1988) New insights on infant ABR hearing screening. Stand. Audiol. (Suppl. 30).
Kim, D.O. (1980) Cochlear mechanics: implications of electrophysiological and acoustical observa-
tions Hear. Res. 2, 297-317.
Kok, M.R., van Zanten, GA. and Brocaar. M.P. (1992) Growth of evoked otoacoustic emissions
during the first days postpartum. Audiology 3 I, 140- 149.
Kruger, B. (1987) An update on the external ear resonance in infants and young children. Ear Hear.
8, 333-336.
Lasky, R., Perlman, J. and Hecox, K. (1993) Distortion-product otoacoustic emissions in human
newborns and adults. Ear Hear. 13, 430-441.
Lafreniere, D., Jung, M.D., Smurzynski, J., Leonard, G.. Kim, D.O. and Sasek J. (1991) Distortion-
product and click-evoked otoacoustic emissions in healthy newborns. Arch. Otolaryngol. Head
Neck Surg. 117, 1382-1389.
Long, J.G., Lucey, J.F. and Phillips. A.G. (1980) Noise and hypoxemia in the intensive care nursery.
Pediatrics 65, 143-145.
Lutman, M.E., Mason, SM., Sheppard, S. and Gibbin, K.P. (1989) Differential diagnostic potential
of otoacosutic emissions: a case study. Audiology 28, 205-210.
McCall, S. and Ferraro, J.A. (1991) Pediatric ABR screening: pass-fail rates in awake versus asleep
neonates. J. Am. Acad. Audiol. 2, 18-23.
Margolis, R.H. and Shanks, J.E. (1990) Tympanometry: basic principles and clinical applications.
In: Rintelman, W. (Ed.), Hearing Assessment (2nd Edn.). Pro-Ed. Austin.
Mitchell, S.A. (1984) Noise pollution in the neonatal intensive care nursery. Sem. Hear. 5. 7-24.
National Institues of Health (1993) NIH Consensus Development Conference Statement, Early iden-
tification of hearing impairment in infants and children. Int. J. Pediatr. Otorhinolaryngol. 27,
Norton, S.J. and Widen, J.E. (1990) Evoked otoacoustic emissions in normal-hearing infants and
children: emerging data and issues. Ear Hear. I I. 121- 127.
Owens, J.J., McCoy, M.J., Lonsbury-Martin, B.L. and Martin, G.K. (1992) Distortion-product
emission measures in children with acquired hearing loss. Sem. Hear, 13, 53-66.
Peters, J.G. (1986) An automated infant screener using advanced evoked response technology. Hear.
Journal 39, 25-30.
Plinkert, P.K., Sesterhenn, G., Arold, R. and Zenner, H.P. (1990) Evaluation of otoacoustic emis.
sions in high-risk infants by using an easy and rapid objective auditory screening method. Eur. Arch.
Otorhinolarynol. 247, 356-360.
Prieve, B.A. (1992) Otoacoustic emmisions in infants and children: basic characteristics and clinical
application. Sem. Hear. 13, 37-52.
Probst, R., Coats, A.C., Martin, G.K. and Lonsbury-Martin, B.L. (1986) Spontaneous, click, and
tone burst evoked otoacoustic emissions from normal ears. Hear. Res. 21. 261-275.
Probst, R. (1990) New aspects ofcochlear mechanics and inner ear pathology. In: Pfaltz, C.R. (Ed,),
Advances in Oto-rhino-laryngology, Vol. 44. Karger, New York.
Robinette, M.S. (1992) Clinical observations with transient evoked otoacoustic emissions with
adults. Sem. Hear. 13, 23-36.
Stevens. J.C., Webb, H.D., Hutchinson, J., Connell, J., Smith, M.F. and B&in. J.T. (1989) Click
evoked otoacoustic emissions compared with brain stem electric response. Arch. Dis. Child. 64.
Stevens, J.C., Webb, H.D., Hutchinson, J., Connell, J., Smith. M.F. and Buffin, J.T. (1990) Click
evoked otoacoustic emission in neonatal screening. Ear Hear. I I, 128-133.
Stevens, J.C., Webb, H.D., Hutchinson, J.. Connell, J., Smith, M.F. and Buffin, J.T. (1991) Evalu-
ation of click-evoked oto-acoustic emissions in the newborn. Br. J. Audiol. 25. 1 I- 14.
248 J . T J acobson, C.A. J acobson /1n1. J . Pediarr. Otorhinolaryngol. 29 (1994) 235-248
Stevens, J.C., Webb, H.D., Smith, M.F., Buff’n, J.T. and Ruddy, H. (1987) A comparison of oto-
acoustic emissions and brain stem electric response audiometry in the normal newborn and babies
admitted to a special care baby unit. Clin. Phys. Physiol. Meas. 8, 95-104.
Stevens, J.C., Webb, H.D., Smith, M.F. and Bufftn, J.T. (1990) The effects of stimulus level on click
evoked oto-acoustic emissions and brainstem responses in neonates under intensive care. Br. J.
Audio]. 24, 93-300.
Thornton, A.R. (1978) Improved detection of auditory potentials. Asha. 20, 765.
Thornton, A.R. and Obenour, J. (1981) Auditory response detection method and apparatus. U.S.
Patent No. 4,275,744, 1981. Reviewed in J. Acoust. Sot. Am. 70, 1814.
Uziel, A. and Piron, J.-P. (1991) Evoked otoacoustic emissions from normal newborns and babies
admitted to an intensive care baby unit. Acta Otolaryngol. Suppl. 482, 85-91.
White, K.R. and Behrens, T.R. (1992) Early identification of hearing loss using transient evoked
otoacoustic emissions. Presentation to the Joint Committee on Infant Hearing, San Antonio, TX.
White, K.R., Maxon, A.B., Behrens, T.R., Blackwell, P.M. and Vohr, B.R. (1992) Neonatal hearing
screening using evoked otoacoustic emissions: the Rhode Island hearing assessment project. In: Bess,
F.H. and Hall, J.W., III (Eds.), Screening Children for Auditory Function, Nashville, Bill Wilkerson
Center Press.
Wilson, J.P. (1980) Evidence for a cochlear origin for acoustic reemissions, threshold line structure
and tonal tinnitus. Hear. Res. 2, 233-252.
Zwicker, E. and Schorn, K. (1990) Delayed evoked otoacoustic emissions - an ideal screening test
for excluding hearing impairment in infants. Audio]. 29, 241-251.