We are now in the era of very early identification of hearing loss. With the advent of universal newborn hearing screening, it is common for audiologists to see infants less than 2 to 3 months of age who, during the newborn period, have been identified as being at risk for significant hearing loss. In order to select appropriate intervention strategies, the next step is to determine the severity, configuration, and nature (sensorineural, conductive, or mixed) of an infant's hearing loss.
The tone-evoked auditory brainstem response (tone-ABR) is currently the only measure that can reliably provide such information in young infants.* This article will (1) review some of the evidence supporting the use of tone ABRs to estimate the audiogram in young infants, (2) discuss how to use the tone-evoked ABR to distinguish sensorineural impairment from conductive loss, and (3) provide a detailed discussion of the “sequence of testing” (e.g., what intensity to start with, what frequency).
The last topic is critical. With so many choices (frequency, intensity, ear, and air vs. bone) and so little time (infants have a habit of waking up too soon), it is essential to use a test sequence that is fast, efficient, and provides the greatest increase in clinical information with each successive step. A detailed presentation of the recording and stimulus set-ups, calibration, and the rationale behind choices is beyond the scope of this article. More detailed reviews of our tone-ABR research, as well as our recommended parameters and protocols, can be found on our “Clinical ABR” web site (www.audiospeech.ubc.ca/haplab/clinic.htm) and in our recent review papers.1,2
EVIDENCE FOR THE ACCURACY OF TONE-ABR THRESHOLDS
The ABR to brief tones has been used successfully for threshold assessment for more than 30 years, since the first publications in the 1970s.3–6 Nevertheless, despite early and subsequent success, some clinicians erroneously believe that the tone-ABR does a poor job of predicting pure-tone behavioral thresholds, especially for low frequencies. This belief may stem from the few research articles that concluded that the tone-ABR has problems. However, most of these “dissenting” articles have technical problems and few present results for groups of individuals (i.e., they present only a few cases).
The great majority of research papers considering the tone-ABR for threshold estimation show reasonably accurate results. Certainly, our own research has consistently shown good threshold accuracy for the tone-ABR.7,8 To provide a broader view, we recently analyzed the tone-ABR threshold estimation literature.9 We assessed 32 studies providing tone-ABR threshold results from a total of 1203 individuals (524 adults, 679 infants and young children); 815 of these individuals had normal hearing sensitivity, and 388 had sensorineural hearing loss.
The results of this meta-analysis, summarized for infants and children in Table 1, show several things. First, threshold results across the 32 studies were quite consistent, with 95% confidence intervals no larger than ±5 dB. Second, tone-ABR thresholds in individuals with normal hearing are typically 10–20 dB nHL. Third, most infants and children with sensorineural hearing loss (SNHL) show tone-ABR thresholds that are within 10 dB of their pure-tone behavioral thresholds (depending upon frequency, see Table 1). And, contrary to what some clinicians believe, the results for 500 Hz are quite accurate and no more variable than those for 1000 and 2000 Hz. Clearly, when recorded and interpreted properly, the tone-ABR can provide a reasonably accurate estimate of the pure-tone behavioral audiogram in infants and young children.
TONE-EVOKED ABR FOR IDENTIFYING TYPE OF HEARING LOSS
As with behavioral audiometry with adults, comparison of ABR results obtained with air- versus bone-conducted tonal stimuli provides information about the nature (conductive, sensorineural, mixed) of the hearing loss. Although immittance results provide helpful information (particularly when normal), they cannot quantify the conductive component (e.g., a reduced middle ear mobility may be associated with a 5-dB or a 40-dB conductive component). Furthermore, reliable immittance results are often unobtainable in young infants. ABR measures, such as wave latencies, to air-conducted stimuli alone also cannot reliably indicate the presence of a conductive component or, especially, its degree.11 Thus, the only reliable method is to compare results for air- and bone-conducted stimuli.
Bone-conduction (BC) ABR results provide information that is critical for the management of an infant with an elevated threshold, as well as information that can be of considerable comfort to the infant's parents, should the elevation be conductive in nature. Because of its importance, bone-conduction testing should occur quite early in the sequence of ABR testing, usually immediately after both ears have been tested at an initial air-conduction (AC) “screening” intensity, and a “no response” for one or both ears obtained (see Figure 1).
