Otoacoustic Emissions in Children

Unidentified hearing loss in infants and young children can lead to delays in speech and language acquisition (Yoshinaga-Itano et al., 1998; Moeller, 2000). Identification of hearing loss presents an additional challenge because these patients may be unable to provide voluntary responses to sound. Otoacoustic emissions (OAEs) are an effective means to identify hearing loss in young children because they are related to the integrity of the peripheral auditory system and do not require voluntary responses from the patient.

OAEs are by-products of normal, nonlinear cochlear function, the source of which is the outer hair cell (OHC) system. They may be evoked by single tones (stimulus frequency, SFOAEs), pairs of tones (distortion product, DPOAEs), or transient stimuli (transient evoked, TEOAEs), and take from several seconds to several minutes to measure. Although all OAEs do not require a behavioral response, only TEOAEs and DPOAEs have been used widely to identify hearing loss. Since OHC damage results in hearing loss, OAEs, which are generated by OHCs, should be present when cochlear function is normal and reduced or absent when it is not. These facts have led to the application of OAE measurements in efforts to describe auditory function in humans, especially infants and young children. Unfortunately, OAE response properties from normal and impaired ears are not completely distinguishable; thus, diagnostic errors are inevitable. Below, we provide brief descriptions of TEOAEs and DPOAEs in normalhearing infants and young children, followed by a description of these two OAEs in patients with hearing loss. Robinette and Glattke (2002) and Hall (2000) provide more background information and extensive reference lists on OAEs. Norton, Gorga, et al. (2000) and Gorga, Norton, et al. (2000) provide comprehensive descriptions of OAEs in the perinatal period.

Infants and young children produce larger OAEs than older children and adults (Prieve, Fitzgerald, and Schulte, 1997; Prieve, Fitzgerald, Schulte, and Kemp, 1997; see Widen and O'Grady, 2002, for a review). There are several explanations for this difference. Very young children have not been exposed to environmental factors that might result in OHC damage. Also, their middle ears transmit energy to and from the cochlea differently than adults (Keefe et al., 1993), which might alter OAE levels in the ear canal. In addition, infant ear canal resonances show greater level at high frequencies, compared to spectra measured in adult ear canals. If stimuli differ in adult and infant ear canals, then responses may differ as well. Finally, the space between the measuring microphone and the eardrum is smaller in infants than in adults. If equivalent OAEs were generated in the cochlea, that signal would be larger in the infant ear canal because it was recorded in a smaller space.

While differences in infant and adult OAEs exist, the larger question revolves around whether OAEs can be used to distinguish ears with hearing loss from those with normal hearing. This dichotomous decision is made whenever OAEs are used in screening programs, regardless of the target population. The following discussion describes work in this area.

The clinical value of OAE measurements was recognized starting with their discovery (Kemp, 1978). Several studies describe the accuracy with which OAEs identify auditory status (e.g., Martin et al., 1990; Prieve et al., 1993; Gorga et al., 1993a, 1993b, 1996, 1997, 1999, 2000; Glattke et al., 1995; Kim et al., 1996; Hussain et al., 1998; Dorn et al., 1999; Harrison and Norton, 1999; Norton, Widen, et al., 2000). In general, both TEOAEs and DPOAEs identify auditory status with greater accuracy for middle and high frequencies than for lower frequencies. This occurs because noise levels decrease as frequency increases during OAE measurements. The noise interfering with OAE measurements (1) is acoustical, (2) results mainly from patient breathing and/or movement, and (3) contains mostly lower frequency energy. Noise adds variability and reduces measurement reliability. Thus, OAE test performance depends heavily on the frequencies at which predictions about auditory status are being made, in large part because noise level depends on frequency.

Test performance also depends on stimulus level (Whitehead et al., 1995; Stover et al., 1996; Harrison and Norton, 1999). Moderate-level stimuli result in the fewest false positive and false negative errors. Lower or higher stimulus levels decrease one of these error rates at the expense of increasing the rate of the other. This occurs for simple reasons. If low stimulus levels are chosen, virtually every ear with hearing loss will fail the test, resulting in a false negative rate of zero. However, the number of ears with normal hearing not producing responses will also increase as stimulus level decreases, increasing the false positive rate. If highlevel stimuli are used, the vast majority of ears with normal hearing will produce responses, resulting in a low false positive rate. Unfortunately, some ears with hearing loss, especially ears with mild or moderate losses, will produce a response to high-level stimulation, increasing the false negative rate. Moderate-level stimuli result in optimal combinations of false positive and false negative rates. Thus, primary levels of 50-65 dB SPL for DPOAE measurements or 80-85 dB pSPL for clicks during TEOAE measurements are recommended.

