Using Contemporary ABR Protocols to Get Accurate Results : The Hearing Journal

Journal Logo

TOT 10

Using Contemporary ABR Protocols to Get Accurate Results

Smith, Joanna T. MS; Wolfe, Jace PhD

Author Information
The Hearing Journal 67(5):p 36,38-40, May 2014. | DOI: 10.1097/01.HJ.0000449904.87362.60
  • Free
Figure 1:
It is best to conduct ABR assessment while the child sleeps in a basinet/baby bean bag or car seat. © iStock/McIninch

Have you ever shopped around for a new car? It can be a maddening experience, with different dealerships offering you different prices, on different models, with different extras.

Joanna T. Smith, MS

But is it possible that a family who took a new baby with hearing loss to three different audiology clinics would have a similar experience? We think it is, because audiology clinics across the United States are likely to use a variety of protocols for auditory brainstem response (ABR) assessment.

Jace Wolfe, PhD

Although we don't have the comprehensive, evidence-based ABR protocols that they do in Great Britain, Australia, and the Canadian provinces, we can refer to these protocols, as well as to the guidelines of our professional organizations and contemporary manuscripts and textbooks.

This month's column addresses ABR acquisition in infants, while next month's column will discuss interpretation.

We dedicate these columns to a handful of experts who are our mentors in auditory electrophysiologic assessment, including, but not limited to, David Stapells, Jay Hall, Guy Lightfoot, Michael Dennis, Richard Talbott, Yvonne Sininger, Linda Hood, Chuck Berlin, and Michael Gorga.


One of the most important factors in obtaining accurate ABR results is reducing interfering noise as much as possible. The biggest culprit for noise in evoked potential testing is myogenic (muscle) noise, and the best way to attenuate that noise is to test the infant while she sleeps.

Since most infants sleep frequently during the first few months of life, it is imperative that we schedule diagnostic assessment as quickly as possible, ideally giving a time slot around the baby's nap time.

During scheduling, we instruct the family to keep the baby awake for at least two hours prior to the assessment and to withhold feeding until right before testing. It is also important to have a second adult sit next to the baby's car seat during the drive to the appointment to keep the baby awake.

When the baby arrives, we attempt to conduct acoustic immittance and otoacoustic emission testing prior to ABR. If the infant becomes very agitated during this testing, we move onto ABR assessment.

We apply the electrodes before the parents feed the baby. Once the electrodes are attached and impedance is satisfactory, we darken the room.

We then ask the parent to make certain the child has a dry diaper, and if so, feed the baby.

Although opinions differ, we believe it is ideal to keep electrode pads taped to the child's head but remove the snap-on electrode leads until it is time to begin testing so that they are not pulled during feeding.

We also believe that it is best to test the child while she sleeps in a basinet/baby bean bag or car seat, which allows us to check electrodes and insert earphones easily.

Experience has suggested that artifact may arise from the parent holding the child. For children who are only comforted when they are close to their mother, though, it is likely preferable to have the mother hold the baby.


If the right techniques are used, excellent electrode impedance and adherence may be achieved with little or no discomfort for the infant.

We start by using an alcohol swab to clean each electrode site, which helps with the removal of skin oils. Then, we rub off any oils or excess skin with a cotton swab and alcohol pad. Next, we use a cotton swab to scrub each site vigorously with an abrasive liquid.

Figure 2:
Since bone-conduction (BC) headbands are made for adults, the authors prefer to hold the headband on a baby.

We recommend using disposable electrode pads, which are quick to apply and sanitary. Although these electrode pads have an adhesive underside, we also place surgical tape in an “x” over each pad.

The electrode leads should be positioned as far away as possible from any stimulus cables. It may be helpful to braid the electrodes. We recommend a four-electrode montage to allow for two-channel recordings.

The non-inverting electrode should be placed at midline, as close as possible to the baby's hairline. Inverting electrodes should be placed as low as possible on the mastoid bone, just posterior and inferior to the auricle.

The ground electrode may be placed on the lower forehead, at least 3 to 4 cm from the non-inverting electrode.

The absolute electrode impedance should be no higher than 3 kilohms, and the difference in impedance between any two electrodes should not be greater than 1 kilohm.


We like the approach of the Ontario Infant Hearing Program's Audiologic Assessment Protocol and the British Columbia Early Hearing Program's Diagnostic Audiology Protocol, which target a behavioral hearing threshold of 30 dB HL or greater.

Based on a landmark study by Dr. Stapells and colleagues, the Ontario and British Columbia hearing programs determined that the following frequency-specific correction factors are good predictors of behavioral threshold: a) 500 Hz = 10 dB, b) 1,000 Hz = 10 dB, c) 2,000 Hz = 5 dB, and d) 4,000 Hz = 0 dB ( Ear Hear 1995;16[4]:361-371

For the target hearing loss of 30 dB HL or poorer, the desired ABR threshold is 25 dB eHL plus the correction factor. At 2,000 Hz, the correction factor is 5 dB, so the lowest level tested for infant ABR assessment is 30 dB nHL.


