Distortion Product Otoacoustic Emissions: A Tool for Hearing Assessment and Scientific Study

Abstract

Distortion product otoacoustic emissions (DPOAEs) reflect outer hair cell integrity and cochlear function. When used appropriately in the audiology clinic, they are an effective diagnostic tool and can detect hearing loss with accuracy. DPOAEs are easily and rapidly recorded in newborns and children, and provide basic hearing screening information as well as detailed diagnostic information in cases of suspected hearing loss. In the past decade, solid guidelines hnve been established to select the most effective recording parameters, thereby optimizing the DPOAE’s diagnostic potential.

DPOAEs also provide hearing scientists with a frequency-specific and noninvasive probe of the cochlea and cochlear amplifier function. Sophisticated and complex DPOAE-based experimental paradigms hnve been developed and applied to address scientific questions about cochlear function in humans. One such paradigm, DPOAE ipsilateral suppression, has been used effectively in our laboratory to study the maturation of cochlear function in newborns. Because of its proven accuracy as a clinical tool for the detection of hearing loss and its extensive use as a scientific tool for cochlear exploration, DPOAEs are likely to enjoy continued popularity and application in both the audiology clinic and the hearing science laboratory.

Introduction

Neonatal hearing screening currently is mandated in 37 states and the District of Columbia, is pending in four states, and has been partially implemented on a volitional basis in four other states (National Campaign for Hearing Health, 2003). The widespread implementation and interest in neonatal hearing screening stems from recent research showing that identification of hearing loss prior to 6 months of age provides a child an excellent chance to acquire normal language and speech (Yoshinaga-Itano, Sedey, Coutler, & Mehl, 1998). In stark contrast, children identified with hearing loss after 6 months might fall considerably behind their early-identified peers and could show delayed speech and language throughout childhood.

The movement toward state-mandated newborn hearing screening is also related to the availability of objective tests that provide accurate detection of hearing loss. These tests involve the measurement of physiological processes associated with hearing and do not require a behavioral response. The best recognized of these tests is the auditory brainstem response (ABR) test. However, the advent of otoacoustic emissions (OAE) and their incorporation into the auditory assessment battery has provided an alternative or complementary choice of objective tests for the evaluation of hearing. OAE tests provide a preneural, noninvasive look into the human cochlea and typically require only a few minutes of testing. Their accuracy in the detection of hearing loss when used appropriately has made them a popular tool in the neonatal hearing assessment arsenal (Norton et al., 2000). In the following sections, we will describe: (a) the physiological basis for the generation of distortion product otoacoustic emissions (DPOAEs) and how our view of the human cochlea has changed greatly in the past 20 years; (b) appropriate implementation and interpretation of DPOAEs for detection of hearing loss; and (c) a specific DPOAE paradigm that has been used successfully to address scientific questions about cochlear maturation in humans.

The Active Cochlea

Hearing scientists have learned much about cochlear function in the past three decades. The cochlea is unique among sensory organs: it codes and preliminarily processes sound en route to the auditory cortex, but it is also capable of producing sound. The cochlea is a nonlinear mechanism, meaning it does not receive and transmit auditory information in an unaltered form. The cochlea acts upon incoming sound with its own intrinsic energy source and, in so doing, changes the characteristics of vibration along the delicate basilar membrane. Acoustic emissions, or “OAEs,” are created as byproducts of this process.

In the 1960s, von Bekesy (1960) established that the basilar membrane was tonotopically organized, and he described a “traveling” wave or vibratory fluid motion that traversed the membrane from base to apex. According to von Bekesy, the vibratory motion on the basilar membrane reached its peak of membrane displacement at a region related to frequency of the incoming sound. This basilar membrane region associated with the characteristic frequency of a sound is determined by the acoustic properties of the sound, as well as the mass and stiffness of the membrane (Pickles, 1988).

Tonotopicity, however, provides only a partial explanation of cochlear processing. Our most recent knowledge indicates that as the traveling wave motion is building and reaching its peak vibration, an amazing process is triggered. Outer hair cells (OHCs) that are lined up in three rows atop the basilar membrane begin to elongate and contract at rates that are well beyond contraction rates for muscle tissue (Brownell, 1990). Their motility is triggered by a significant change in resting potential. This voltage change is produced by the traveling wave motion; therefore, only those OHCs that are maximally stimulated (on the upswing of the traveling wave) become motile. And, because they are motile in only a focal region of the membrane, the displacement around this narrow cochlear region (i.e., characteristic frequency) is made larger, powered by OHCs that are pulling downward on the tectorial membrane and upward on the basilar membrane. This mechanical amplification of basilar membrane motion sends a clear and robust message to the brain about the acoustic input. This robust vibration pattern on the basilar membrane translates into excellent detection of sound. Enhanced membrane displacement in a focal region, around the tip of the traveling wave and just basal to it (Neely & Kim, 1986), also produces a more highly tuned signal, which translates into excellent frequency resolution.