Figure 1 shows representative examples of DPOAE and TEOAE signal and noise levels for three hearing loss categories. In general, robust responses above the noise floor are observed when hearing is normal (top row). When borderline normal hearing or mild hearing loss exists (middle row), the response is either reduced in level or absent. In cases of moderate or greater hearing loss (bottom row), the response typically does not exceed the noise floor, even when the noise level is low. These examples are consistent with general response patterns in these hearing loss categories, but it is important to

750 1000 1500 2000 3000 4000 6000



: C

3 C

> A


A z


A ;




A i

i : a





750 1000 1500 2000 3000 4000 6000

500 750 1000 1500 2000 3000 4000 6000


Figure 1. OAE (circles) and noise (triangles) levels as a function of frequency, with DPOAE and TEOAE data shown in the left and right columns, respectively. Note that the y-axis is not the same for DPOAEs and TEOAEs. Representative examples are shown for normal hearing (top row), borderline normal hearing/mild hearing loss (middle row), and moderate to severe hearing loss (bottom row). See Gorga et al. (1997) for an explanation of the shaded areas in the DPOAE column. Although Hussain et al. (1998) developed a similar grid for TEOAE data, those data were collected with a different paradigm, and thus cannot be applied to the present set of data.

remember that both measurements will produce some diagnostic errors.

One of these errors (false positives or false negatives) may be more important for certain clinical applications. For example, infants or young children who are brought to a speech and hearing clinic or an otolaryngology clinic because of concern about hearing loss are at higher risk than the general population. In this case, one might choose stimuli or criteria that provide higher sensitivities, despite the higher false positive rates, because missing a hearing loss is the greater concern. In contrast, one might choose stimuli or criteria that provide higher specificity, despite the lower sensitivity, when the target population includes only well babies without risk for hearing loss. In this group, the probability of hearing loss is so low that it may be more important to minimize false positive errors. It is impossible to recommend a single set of stimuli and/or criteria because individual clinics must decide which error is more important for their needs.

OAE test performance also depends on the audio-metric criterion defining the border between normal hearing and hearing loss. Both TEOAEs and DPOAEs perform best when thresholds <20-30 dB HL are used to define normal hearing. There are data suggesting that TEOAEs are more sensitive than DPOAEs to mild hearing loss (for a review, see Harris and Probst, 1997). However, direct comparisons failed to reveal large differences in test performance between TEOAEs and DPOAEs when audiometric criterion was varied (Gorga et al., 1993b; Norton, Widen, et al., 2000). Still, some differences across frequency have been observed. TEOAEs tend to perform better at detecting hearing loss for lower frequencies while DPOAEs tend to perform better at detecting high-frequency hearing loss (Gorga et al., 1993b; Kemp, 1997), because of how each measurement is made.

During TEOAE measurements, a fast Fourier transform (FFT) is performed on the ear canal waveform. The status of the cochlear region associated with specific frequencies is determined by examining the energy (or signal-to-noise ratio) at those frequencies. For example, one would conclude that the 1000-Hz region of the cochlea is functioning if energy is observed in the FFT at 1000 Hz. For DPOAE measurements, two tones are presented simultaneously (f1 and f2), interact at the cochlea place close to where f2 is represented, and produce distortion products (DP), the most prominent of which is the 2f1 -f2 DP. The level of this component is then measured to determine if the cochlea is functioning at the point of its initial generation (f2). However, 2f1-f2 occurs at a frequency that is about one-half octave lower than f2. Thus, the measured response may occur in a region in which noise floors are less favorable, thus reducing measurement reliability. As a consequence, DPOAEs are less accurate than TEOAEs for lower frequencies.

During TEOAE measurements, the first 2.5 ms of the ear canal waveform following stimulation usually is zeroed to ensure that stimulus artifact does not contaminate the measured response. However, TEOAE energy generated in the high-frequency (basal) end of the cochlea will return with the shortest latency. Zeroing the first 2.5 ms of the ear canal signal may remove some of the high-frequency cochlear response. DPOAEs are not susceptible to this problem. Thus, they predict cochlear status better than TEOAEs do at higher frequencies.

Implicit in the above discussion is that errors are inevitable regardless of OAE measurement, stimulus level, OAE criterion value, or the definition of normal hearing. Both TEOAEs and DPOAEs will miss some ears with hearing loss and/or will incorrectly label some ears with normal hearing as hearing impaired. In addition, OAEs are not useful measurements of sensory function when middle ear dysfunction exists, which is frequently the case in children. Furthermore, OAEs will not identify patients with pathologies central to the OHCs, because OAEs test only the OHC system. Since the majority of hearing losses arise from OHC damage, however, OAEs are well-suited to the task of determining auditory status in infants and children.

Was this article helpful?

0 0
Micro Expression Master

Micro Expression Master

If You Could Read Everyone Life A Book You Can Have Better Career, Great Relationships And Become Successful. This Book Is One Of The Most Valuable Resources In The World When It Comes To Reading the smallest and tiniest body Language and know what people are thinking about.

Get My Free Ebook

Post a comment