Without an American National Standards Institute (ANSI) standard describing calibration of tone-burst evoked potentials, we once again turn to the experts in Canada.

The Ontario and British Columbia infant hearing program guidelines provide values for acoustic calibration at 0 dB nHL. These values are frequency and transducer specific, and are based on the Stapells et al data used to generate the frequency-specific correction factors.


Most evoked potential systems allow the clinician to review the ongoing EEG without averaging the ABR response.

If the ongoing EEG is noisy, then the clinician should wait for the child to fall into deeper sleep or relaxation. If several minutes pass and the EEG activity does not change, then the clinician should reposition the child to reduce tension in and around the head and shoulders.

Also, if the parent is holding the baby, it often helps to position the baby on a pillow resting on the parent's lap. If noise persists, the clinician should check to make certain that electrode impedance values are still acceptable.

According to most contemporary protocols, an ABR waveform should only be labeled “no response” if the residual noise after averaging is less than 20 to 25 nV.

Filtering and the artifact reject system are also effective means to reduce electrical noise. The large negativity following wave V of the infant ABR may extend beyond 20 msec.

If we err on the side of caution and say that the latest response we might wish to measure could fall within 20 to 25 msec, we would have a response that repeats approximately every 25 msec.

If we divide 1,000 msec (1 sec) by a response window of 25 msec, we see that the slowest potentials may repeat around 40 times per second.

With that in mind, the high-pass filter should be placed lower than 40 Hz. The inter-peak latencies of waves I through V can be as short as 1 msec, so the low-pass filter must be set to at least 1,000 Hz.

Since there is not much energy in the infant ABR above 1,000 Hz, we use a low-pass filter cutoff of 1,500 Hz at Hearts for Hearing and a high-pass filter setting of 30 Hz.

Martin Hyde of the Ontario Infant Hearing Program has suggested that the artifact reject system and amplifier gain should be set so that approximately five to 10 percent of the sweeps are rejected while a baby rests quietly. At Hearts for Hearing, we have had good luck with using an amplifier gain setting of 100,000 and an artifact rejection setting of 10 µv.

Signal averaging is one of our best friends in the battle to reduce noise. During the first few hundred sweeps, the noise is reduced dramatically, with lesser reductions in noise achieved with further averaging.


If we select unproven, insufficient (garbage) ABR stimulus parameters to elicit the infant's response, then who knows what garbage we will actually record.

At Hearts for Hearing, we use an alternating polarity stimulus. Additionally, we have adopted a two-one-two rise-plateau-fall cycle duration with a linear window for signal duration, which should provide equal results to five-cycle exact-Blackman tones.

Further, we generally set the maximum number of recorded sweeps per waveform to 6,000 to allow for sufficient averaging near the threshold, if needed.


Legendary UCLA basketball coach John Wooden was fond of instructing his players to “be quick, but don't hurry.” The same can be said for ABR measures.

For a 25-msec window, the maximum stimulus rate that allows for a rapid acquisition of the response without overlapping responses to preceding stimuli is 40 stimuli per second. Since we want to avoid stimuli that will interfere with 60-Hz electrical noise, we generally use a stimulus rate of 39.1 tone bursts per second.

Of course, for high-frequency tone-burst responses, which possess shorter latencies, the stimulus rate could be significantly higher. For the sake of simplicity and uniformity, though, we use a stimulus rate of 39.1 tone bursts per second for all tone-burst frequencies.

We don't want to spend a lot of time running unnecessary sweeps if an obvious response exists. The British Columbia ABR protocol suggests a minimum of 1,000 total sweeps to conclude that a response is present or absent.


The work of Dr. Stapells, Dr. Gorga, and their colleagues shows that bone-conduction (BC) ABR can be reliably recorded to indicate an infant's permanent sensory hearing acuity. When the air conduction threshold suggests an estimated hearing level of 30 dB or poorer, we prioritize the attainment of a bone-conduction response.

Since bone-conduction headbands are made for adults, it is almost impossible to get one to fit well on the head of a baby. We prefer to hold the headband so that the BC stimulator is placed on the mastoid, just above and behind the auricle (see figure 2).

If the clinician does not wish to use handheld placement, then the BC stimulator can be retained with an elastic band (see the British Columbia protocol for details).

Susan Small and colleagues showed that bone-conduction results obtained with handheld placement are similar to those obtained with a headband ( Ear Hear 2007;28[1]:83-98

The dynamic range of bone-conduction stimulation is typically on the order of 20 to 30 dB for ABR testing. As a result, bone-conduction testing can be used to rule out a conductive hearing loss but cannot evaluate much more than a mild sensory component.