This OHC-mediated amplification process produces an enhanced representation of the auditory signal to the brain and has been collectively termed the “cochlear amplifier” (Davis, 1983). The cochlear amplifier (CA) works to improve perception of sounds that are low-to-moderate in level only. The cochlear amplifier’s effectiveness saturates with high level input. It is not difficult to hypothesize the purpose of this amplifier and why it might have evolved. Without sensitive detection and exquisite frequency resolution, humans would have trouble detecting some of the critical nuances of speech; nuances that are required for adequate speech perception and discrimination. In fact, when an individual has mild-to-moderate amounts of sensorineural hearing loss he or she has typically lost OHCs (cochlear amplifier) and often scores poorly on speech discrimination tasks in noise.

In the process of amplifying and enhancing sound for efficient decoding by the brain, the cochlea creates a byproduct. Just like sound that comes into the ear from an external source, the cochlear byproduct produces its own physical vibration along the basilar membrane. The DPOAE is generated by the cochlea when the ear is presented with two simultaneous pure tones (f1 and f2). The DPOAE travels from its generation site on the basilar membrane around the f1 and f2 frequency to the region of its own characteristic frequency. It also travels in reverse direction from its generation site, through the middle ear, and into the ear canal. The cochlear-generated distortion becomes an acoustic product once it is in the ear canal and can be measured with a sensitive microphone placed at the auditory meatus. When the DPOAE is present in the ear canal, it indicates that the mechanism generating it (i.e., the cochlear amplifier) is functional; when the DPOAE is absent, it indicates that the amplifier is nonfunctional or dysfunctional and that there is hearing loss. In this way, a distortion tone produced by the ear as a byproduct and measured in the ear canal has the useful characteristic of reflecting cochlear integrity.

Because we cannot access the neonatal cochlea in any other way, DPOAEs provide a useful alternative. As is evident, this association between otoacoustic emissions and cochlear function is an incredibly beneficial feature, one that makes the DPOAE an excellent tool for hearing assessment and for the scientific study of the cochlea. There are other types of otoacoustic emissions: spontaneous otoacoustic emissions and transient-evoked otoacoustic emissions. Spontaneous OAEs (SOAEs) are pure tones produced by the ear in the absence of eliciting stimuli. They are present at fixed frequencies in approximately 60–70 percent of people with normal hearing. Transient-evoked or dick-evoked emissions (TEOAEs) are responses produced when an abrupt-onset, transient stimulus is presented to the ear. TEOAEs provide an overall, wideband look at cochlear function, across a broad range of frequencies.

This article focuses only on the 2fl-f2 DPOAE. The following section will describe how the DPOAE is used for clinical assessment of hearing in infants and children, including the appropriate recording parameters and appropriate interpretation of DPOAE results.

DPOAEs as a Tool for Clinical Assessment of Hearing

DPOAEs can be used effectively to diagnose or detect hearing loss in infants and children. Although they are technically not a measure of “hearing,” they are correlated with hearing. As such, they are useful in the audiology clinic. Under good-to-excellent test conditions, the correlation is fairly straightforward: When DPOAEs are present and normal in amplitude and configuration, their presence indicates that the cochlear amplifier (i.e., OHC motility) is normally functional. In the absence of neurological or isolated inner hair cell dysfunction, this result is consistent with normal hearing. When DPOAEs are absent, their absence indicates that there is some dysfunction in the cochlea, though the level of dysfunction and, thus, degree of the hearing loss is not clear. This reliable correlation between DPOAEs and hearing allows for the effective clinical application of this easily recorded response.

Various studies have shown DPOAEs to be effective in correctly identifying subjects with known sensorineural hearing loss (Gorga et al., 1993, 1999, 2000; Norton et al., 2000; Stover et al., 1996). An important requirement to ensure their clinical usefulness, however, is that DPOAEs be measured using the appropriate recording and stimulus parameters—that is, the parameters that produce the most robust response and the response that best reflects cochlear status and integrity.

DPOAEs are generated when two pure tones (fl, f2) are presented simultaneously to the ear. In the clinic, they are typically generated across a broad range of f2, f1 frequencies at one pre-established stimulus level. The resulting graph is called a “DP-gram.” This is a graph of DPOAE amplitude as a function of f2 frequency. When a newborn patient is brought into the clinic by his or her parents after failing the initial hospital hearing screening, or a toddler who is not developing speech and language as expected walks into an audiologist’s sound booth, four stimulus-related parameters need to be selected by the clinician before recording DPOAEs: (a) absolute level of the primary tones; (b) their level separation (i.e., level relative to one another); (c) frequency of the primary tones; and (d) the frequency ratio between primary tones (i.e., the frequency relationship between the tones).

The first factor, primary tone level, is critical to recording a clinically diagnostic DPOAE. Stover and colleagues (1996) found that moderate level primary tones, presented in the range of 55–65 dB SPL, provide the best accuracy in separating normal hearing subjects from subjects with hearing loss. These researchers found that primary tones presented at either higher or lower levels produced more error in the detection of hearing loss. These errors could translate into clinical misdiagnoses. High-level primary tones underestimated hearing loss; therefore, more children with hearing loss might be missed and classified as having normal hearing. The use of low-level primary tones overestimated hearing loss because the reduced signal-to-noise ratio (SNR) made DPOAE detection difficult and less reliable. This created false-positive errors or classification of individuals with normal hearing as having a hearing loss.

This information on primary tone level in humans is consistent with more experiments conducted on laboratory animals. In small mammals, DPOAEs produced by high-level primary tones are not based in cochlear amplifier activity. These DPOAEs are present and robust even when drugs are given that eliminate OHC motility and, thus, eliminate cochlear amplifier function (Brown, McDowell, & Forge, 1989; Mills & Rubel, 1996; Whitehead, Lonsbury-Martin, & Martin, 1992a, 1992b). DPOAEs generated with low-level primary tones, in contrast, are reduced in amplitude or abolished completely by drugs that eliminate OHC motility (e.g., salicylates or aminoglycosides). This indicates that DPOAEs recorded with low-level stimuli are, in fact, reflecting integrity of cochlear amplifier function and DPOAEs recorded with high-level primary tones are reflecting something at least partially unrelated to cochlear amplifier function. Comparable research cannot be conducted on humans to confirm these findings because the drugs used can be damaging. However, the results with humans (Stover et al., 1996) also suggest that high-level DPOAEs are not optimally diagnostic for hearing loss.

The second parameter determined when recording DPOAEs for clinical purposes is level separation. Although we know from the above-cited work that mid-level tones produce the most diagnostic DPOAE, this work does not tell whether both primary tones should be presented at the same moderate level or whether one of the primary tones should be presented at lower or higher levels than the other. Research with human adults has shown that a 10 dB level separation produces the largest amplitude, and the most robust DPOAE when in the mid-level range (Gaskill & Brown, 1990; Popelka, Osterhammel, Nielsen, & Rasmussen, 1993). This same finding has also been reported for newborns born full term or prematurely (Abdala, 1996). Thus, it appears that a 10–15 dB level separation, with L1 (lower frequency primary tone) greater than L2 produces the largest amplitude DPOAE in both adults and infants. For optimal accuracy in detection of hearing loss, most clinical instrumentation will include default level settings of 65–55 dB SPL or 60–50 dB SPL when recording a DP-gram.

Primary tone frequency is the third factor that needs to be determined prior to recording DPOAEs. The DPOAE is most diagnostically effective (i.e., has the greatest “hit” rate for detection of hearing loss) when recorded in the range of f2 frequencies between 2000Hz and 8000Hz (Gorga et al., 1999). At an f2 frequency of 500 Hz, the ability to separate individuals with normal hearing from those with hearing loss approaches chance because of the elevated noise floor. At 1000 Hz, it is only slightly better. Accuracy in the detection of hearing loss using DPOAEs is optimal when using primary tones in the mid-to-high frequency range. A clinical DP-gram will typically include f2 frequencies ranging from approximately 800Hz through 6000Hz or 8000 Hz.

The fourth recording parameter to be selected is frequency ratio—the separation between primary tone frequencies. On average, in both adults and newborns, 1.2 is the f2/ f1 ratio that produces the most robust DPOAE and is considered the optimal frequency ratio (Abdala, 1996; Gaskill & Brown, 1990; Harris et al., 1989)—that is, the f2 tone is presented at a frequency 1.2 times higher than the f1 tone. Optimal frequency ratio varies slightly with stimulus level and frequency. At low levels and low frequencies, f2/fl needs to be slightly smaller than 1.2 to produce optimal amplitude. However, these kinds of minute adjustments of ratio are probably not necessary in clinical practice. In most cases, a 1.2 ratio will produce a robust DPOAE, or at least maximize whatever response amplitude is available.

If optimal recording parameters are used and the DPOAE test is conducted under stable circumstances (e.g., low noise, adequate fit of the probe, minimal subject movement), the result can typically be classified into one of three outcome categories and can provide a reliable interpretation of hearing status in subjects that cannot be tested using behavioral techniques. The three outcome categories are: (a) DPOAE is absent; (b) DPOAE is clearly present and normal; and (c) DPOAE is present but not normal.

An absent DPOAE is easy to classify. If the DPOAE is not present at two or more f2 frequencies with sufficient signal-to-noise ratio (SNR), it is an absent response. What is sufficient SNR? This depends on noise floor calculations used by the instrumentation. Some instrumentation calculates the mean noise floor with response variability (standard deviations) incorporated into the measurement. If this is the case, a 3–4 dB SNR is sufficient to detect a response.

If the mean noise floor is used as a reference with no variability calculation included, then a 5–6 dB SNR or greater is often required. Therefore, an absent result is one in which the DPOAE is not present with at least 3–6 dB SNR for more than one f2 frequency. Figure 1a displays a DP-gram that shows a clearly absent DPOAE response. All data points are embedded in the noise and do not have sufficient SNR to be considered valid responses.

Figure 1.

(a). DP-gram showing a clearly absent DPOAE. All data points are embedded in the noise floor; (b). DP-gram showing a present and normal DPOAE from a 6-year-old child. Both SNR and absolute amplitude values are robust and meet criteria; (c). DP-gram showing a present but not normal response from an 8-year-old child with mild sensorineural hearing loss. The DPOAE is absent at some f2 frequencies, present, but low amplitude at others and completely normal at the very highest frequencies.

Adequate SNR at only one single primary tone frequency is not likely reflecting a true cochlear response, since hearing loss does not usually produce such sharp boundaries of cochlear function/ dysfunction. A single “spike” in a DP-gram at only one f2 frequency should not be considered indicative of a present response. If the test session was considered to have good reliability and there is no middle ear pathology present, the absent DPOAE result indicates cochlear dysfunction. The dysfunction and associated degree of hearing loss may range from mild to profound. The DPOAE test cannot determine degree of hearing loss. Threshold tests, such as an ABR or a behavioral audiogram, will need to be conducted to obtain severity information and to guide (re)habilitation efforts with infants and children.

A clearly present and normal outcome must show a DPOAE with: (a) greater than 3–6 dB SNR at approximately 70 percent of the collected data points (i.e., at four or five of six f2 frequencies); and (b) appropriate absolute amplitude for the patient’s age. It is important to note that both criteria must be met to consider a response completely normal. For example, a newborn DPOAE must have at least a 3–6 dB SNR from 2000 to 8000 and the amplitude at these frequencies must be in the range of approximately 5–25 dB SPL because newborns with normal hearing have robust OAE energy and large amplitude responses. Likewise, a −10 dB SPL DPOAE response from a 3-year-old child could not be considered normal, even if the noise floor was −20 dB SPL and the SNR criterion was met. This response could not be considered normal because a −10 dB SPL DPOAE is not in the normal range of response amplitude for a toddler with normal hearing (see Prieve, Fitzgerald, Schulte, & Kemp, 1997, for a review of normative DPOAE amplitude as a function of age). Thus, both SNR and absolute amplitude criteria must be considered for a response to be classified as unequivocally present and normal. Figure 1b shows a completely normal and present DPOAE response from a 6-year-old child.

A present and normal DPOAE, however, does not require that each f2 frequency measured on the DP-gram show criterion SNR and/ or amplitude. This excessively stringent criterion would lead to the detection of “hearing loss” where none was present. In newborns with normal hearing, absent DPOAEs are commonly found in the low-frequency range only because of the high noise floor in this patient population at low frequencies (Lasky, 1998; Prieve et al., 1997). Likewise, in adults with normal hearing, absent DPOAEs are common at the highest frequencies tested, probably reflecting age-related OHC loss in this frequency range. In addition, ear-canal and middle-ear resonance effects often produce a “dip” or low amplitude trough in the mid-frequency range of the DP-gram (Lonsbury-Martin et al., 1990), but this should not be interpreted as a region of hearing loss. For these reasons, the DPOAE does not need to be present at all f2 frequencies tested to be considered present and normal.

What can an audiologist say about hearing when a DPOAE outcome is present and normal? It is appropriate to state that cochlear function, and most likely hearing, is normal. Certainly if there are neurological issues that might impede the coding of auditory information at higher levels of the system, normal DPOAEs do not ensure normal hearing. However, in the absence of this type of pathology, normal cochlear function, as measured by clearly present DPOAEs, is typically consistent with normal hearing.

The third DPOAE outcome category is defined less rigidly than the first two and perhaps creates the greatest challenge for diagnosticians: the DPOAE is present but not normal. It is true that some clinicians do not consider this result or outcome and seem to prefer a two-category diagnostic scheme, perhaps because of its simplicity. However, the two previously described categories (absent and present) alone cannot aptly describe DPOAE outcome in infants and children. This scheme fails to recognize that OAEs may be present in ears with mild to moderate degrees of hearing loss (≈35–45 dB HL). Without including a third diagnostic category (i.e., present but not normal), a somewhat “suspicious” test result may be erroneously classified as normal and could produce a false-negative result, failing to detect the presence of mild hearing loss.

The present but not normal DPOAE meets or marginally meets SNR criteria (e.g., 3 dB SNR) but might have absolute amplitude that is unreasonably low for a patient’s age. Another example of this outcome could be a DP-gram that shows islands of response in limited frequency ranges. For example, DPOAEs may be present from 500 Hz to 2000 Hz and then absent from 2500 Hz to 8000 Hz, or visa versa. Figure 1c shows an example of a DP-gram recorded from an 8-year old child with mild, sensorineural hearing loss. This child has present but low amplitude DPOAEs at some f2 frequencies, and absent responses at other frequencies.

When a child has excessively low-amplitude DPOAEs or frequency “islands” of absent response, it is not correct to state that the DPOAE is completely absent. However, it is also inaccurate to state that the DPOAE is normal. The risk in not considering this third outcome category is that an audiologist might miss a mild sensorineural hearing loss. As long as the test conditions were good and the test considered reliable, a present but not normal DPOAE outcome can be interpreted to indicate that there is some dysfunction in the cochlea and, thus, some hearing loss present (probably mild in nature). The degree and configuration of the hearing loss cannot be definitively determined from OAE testing only.

When recorded with optimal parameters and interpreted appropriately, DPOAEs provide important information about the integrity of the cochlea and, consequently, about hearing. When an infant has failed a newborn hearing screening and is referred for audiological follow-up, DPOAEs are often the initial diagnostic test conducted in the clinic. They can be an effective tool either to confirm the presence of hearing loss or rule it out. DPOAEs are most effectively applied as one component of a multifaceted hearing assessment battery. The final section of this article will describe one way in which the DPOAE has been used as a scientific tool to investigate the maturation of human cochlear function in newborns.

DPOAEs as a Tool for Scientific Investigation

In addition to their utility as a clinical hearing assessment tool, DPOAEs are widely used in auditory research. In the Infant Auditory Research Laboratory of the House Ear Institute, DPOAEs have been applied extensively to study the maturation of human cochlear function in newborns. As a research tool, DPOAEs provide a pre-neural, frequency-specific probe of outer hair cell (OHC) integrity. When they are studied as a function of age, as in our laboratory, DPOAEs provide an effective and extremely useful tool for describing maturation of human cochlear function.

The DPOAE paradigm that has been widely used within our laboratory to explore maturation of human cochlear function is ipsilateral DPOAE suppression. A DPOAE is generated by two simultaneously presented pure tones: f1 and f2. The combination of these two tones and the subsequent vibration patterns they produce on the basilar membrane creates intermodulation distortion. Because distortion tones create their own traveling waves on the basilar membrane, their vibration patterns can be disrupted and suppressed by an external tone. The ipsilateral suppression paradigm involves the suppression (or amplitude reduction) of a distortion product by a third tone presented simultaneously with primary tones f1 and f2. To get a clear picture of how a given DPOAE at a specific f2 frequency is affected by suppression, many suppressor tones (fs) of varying frequency must be presented around f2. For example, to study the cochlear region around 6000 Hz, suppressor tones are presented ranging from 4200 Hz to 7200 Hz. This provides a comprehensive picture of suppression patterns around the target frequency of 6000 Hz.

In our laboratory, a DPOAE suppression tuning curve (STC) is generated when the level required for a given suppressor tone to reduce DPOAE amplitude by 6 dB is plotted as a function of that suppressor tone’s frequency (see Figure 2). In the normal adult ear, suppressor tones nearest f2 require the lowest levels to suppress the DPOAE amplitude because they are close to the primary generation site of the distortion product (Martin et al., 1987). In contrast, suppressor tones either significantly above or below f2 frequency require higher levels to achieve the same suppressive effect. Thus, a DPOAE STC has a sharply tipped, narrow, bowl-shaped configuration. When we generate these suppression tuning curves or DPOAE STCs in infant subjects of various age, we can learn about the time course for the maturation of cochlear function in humans.

Figure 2.

DPOAE suppression tuning curve (STC) generated at an f2 frequency of 3000 Hz. Suppressor tones presented to record this STC ranged from 1800Hz to 4000Hz. Note that the tip region is centered around the f2 frequency.

To study maturation of the cochlea, we quantify and analyze several aspects of the DPOAE STC in infants. In addition, we quantify suppression growth patterns—that is, the rate at which suppression grows (i.e., the rate at which DPOAE amplitude drops) as the suppressor level is increased. Figure 3, for example, shows suppression growth patterns for 12 different suppressor tones centered around 3000Hz. It is evident that DPOAE amplitude drops at different rates depending on the frequency of the suppressor. Suppressor tones lower than f2 (see upper panel of Figure 3) produce steep, effective suppression in the normal adult cochlea. Suppressor tones higher than f2 (lower panel) produce shallow suppression (Harris, Probst, & Xu, 1992; Kummer, Janssen, & Arnold, 1995). A slope value is calculated from each one of these functions using a standard linear regression equation in order to capture the frequency-dependent pattern of suppression growth. Figure 4 illustrates this process. The larger the slope value, the steeper the suppression growth and the more effective the suppressor tone.

Figure 3.

DPOAE amplitude as a function of suppressor level for 12 different suppressor tones centered around an f2 frequency of 3000 Hz. The upper panel displays growth of suppression for tones lower than f2 (fs<f2). The lower panel displays growth of suppression for tones higher than f2 (fs > f2). Note that high-frequency suppressor tones (lower panel) produce much more shallow growth of suppression than low-frequency tones.

Figure 4.

Suppression growth is measured by fitting a line to each function and calculating a slope value (dB/ dB). The slope value reflects the rate at which suppression grows (DPOAE amplitude drops) as the suppressor tone is increased in level. An example of steep (1.3) and shallow (.2) suppression growth is provided in this figure.

It is not clear exactly which aspect of cochlear function is being explored with the DPOAE suppression paradigm. Some early researchers suggested that it provides a good indication of cochlear tuning (Brown & Kemp, 1984). However, recent work with subjects with hearing loss indicates that DPOAE suppression tuning does not reflect cochlear frequency resolution in a traditional manner (Abdala & Fitzgerald, 2003; Howard et al., 2002). Some scientists have hypothesized that DPOAE suppression tuning provides an outline of the DPOAE generation site (Martin, Jassir, Stagner, & Lonsbury-Martin, 1998) and recent work from our laboratory strongly suggests that it reflects cochlear non-linearity in the same fashion as more direct measures of basilar membrane displacement (Abdala & Chatterjee, 2003). Whatever the exact underlying “process” that is being measured, there is a systematic, highly frequency-dependent and expected pattern of suppression in adults with normal hearing; therefore, an altered or atypical pattern of suppression observed in human newborns could indicate an immaturity in cochlear non-linearity and frequency coding.

In our first study of cochlear maturation, we tested 16 newborns that were born full term and 15 adults with normal hearing at three f2 frequencies: low (1500Hz), mid (3000Hz), and high (6000Hz), using optimal recording parameters (Abdala, Sininger, Ekelid, & Zeng, 1996). We also recorded psychoacoustic tuning curves in 3 of our adult subjects to confirm their similarity with DPOAE suppression tuning curves. The objective of this study was two-fold: (a) to characterize DPOAE suppression tuning in adults with normal hearing, and (b) to initiate the study of cochlear maturation in newborns who were born full term. We analyzed DPOAE STCs by quantifying their width (Q10), the steepness of their sides or “flanks” on both low and high frequency sides (dB/octave), and by characterizing tuning curve tip level and tip frequency. Our results show that adult DPOAE STCs resembled other more traditionally recorded frequency tuning curves from the auditory system. They were asymmetrical in shape with a steep high frequency flank, narrow width, and a sharp tip region. This morphology is similar to tuning curves recorded from the basilar membrane, the VIIIth-nerve of the auditory system, and to those using behavioral, psychoacoustic methods (Liberman & Dodds, 1984; Moore, 1978; Rhode & Cooper, 1993).

Results also indicated that DPOAE suppression tuning curves from newborns born full term were similar in width and slope to adult STCs, although there were some age differences noted, primarily related to tuning curve tip features. In this first study, a limited number of neonatal STCs were collected at each frequency (as few as 7 in a group) and the initial analysis scheme was simple. This experiment served the purpose of establishing the DPOAE suppression technique as a viable tool for assessing cochlear function in newborns. Although the initial results suggested that newborns born full term were adult-like in cochlear function, more detailed results generated later with larger subject groups and more sophisticated experimental design and analyses challenged this early conclusion.

In our second DPOAE suppression study, we sought to test younger subjects and study cochlear function during an earlier period in auditory development (Abdala, 1998). We did this by including a group of 85 newborns born prematurely ranging in age from 31 weeks postconceptional age (PCA) to 41 weeks PCA. Subjects varied in gestational age at birth; however, for the purposes of this study, they were classified by their postconceptional age (weeks since conception) on the test day to better reflect maturational status. A group of 33 newborns born full term and 14 young adults with normal hearing were included for comparison. The same three frequencies (1500 Hz, 3000 Hz, and 6000 Hz) were investigated using optimal frequency ratio and level/level separation values (f2/fl = 1.2; 65-50 dB L1 > L2). DPOAE STCs were analyzed as described previously, but measurements of DPOAE suppression growth were added. These were analyzed by calculating a slope value to indicate the rate at which DPOAE amplitude dropped as a given suppressor tone was increased in level (see Figure 4). Figure 5 shows how this slope value, plotted as a function suppressor frequency, generates a recognizable pattern reflecting the frequency-dependent pattern of suppression growth in the normal cochlea. Data points above the dashed line reflect steep growth of suppression and points below the line reflect shallow growth.

Figure 5.

Slope values (reflecting suppression growth) plotted as a function of suppressor frequency. Slope values above the horizontal dashed line reflect steep growth of suppression and values below the line reflect shallow growth. As is evident, the growth of suppression is highly frequency-dependent.

We found that (a) newborns born prematurely had narrower and sharper tuning than adults at both low (1500 Hz) and high (6000Hz) frequencies, and full term neonates showed sharper tuning than adults at 6000 Hz only; (b) both premature and full-term neonates showed a steeper flank on the low-frequency side of the STC for 6000 Hz (all ages were comparable with respect to steepness on the high frequency tuning curve flank); and (c) the adult STC tip, at 6000 Hz, was elevated relative to newborns. Thus, with a larger data set and more detailed analyses, immaturities in DPOAE suppression tuning were detected even in the full-term newborns, in particular at 6000Hz.

In addition to the three above-noted immaturities in suppression tuning, the pattern of suppression growth across suppressor frequency showed age effect. Recall that the slope value measured from a suppression growth function reflects the rate at which DPOAE amplitude drops for a given suppressor tone. At both 1500 Hz and 6000Hz, the DPOAE was more difficult to suppress in newborns when using suppressor tones lower in frequency than f2. The growth of suppression in newborns for the lowest suppressor tones was shallow and ranged from .2 dB/dB to .8 dB/dB. In contrast, adults typically showed steep suppression growth (> 1.0 dB I dB) using these same low-frequency suppressor tones.

Our results showing narrower DPOAE suppression tuning and more shallow growth of suppression for low-frequency tones may reflect a subtle immaturity of cochlear function in full term and prematurely born neonates. To delve into this question more thoroughly, we designed a large-scale DPOAE suppression study to test larger groups of prematurely born neonates, classified with narrow age intervals for better accuracy in the detection of age effects (Abdala, 2001). This third DPOAE suppression experiment from our laboratory included 300 newborns: 202 were premature and 98 were full term (Abdala, 2001). Newborns born prematurely were tested in the following PCA categories: 31–33 weeks; 34–36 weeks, and 37–41 weeks. The study also included 30 adults with normal hearing. Subjects were tested at five frequencies spanning from 1500 Hz to 12,500 Hz at three primary tone levels.