Because of bone oscillator output limitations, stimulus intensities for bone-conducted brief tones are restricted to maxima of 45–60 dB nHL at 500 Hz and 60 dB nHL at 2000 Hz. This range nevertheless allows the audiologist to assess sensorineural loss up to about 50–60 dB HL and, importantly, to classify elevated air-conduction results as being from a normal (i.e., conductive loss) versus an impaired (i.e., sensorineural loss) cochlea.
Masking of the contralateral ear, typically required with bone-conduction testing in adults, is neither feasible nor necessary for ABR audiometry in young infants. Masking is not feasible because (1) effective masking levels for bone-conduction brief-tone stimuli in infants are not known and (2) time is too limited to record using several masking levels, as one might do in attempting plateau masking.
However, because of their immature skulls, young infants show substantial interaural attenuation of bone-conducted stimuli, as much as 25 dB.11 Thus, stimuli presented to the temporal bone at the low stimulus levels (20–30 dB nHL) required to demonstrate normal versus impaired cochlear function will stimulate primarily the cochlea ipsilateral to the oscillator placement.
Also, the laterality of ABR origin (i.e., which cochlea is reflected in the recorded ABR?) can be determined by using two-channel recordings and observing the large ipsilateral/contralateral wave V latency and amplitude asymmetries present in infants and young children (but not in older children or adults). Wave V is larger and earlier in the EEG channel ipsilateral to the stimulated cochlea.12,13 Thus, if one sees this pattern in the channel on the same side as the bone oscillator, one can infer that stimulation of the cochlea on the same side has resulted in the ABR; if it shows the opposite pattern, then the opposite cochlea has produced the response, and a sensorineural impairment is present.1,12 While the ipsi/contra technique is reasonably well-tested in infants with conductive loss,12 it requires further assessment in infants with sensorineural or mixed loss, although the author's clinical experience indicates reasonable results in these latter groups.
THE SEQUENCE OF TESTING
As stated previously, it is essential to use a test sequence that is fast, efficient, and provides the greatest increase in clinical information with each successive step.
Several principles guide the general strategy of stimulus conditions:
- ❖ Test time is limited. The infant may awaken at any moment, so the most important question must be answered first.
- ❖ The choice of stimulus condition should be based on the most probable outcome. For example, most infants coming to the diagnostic ABR stage after “refers” during the newborn period have normal hearing; starting at a low “screening” intensity will therefore quickly obtain the necessary results for most infants.
- ❖ The choice of stimulus condition should be based on obtaining results that make a difference in management as well as providing information to the parents. For example, when no response is present at the air-conduction screening intensity, spending time collecting precise air-conduction threshold information is less useful than obtaining bone-conduction results, because the air-conduction threshold associated with most infant conductive hearing loss is a “moving target” (i.e., it changes over time). Having information about the type of impairment directs subsequent management (including medical management) and provides more certain information for the family.
- ❖ Efficient strategies require clinicians to switch ears and mode (AC vs BC) of stimulation frequently; insert earphones should be placed in both ears at the beginning of testing and the bone oscillator should be ready for application. For example, after obtaining a “no response” at the air-conduction screening intensity in the first ear, the audiologist should switch to the other ear rather than seek threshold. Otherwise, one may spend time determining threshold for one ear only to have the infant wake up before it has been discovered that the other ear is normal (determined on a second ABR appointment). Obviously, it is better, both for management and for the family, to find out as soon as possible if at least one ear is normal.
Figure 1 depicts a tone-ABR test sequence flowchart based on the above principles. In this flowchart, “normal” response levels (levels, in dB nHL, at which a response is required for the result to be considered normal) are 30–40 dB for AC 500 Hz, 30 dB for AC 2000 Hz, 30 dB for BC 2000 Hz, and 20 dB for BC 500 Hz (see our recent reviews1,2).
It should be noted that the flowchart is incomplete. It does not show details such as intensity step size when searching for threshold, nor does it show details of testing for frequencies other than 2000 Hz or 500 Hz. For example, if one obtains a threshold for air-conducted 2000-Hz tones at 60–80 dB, more information is obtained by starting 500-Hz testing at 30–40 dB rather than at 60–80 dB. A 500-Hz response present at 30–40 dB indicates a sloping loss, whereas a 500-Hz response at 60–80 dB indicates only that the configuration is either flat or sloping.