Additionally, two-channel recordings should be conducted to measure the bone-conduction response. This allows for a comparison of the response recorded in each channel to surmise the ear that is responding.


At Hearts for Hearing, our ABR “agenda” is a combination of the Canadian (British Columbia and Ontario) and British protocols, with some deviations from the former.

We typically begin auditory brainstem response testing with a 2,000-Hz tone burst presented at the stimulus level consistent with our minimum testing level (i.e., 30 dB nHL at 2,000 Hz) and obtain two responses.

If a repeatable response is present, we then borrow a page from the British Newborn Hearing Screening Program's Guidance for Auditory Brainstem Response Testing in Babies and go up in level by 10 dB to obtain two more waveforms, again looking for a high degree of repeatability.

Collecting waveforms 10 dB above the minimum testing level accomplishes two important objectives. First, if the higher-level waveforms are indeed obtained at a greater amplitude and earlier latency, they provide validation of the waveforms obtained at the minimum testing level.

Second, if there was any uncertainty about the validity of the response obtained at the minimum testing level, the attainment of an obvious response at a level 10 dB higher will limit any possible error in interpretation to no greater than 10 dB.

Clinicians who are inexperienced with auditory brainstem response testing or who do not possess a high level of confidence in their ABR interpretation skills may find it helpful to begin the ABR with a 2,000-Hz tone burst presented at 60 dB nHL for naturally sleeping babies and 70 dB nHL for sedated children.

This approach should result in a robust response for the normal-hearing infant (see figure 3).

Once testing is complete at 2,000 Hz for the first ear tested, we switch to the opposite ear and repeat the same testing.

If another normal response is obtained, we then test 500 Hz in each ear, starting at 35 dB nHL.

If an elevated response is obtained at any of the aforementioned frequencies, we obtain a bone-conduction threshold at that frequency. We then immediately switch to the opposite ear and test the same frequency as well as obtain a bone-conduction threshold if the opposite ear also possesses an elevated threshold.

This approach provides ear-specific results and also delineates whether the loss is conductive and in need of medical referral or sensorineural and also in need of amplification.

Finally, we assess 4,000 Hz for each ear, starting at 25 dB nHL. If no clear response is obtained at the starting levels of any of the aforementioned frequencies, then the presentation level is increased in 20- to 30-dB increments until a response is obtained or equipment limits are reached. Once a response is obtained, a 10-dB bracketing procedure is utilized.

As soon as we are able to confidently identify that a response is present, then we stop averaging and initiate a new waveform to obtain a repeatable response.

Figure 3:
Beginning the ABR with a 2,000-Hz tone burst presented at 60 dB nHL for naturally sleeping babies and 70 dB nHL for sedated children results in a robust response for the normal-hearing infant.

If the response is not clearly present at the onset of testing, then we average until we obtain a response or until we obtain a sufficiently low noise level without a present response.


Several years ago, we abandoned the routine use of the click stimulus in infant auditory brainstem response assessment. The tone-burst ABR described provides a frequency-specific assessment of auditory function with the means to define type of hearing loss as well.

Additionally, well-formed tone-burst ABR waveforms indicate good neural synchrony and integrity, especially when obtained at levels consistent with normal hearing. (Stapells DR. Frequency-specific threshold assessment in young infants using the transient ABR and the brainstem ASSR. In: Seewald R, Tharpe AM, eds. Comprehensive Handbook of Pediatric Audiology. San Diego, CA: Plural Publishing Inc; 2011:409-448.)

If ABR tone-burst responses are only present at levels consistent with a severe to profound hearing loss, then the clinician should consider measuring a click-evoked auditory brainstem response, following these recommendations:

  • Measure the click-evoked ABR with a 90-dB nHL click, presented with at least 1,000 sweeps of rarefaction clicks and 1,000 sweeps of condensation clicks.
  • If no responses are obtained, increase the presentation level in 10-dB steps until a response is identified or equipment limits are reached.
  • Compare the separate condensation and rarefaction waveforms to look for evidence of a cochlear microphonic (CM).
  • When the CM is present, clamp the earphone tubing and remeasure the response. If the previous response is truly a cochlear microphonic, then it will disappear when the tubing is clamped. If the activity in the waveform is actually a stimulus-related artifact, then the response remains.
  • Use a stimulation rate within the range of 11.1 to 19.1 clicks per second to attempt to elicit a neural response to a click. A filter setting of 30 to 1,500 Hz along with an analysis window of 10 to 12 msec should suffice.

Next month, we will discuss interpretation of the cochlear microphonic response in infants.

Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved.