Brain Stem Evoked Response Audiometry in

Newborn Hearing Screening

Carol Schulman-Galambos, PhD, Robert Galambos, MD, PhD

\s=b\ Brain stem evoked response audiom- nique, brain stem evoked response was performed on 220 newborn infants in
etry (BERA) has been used as an auditory audiometry (BERA), has been devel¬ the nursery on the obstetrical ward of a
screening procedure in three groups of oped. The BERA technique permits service that has approximately 3,000 births
newborn infants. Group 1 consisted of 220 each year. Measurements were made only
normal-term infants who were tested with-
objective and precise measurement of on infants who were judged by the attend¬
the function of the peripheral audi¬
in 72 hours of birth; no hearing abnormal-
tory system. ing physicians to be normal, healthy, and
ities were uncovered (386 ears), and their full term. They ranged in postnatal age
threshold responses (for clicks) lay The BER, recorded from scalp elec¬ from 7 to 72 hours and in estimated gesta-
between a 10- and 20-dB hearing level (re: trodes, consists of a series of waves tional age from 38 to 42 weeks (S 39.8=

adult). Group 2 consisted of 75 newborns that reflect the activation of the weeks). The sample in which measure¬
who were treated in an intensive care unit eighth nerve and, successively, of ments were made represented about a
for one to 14 weeks; four were found to auditory brain stem structures fourth of the population of the nursery
have severe sensorineural hearing loss through at least the inferior collicu- during the time of the study. No a priori
(seven ears) at the time of discharge. lus.-'"4 When recorded at low-stimulus selection of these subjects was made; the
Group 3 consisted of a group of 325 technician made recordings on as many
infants, 1 year or older, who had
rates (one to ten per second), seven
available infants as she had time for
waves can be identified in adults;
previously been discharged from the when recorded at faster rates (30 to 45 during the hours she worked (six hours
same intensive care unit; of these infants, each day, four days each week).
an additional four showed severe sensori- per second as in the present studies),
neural hearing loss. All abnormalities that only wave V, according to Jewett and Procedure
were identified by BERA were subse- Williston's5 nomenclature, is regularly
quently confirmed by conventional au- seen. The principles and procedures for record¬
diometric measures. The estimate of an The latency of this wave V has been ing the BER, used in our laboratory, have
incidence of severe hearing loss in one of shown to vary systematically as a been described in detail elsewhere.'3 Mea¬
50 infants who required intensive care in function of stimulus intensity for surements were made on a clinical averag¬
the neonatal period calls for careful test- er (Nicolet model CA-1000), which was
animals and man within a given age
ing of this population. located in a quiet (but not sound-treated or
group, and as a function of increasing
(Arch Otolaryngol 105:86-90, 1979) maturation of the organism.'' " electrically shielded) room that was adja¬
cent to the nursery. Infants were scheduled
We have reported previously on the for testing only after feeding when they

The of the Conference

consensus
Newborn Hearing Screening,
on
which was held in San Francisco in
normal BER of infants and children7
and of premature infants," and de¬
scribed one year's experience with the
would be in natural sleep.
Recordings were made as already de¬
scribed," through surface electrodes that
1971,' was that no acceptably objective clinical application of the BERA were attached to the scalp at the vertex
and reliable technique for newborn method in a pediatrie population.'- and the active mastoid, with the contralat¬
hearing screening was then available. The application of this new method to eral mastoid grounded. Clicks of 0.1-ms
Since that conference, a new tech- the problem of newborn hearing duration were presented at a rate of 30 or
45 per second through an earphone that
Submitted for publication May 15, 1977; screening, both in a normal newborn was taped to the subject's ear. The electri¬
accepted Oct 10, 1977. population and in a special-risk popu¬ cal activity of the brain was amplified
From the Speech, Hearing, and Neurosensory lation of neonatal intensive care unit xlO', filtered between 150 and 1,500 Hz,
Center, Children's Hospital and Health Center,
and the Department of Neurosciences, Universi- (ICU) survivors, is the subject of this and responses to 1,000,2,000, or 4,000 clicks
ty of California, San Diego. communication. were averaged. Each measurement was
Deceased. replicated at least once. The latency of
Reprint requests to Speech, Hearing, and NORMAL NEWBORN SAMPLE
wave V in milliseconds was obtained by
Neurosensory Center, Children's Hospital and Subjects
Health Center, 8001 Frost St, San Diego, CA placing the computer's cursor on the peak
92123 (Dr Galambos). Brain stem evoked response audiometry of the wave and by reading off the digital-

Downloaded From: http://archotol.jamanetwork.com/ by a Monash University Library User on 06/17/2015

 Normal Newborn ICU Infant
ly displayed time.
The screening protocol called for mea¬
suring the response at a 60-dB hearing
level (HL) with one replication and at 30 dB
HL with one replication for one ear; if time 60 dB
permitted, recordings at 30 and 60 dB (also
with one replication) were made on the
second ear also.
In addition, the response threshold was
obtained on a series of 20 subjects; for 15 of
these subjects, measurements were made
at 60, 30, 20, and 10 dB and for the remain¬
ing five subjects, in 10-dB decrements
from 60 down to 10 dB. 30 dB
The following data were recorded for
each subject: sex, estimated gestational
age, age in hours when the BERA measure¬
ments were made, type of delivery (normal
vs cesarean section), latency of wave V for
each ear at each intensity measure, and
any comments that were relevant to the
test. Reliability of the test was rated 20 dB
"good," "fair," or "poor" on the basis of
how noisy the records were; noise came
from equipment (especially 60-Hz line
noise), from the patient (subject was, or
became, restless during the procedure), or
from the environment (occasional interrup¬
tions by the nursing staff in the normal
course of their running the newborn
10 dB
service). Approximately one hour was
required to obtain data from each patient,
half for the response measurements, and
half for obtaining parental permission,
Fig 1 .—Intensity series for two infants. Normal newborn (at left) showing wave V (arrows)
charting, and record keeping. down to 20 dB and absence at 10 dB. Infant from intensive care unit (ICU) with
gestational age of 37 weeks ready for discharge (at center); recording made with
RESULTS equipment (Nicolet CA-1000) as in normal newborn (at left). Recording (at right) of ICU
infant same as in (center) recording but using laboratory equipment. Duration of all
Measurements were made on a total recording was 20 ms.
of 386 ears. Sixty-five percent of these
tests were judged to have good relia¬ change of latency as a function of dB although test reliability was
bility, 20% were fair, and 15% were maturation is rapid in this period." "
judged good, ie, the subject slept
poor. Patient noise and equipment If one considers only data from those quietly throughout, and there was no
noise contributed about equally when normal newborns where reliability of 60-cycle interference. This patient
reliability was less than "good," with measurements was judged "good" and was referred to the Speech, Hearing
environmental noise interfering rare- who 39 to 40 weeks of estimated
were and Neurosensory Center, Children's
ly· gestational age, variability is consid¬ Hospital and Health Center, in San
Figure 1 shows a typical intensity erably reduced. These data are also Diego, for BERA retesting at 6 weeks
series normal newborn (at left).
on a shown in Fig 2 (at center). of age, at which time normal
Latency of wave V can be seen to For the subsample of 20 infants in responses were obtained bilaterally.
increase systematically as the intensi¬ whom threshold determinations were Therefore, this case must be consid¬
ty of the stimulus is decreased, as attempted, responses were obtained ered a false-positive outcome of the
expected.7 at 20 dB but not at 10 dB in all cases. screening procedure, probably the
Figure 2 shows the distribution of Figure 3 shows the latency-intensity result of a technical error that was
wave V latencies for the ears in function for all these ears. undetected during the initial testing.
normal newborns (at top) for the 60- Two subjects in this sample are of
and 30-dB stimuli. The variability in particular clinical interest. One was HIGH-RISK NEONATAL SAMPLE
this clinical sample, which is consider¬ the offspring of congenitally deaf Subjects
ably larger than we have found under parents; normal responses wereob¬ The Neonatal Intensive Care Unit of
laboratory conditions,7 was due in part tained bilaterally from this infant. Children's Hospital and Health Center, San
to the noisiness of some of the records, For the other infant, no response Diego (to which the Speech, Hearing and
and in part to the fact that the rate of could be obtained from either ear at 70 Neurosensory Center is attached both

Downloaded From: http://archotol.jamanetwork.com/ by a Monash University Library User on 06/17/2015

 60 dB HL 30 dB HL and for the remaining ten infants, in 10-dB
decrements from 60 down to 10 dB.
Normal Newborn ( =
386)
RESULTS
50
Figure (at center and right) shows
1
40 for a typical subject in this group, an
30 intensity series made with each of the
20 two instruments used in this study.
10 Morphologic differences between the
0 records are due to differences in the
Normal Newborn ( =
62) high- and low-frequency filtering
characteristics of the different am¬
50
40
plifiers, in the averaging computers
and in the properties of the X-Y plot¬
30 ters. Typical individual differences
20 between subjects because of variables,
10 such as the noise level of the EEG, are
0 lili j_ _ _L illustrated by a comparison of the
ICU Infants ( =
54) responses in Fig 1 (at center and
50 right).
Figure 2 shows the distribution of
40
latencies at 60 and 30 dB for infants
30
who were tested on discharge from
20 the ICU (at bottom), and Fig 3 shows
10 the latency-intensity curve for this
0
in r·» io co i^
population.
cr¡ .- > ö in co r·- ^ lo co r--
Threshold determinations yielded
ID i< od co co ·->
ni ( r-> co oó co öS cri
^ œ w m o * co (NIDO =!- co c\i id ó responses at 20 dB but not at 10 dB for
l¿ ID oó od cri IO i< od od öS öS oS
26 infants, at 30 dB but not at 20 dB
Latency of Wave V, ms for three infants, and at 40 dB but not
Fig 2.-Distribution of latencies at 60 dB HL (at left) and 30 dB HL (at right). All ears of at 30 dB for one infant. This last
normal newborns (at top) included those tested under "poor" conditions. Subsample (at result will be discussed further.
center) of all newborns had gestational age of 39 to 40 weeks, with "good" test reliability. Four cases in this sample of 75 are
Infants (at bottom), at discharge from intensive care unit (ICU), with presumed normal
of clinical interest. Three were at high
hearing and gestational age of 37 to 40 weeks. risk at the time of testing. One infant
physically and administratively), is a Procedure of a congenitally deaf mother was
regional center to which patients are Measurements were made in our clinical found to have no BER below 60 dB HL
admitted by transfer only, from a wide BERA laboratory with the use of the bilaterally. Two had been kernicteric
geographic area. Survivors are at risk for instrumentation and procedures that have on admission (serum bilirubin levels of
sensory and CNS disorders for a variety of been already described.'- The instrumenta¬ 38 mg/dL and 45 mg/dL, respective¬
causes, such as prematurity, problems of tion consists of commercially available ly); no BER was obtained for either
oxygénation, and the administration of components (TDH-39 earphone, Grason- ear at equipment limits (90 dB HL)
potentially ototoxic drugs. Stadler timers, click generator and atten¬ from one, while the other showed no
The subjects who were studied in this uator, Grass-Poll amplifier, Nicolet-1070
sample were 75 neonates who were housed response below 90 dB in one ear or
averager, and Hewlett-Packard X-Y plot¬
in the ICU. This sample included 54 below 80 dB in the other. The fourth
ter). The band-pass of the amplifier (Grass- infant was not obviously at risk for
patients who were presumed to have P511) was set at 300 to 1,000 Hz, and its
normal hearing, in whom routine BERA gain was x 10'. The infants were prepared hearing impairment and was actually
measurements were made; there were also for recording as described previously; they identified during our collection of
21 patients who were referred by the were in natural sleep at the time of testing, normative data. This patient had been
attending physician as at special risk for which was conducted in a double-cham¬ admitted for neonatal asphyxia. No
peripheral auditory problems as a result of, bered booth (IAC 1200 series) (the connect¬ BER could be obtained below 85 dB
for example, congenital familial deafness, ing door was left open so that the infants for the left ear or below 65 dB for the
high serum bilirubin levels, or long-term could be observed at all times).
right ear.
administration of potentially ototoxic Measurements were made at click inten¬
drugs. sities of 60 and 30 dB on one ear and at 30 Therefore, in this sample of 75
All patients were between 37 and 40
infants who were tested on discharge
dB (if time permitted, also at 60 dB) on the
weeks of gestational age (x 38.2 weeks)
=
second ear. from the ICU, four (5.3%) were shown
and between 1 and 14 weeks of postnatal In a subsample of 30 of these infants, an by BERA to have severely impaired
age, in open bassinets, and ready for attempt was made to estimate the thresh¬ hearing. These four infants later
discharge at the time the measurements old of response. For 20 of them, measure¬ received complete audiological work-
were made. ments were made at 60, 30, 20, and 10 dB, ups at which time the presence of

Downloaded From: http://archotol.jamanetwork.com/ by a Monash University Library User on 06/17/2015

 severe,irreversible, sensorineural loss
was independently confirmed, and
appropriate treatment was started in
each case.

HIGH-RISK FOLLOW-UP GROUP

These subjects belong to a group of ICU
survivors who participated in an ongoing
study that was designed to evaluate the
mental and motor outcome of disorders
that require intensive care.14 They are
included in this study because we have ICU Infants
8.0
identified four cases of peripheral hearing -

loss among them. The children in this

group are seen at six months, 12 months,
and subsequently at yearly intervals after
7.5
discharge. Their discharge diagnoses in¬ -

clude prematurity (less than 37 weeks of

gestational age), neonatal asphyxia (Apgar
score less than 5 at one minute and less
than 7 at five minutes, or requiring resus¬
citation in delivery room), respiratory
distress syndrome, and inappropriately low
weight for gestational age. Of these
infants, 325 who are 12 months of age or
older have been examined for language
competence. All who demonstrate signifi¬
cant language delay are screened by the
BERA test; if this is abnormal, they 10 30 40
receive a complete audiologic workup.
The manner in which the BERA test was Intensity, dB HL
applied was essentially that described Fig 3.—Latency-intensity functions for normal infants and adults. Differences in mean
previously for the infants in the high-risk latency in infant samples is assumed to be result of slight difference (1.6 weeks) in
neonatal group. Thus far, four of these gestational age.
children who demonstrated language delay
(three of whom are included in a already
published report'2) show severe sensorineu¬ headings: (1) the threshold of hearing difference between them according to
ral hearing impairment that requires (for clicks) in newborns, both the the BER measurements.
amplification. None of them is in the group newly born, and infants of the same According to Bradford,'' the new¬
of 27 who are common to studies 2 and 3 gestational but greater postnatal age; born threshold by respiratory audiom¬
(high-risk neonatal and follow-up groups), (2) the incidence of hearing loss in a etry is at 0 dB (re: ANSI-1969) at 1,000
which argue that infants who passed the
BERA test on discharge from the ICU are
high-risk population of infants; (3) the and 3,000 Hz. By contrast, the heart
unlikely to develop severe auditory impair¬ feasibility of screening newborns rate method estimated the threshold
with the use of BERA. for 500 and 3,000 Hz to be at 90 dB for
ment thereafter. No specific cause can be
newborns."1 Other methods (cribo-
assigned for three of the cases; two infants
weighed less than 1 kg at birth, and one Response Threshold to Clicks gram, behavioral) either do not
had been asphyxiated. The fourth infant attempt to establish threshold or yield
suffered fracture of the right temporal As shown in Fig 3, infants of about uncertain values for it. Though we
bone during a traumatic delivery; BERA 40 weeks of gestational age regularly have had no direct experience with
shows his right ear to be profoundly yield BERA responses to clicks of 20 respiratory audiometry, the ease with
impaired and, for some unknown reason, dB. This result, which is consistent which the BERA method yields reli¬
his left to be mildly impaired. with earlier reports based on some¬ able values certainly makes it an
When these four infants are added to the what smaller samples,7" means that
four who were identified in the high-risk outstanding candidate for those inter¬
the auditory threshold of the newborn ested in auditory threshold determi¬
neonatal group, a total of eight in the
lies below 20 dB, and so must be close nations in newborns and young chil¬
high-risk population of 373 have been
shown to have significant hearing impair¬ to that of adults. From our compari¬ dren.
ment. This is an incidence of 1.0:46.6 or sons of the properties of adult and As a practical matter, we do not
2.14%. newborn BERAs as the acoustic stim¬ routinely attempt to obtain the BERA
uli approach threshold in strength, we threshold, since the response is small,
COMMENT
estimate the infant threshold (for relatively unstable, and time-consum¬
The data from these studies will be clicks) to be about 10 dB above the ing to obtain. For this reason, we stop
discussed under the following three adult. There is, in any event, no large with the 30-dB response that, if it

Downloaded From: http://archotol.jamanetwork.com/ by a Monash University Library User on 06/17/2015

 displays a clearly defined wave shape the first postnatal day of life. Even criterion of the conference for use in
and a wave V latency appropriate for though the stimuli that were used, ie, the normal newborn nursery, where
age, is considered to indicate essen¬ clicks, are not optimal for testing at¬ the expected incidence of hearing loss
tially normal cochlear function. If an tenuation at the low frequencies, is about 1:2,000.
infant shows no response at 30 dB but which such transient middle ear inclu¬ However, when one considers the
does so at 40 dB, as did one of those in sions would introduce, the results population of neonatal ICU survivors
the high-risk neonatal group, we support Keith's findings17'1* of normal who are described in this study, a
suspect the infant to have a mild middle ear function as measured by different picture emerges. According
conductive problem. The infant in this impedance audiometry in the new¬ to our estimate, the incidence in this
group with a threshold at 40 dB would, born. population approximates one in 50.
by this criterion, be regarded as We recognize, of course, that this may
suspect for auditory problems. Hearing Loss in a not be the true incidence in this popu¬
A comparison of the BERA findings
High-Risk Population
lation, given the limited size of our
for infants who are similar in gesta¬ In groups 2 and 3, we examined a sample and the way it was selected.
tional age but different in postnatal population of 373 premature infants, However, the number of infants who
age yields several findings of interest. all of whom had been at risk for are identified and securely docu¬
Thus, Fig 3 shows the wave V latency hearing disorder (among other possi¬ mented as having severe to profound
for the normal newborns to be slightly ble abnormalities), and we identified sensorineural loss in our sample
but consistently shorter than for ICU by BERA eight with severe peripheral argues for a probable high incidence
infants. This difference can be attri¬ losses that required amplification. in this population. For this group, the
buted in part at least (and probably This incidence, 2.14%, is to be com¬ BERA seems to be an efficient screen¬
entirely) to the difference in their pared with the zero incidence in our ing test, and its cost seems to be
mean gestational ages (the normal sample of 220 normal newborns. It justified to us. Therefore, we have
newborn mean is 1.6 weeks older). would appear desirable to examine a recommended to our hospital adminis¬
However, the maturational decrease much larger sample of at-risk infants tration that they institute routine
in wave V latency has been estimated to ascertain how valid the surprisingly BERA testing on all ICU infants prior
at 0.3 to 0.5 ms per week in this high 2.14% estimate may be for this to discharge.
period,7811 and since in our sample, group.
the decrease is somewhat smaller than study was supported by grants HD-10636-
This
Feasibility of BERA for 01 (Dr Schulman-Galambos) and NS11707 (Dr
expected (about 0.1 ms), the question Newborn Galambos) from the University of California, San
of whether auditory maturation pro¬ Screening Diego.
The authors thank Jan Labar, MA, for assem¬
ceeds somewhat more rapidly in As these data show, BERA unques¬ bling the data on the newborn population; the
prematurely born infants is a possibil¬ tionably meets the criteria for an parents of these newborns; the staff of Sharp
ity. Though this seems unlikely, the objective and reliable method for Hospital, San Diego, where the measurements
were made; Alan Schumacher, MD, Morton
question cannot be answered with the assessing newborn hearing according Cohen, MD, the staff of the Intensive Care Unit
present data. to the recommendations of the Con¬ at Children's Hospital, San Diego, and the
Finally, the data from the normal ference Newborn Hearing Screen¬
on parents of the infants who were tested there.
Kristen Gist, MA, took care of the infants in the
newborn sample indicate the absence ing.1 However, the complexity and high-risk follow-up group; Donald Krebs, PhD,
of significant conductive hearing loss cost of the test, requiring as it does and Robert Sandlin, PhD, provided space and
equipment used in testing; and the Nicolet
due to unabsorbed mesenchymal tis¬ expensive instrumentation and a Instrument Co, Madison, Wis, loaned the clinical
sue or fluid in the middle ear during trained technician, does not meet the averager (model CA-1000).

References

1. California State Department of Public 115, 1972. in children. Arch Otolaryngol 103:38-43, 1977.
Health: Conference on Newborn Hearing Screen- 7. Hecox K, Galambos R: Brain stem auditory 13. Schulman-Galambos C, Galambos R: Re-
ing: Proceedings Summary and Recommenda- evoked responses in human infants and adults. cording the Brainstem Evoked Response: A
tions. San Francisco, Feb 23-25, 1971. Arch Otolaryngol 99:30-33, 1974. Manual, 1975 (mimeograph).
2. Starr A, Achor J: Auditory brainstem 8. Schulman-Galambos C, Galambos R: Brain 14. Schulman-Galambos C, Cohen M, Gist K:
response in neurological disease. Arch Neurol stem auditory-evoked responses in premature Developmental followup of neonatal ICU survi-
32:761-768, 1975. infants. J Speech Hear Res 18:456-465, 1975. vors. Proc Soc Pediatr Res, 1976.
3. Starr A, Hamilton A: Correlation between 9. Salamy A, McKean CM, Buda FB: Matura- 15. Bradford LJ: Respiration audiometry, in
confirmed sites of neurological lesions and abnor- tional changes in auditory transmission as Bradford LJ (ed): Physiological Measures of the
malities of far-field auditory brainstem response. reflected in human brain stem potentials. Brain Audio-Vestibular System. New York, Academic
Electroencephalogr Clin Neurophysiol 41:595-608, Res 96:361-366, 1975. Press Inc, 1975, pp 249-317.
1976. 10. Salamy A, McKean CM: Postnatal develop- 16. Schulman CA: Heart rate audiometry: I.
4. Stockard JJ, Rossiter VS: Clinical and path- ment of human brainstem potentials during the An evaluation of heart rate response to auditory
ologic correlates of brainstem auditory response first year of life. Electroencephalogr Clin Neuro- stimuli in newborn hearing screening. Neuropae-
abnormalities. Neurology 27:316-325, 1977. physiol 40:418-426, 1976. diatrie 4:363, 1973.
5. Jewett D, Williston J: Auditory-evoked far- 11. Starr A, Amlie R, Martin W, et al: Develop- 17. Keith RW: Impedance audiometry with
fields averages from the scalp of humans. Brain ment of auditory function in newborn infants neonates. Arch Otolaryngol 97:465-467, 1973.
94:681-696, 1971. revealed by auditory brainstem potentials. Pedi- 18. Keith RW: Middle ear function in neo-
6. Jewett DL, Romano MN: Neonatal develop- atrics 60:831-839, 1977. nates. Arch Otolaryngol 101:376-379, 1975.
ment of auditory system potentials averaged 12. Mokotoff B, Schulman-Galambos C, Gal-
from the scalp of rat and cat. Brain Res 36:101- ambos R: Brain stem auditory evoked responses

Downloaded From: http://archotol.jamanetwork.com/ by a Monash University Library User on 06/17/2015