Generally, intensity step sizes smaller than 20 dB are inefficient, although a final step size of 10 dB is appropriate if time permits. It is inefficient to use a 5-dB step size or test at levels below 20–30 dB nHL. Except perhaps in ototoxic monitoring, management of a 20–30 dB nHL “threshold” is unlikely to be different from that of a 10-dB nHL threshold).
The intensities tested should bracket threshold. For example, if no response is seen at 30 dB (2000 Hz), following up at 40 dB will give little information if there is no response at 40 dB. A better compromise is 60 dB. If both 30 dB and 60 dB at 2000 Hz have been tested (as well as BC 2000 Hz at 30 dB; see Figure 1) before the infant wakes up, then we know the following: (1) whether an impairment exists for one or both ears, (2) if a sensorineural component exists, and (3) whether the loss is mild/moderate (if 60 dB response is present) or more severe (if 60 dB is absent). The use of smaller step sizes, and following a sequence that does not switch ears (and mode), may result in the infant waking up before a clear picture of the status of both ears, as well as the type and severity of loss, has been obtained.
ABR assessment of young infants can be seriously compromised by auditory neuropathy and/or neurologic involvement. In such cases, especially of auditory neuropathy, ABR thresholds typically do not reflect cochlear sensitivity. As a rule, if a clinician sees a distinct wave that is clearly ABR wave V (to a brief tone of any frequency, whether AC or BC), then he can be reasonably confident that an elevated ABR threshold is not the result of auditory neuropathy or neurologic dysfunction. Note that if wave V is clear and the V/I amplitude ratio is normal, the finding of a prolonged wave I-V interpeak latency should not be interpreted as suggesting that any threshold elevation is secondary to neurologic dysfunction.
On the other hand, the absence of a clear ABR wave V in any waveform, even at the highest intensity, may be the result of profound peripheral (conductive and cochlear) impairment or auditory neuropathy/neurologic dysfunction. In such a situation (no tone-ABR response showing a clear wave V), the clinician should obtain recordings to high-intensity clicks (90–100 dB nHL, single polarity, 19/s rate). Of course, it is also important to obtain other measures of auditory responsivity (especially evoked otoacoustic emissions) to cross-check these ABR results. For cases of click-ABR waveforms of unusual morphology (e.g., only early waves present without wave V) or clearly present emissions with absent ABR, repeat ABR testing in a month or so may also be warranted (see Sininger in this issue).
What is the minimum information that should be sought from an ABR assessment? Assuming that results are deemed reliable and no neurologic component to the threshold elevation is suspected (i.e., a clear wave V is present), then, at a minimum, a “complete” tone-ABR evaluation should provide AC thresholds for 500 and 2000 Hz and, if elevated, BC at least for 2000 Hz. With this information, appropriate management can be initiated early, to be modified later as further, behavioral information becomes available.
The question is often asked, how long does tone-ABR testing require? It definitely takes longer than a simple air-conducted, click-evoked ABR, since far more information is being sought. Therefore, clinicians must be skilled in carrying out and interpreting tone-ABR results, and they should use appropriate and efficient test protocols. Nevertheless, even with efficient protocols, there will be infants for whom complete information cannot be obtained in a single test session. With today's very early identification, infants are now seen at a young age when they sleep naturally and sedation is rarely necessary or appropriate. There should be no hesitation in scheduling a second diagnostic ABR session.
Early-identification programs make ABR procedures more important than ever. In most cases, delaying the fitting of amplification until reliable behavioral thresholds have been obtained is no longer acceptable. Although other measures are useful in determining an infant's auditory capabilities, only the tone-evoked ABR provides the frequency-specific air- and bone-conduction threshold information required for fitting hearing aids in very young infants. Clinicians must begin to use these tone-ABR techniques, and therefore audiology training programs and workshops must provide students and professionals with the training and experience necessary to do so.
Discussions and collaborations with many individuals have helped develop these tone-evoked ABR protocols. In particular, I thank Terence Picton, Judith Gravel, and, most recently, Martyn Hyde. In many cases, the “we” referred to in this paper includes these individuals.