The results of this study replicated the earlier findings and confirmed that newborns have narrower, sharper tuning than adults at 1500Hz and 6000Hz. However, suppression tuning was not different among premature newborn groups, and premature newborns were not different from full-term subjects. In addition, adults showed a loss of suppression tuning when stimulus levels were high (75–65 dB SPL), while newborns showed no degradation in tuning with level. This result suggests that adults and newborns may have different saturation thresholds for CA function.

One of the most robust age effects observed in this study involved the pattern of suppression growth. Again, results clearly established that suppressor tones lower in frequency than the f2 were not effective in suppressing the newborn DPOAE. The youngest premature group (31–33 weeks PCA) showed the most non-adult-like suppression growth for low-frequency suppressor tones, and the age groups became progressively more adult-like in suppression growth as they approached 40 weeks PCA (Figure 6).

Figure 6.

Slope values (reflecting suppression growth) plotted as a function of suppressor frequency for adults, newborns born full-term, and newborns born prematurely designated by post-conceptional age in weeks. Adults show larger slope values, indicating steeper suppression growth than newborns for suppressor tones lower in frequency than f2. Newborn data shows a strong developmental trend since suppression growth for low-frequency tones becomes increasingly adult-like with age.

This study clearly indicates that cochlear function is not adult-like in infants born full term or prematurely. The subsequent study used a longitudinal experimental design in an attempt to identify the period for cochlear maturation (Abdala, 2003). Rather than study large groups of infants at established ages (i.e., “cross-sectional design”), a small group of newborns was tested repeatedly as they increased in age from 31 to 33 weeks PCA to the equivalence of full term (i.e., 38–40 weeks).

Nine infants born prematurely were initially tested with the DPOAE suppression paradigm between 31 and 33 weeks PCA and then weekly thereafter for 6–8 sessions. In addition, a control group of adults with normal hearing was tested weekly for five sessions and a group of 10 infants born full term was tested during a one-time visit. DPOAE STCs were generated at an f2 frequency of 6000Hz because results at this frequency had shown to be sensitive to immaturities in cochlear function.

Figure 7 shows mean DPOAE STC width (Q10) and STC slope on the low-frequency flank (dB/octave) for 9 newborns and 5 adults as a function of postconceptional age or test session. Two trends are evident from this figure: (a) newborns continue to have excessively sharp tuning and steep slope during the entire test period (as shown by the larger Ql0 values for newborns and the separation of infant and adult groups along the vertical axis); and (b) newborns do not become adult-like once they reach 40 weeks (as shown by the lack of change in tuning curve width or slope over time). Even the full-term infants used for comparison are clearly not adult-like in suppression tuning characteristics and shows STC width and slope values that are similar to premature newborns. The results of this study indicate that there is continued maturation of cochlear function after a full-term birth and into the postnatal period.

Figure 7.

Mean tuning curve width (Q10) and slope on the low-frequency flank for 9 newborns born prematurely tested repeatedly at weekly intervals for 7–8 test sessions (open circles), 5 adults with normal hearing tested weekly for 5 sessions (filled squares) and 10 newborns born full-term tested during one session (X). Error bars represent +/− 1 sd.

Because our first four studies indicated that final cochlear maturation occurs during the postnatal period, our most recent experiment involves the testing of 3-month-old infants. This study was conducted to observe whether healthy 3-month-olds have become completely adult-like in their cochlear function, as measured by DPOAE suppression (Abdala & Visser-Dumont, 2003). Results indicate that both newborns and 3-month-olds continue to have larger Q10 values (indicating narrower tuning) than adults at an f2 frequency of 6000 Hz. STC width is comparable between newborns and 3-month-olds (see Figure 8). In addition, adult-like suppression growth patterns for low-frequency suppressor tones are still not observed in 3-month-old subjects (data not shown). Both newborns and 3-month-olds continue to show more shallow growth of suppression for low-frequency suppressor tones; however, data points from the 3-month-olds fall between newborn and adult data, indicating maturation in an adult-like direction. This experiment indicates that complete maturation of human cochlear function occurs after 3 months of age. Considering complementary research conducted with the ABR and behavioral paradigms, we hypothesize that the process is complete by 6 months of age.

Figure 8.

Mean tuning curve width (Q10) for three age groups: 3-month-old infants, newborns, and adults. Newborns and 3-month-old infants had comparable STC width and both infant groups had sharper DPOAE suppression tuning than adults.

By using a sophisticated DPOAE-based suppression paradigm to probe cochlear non-linearity and frequency coding, we have been able to explore the time course for maturation of human cochlear function. The DPOAE suppression paradigm is noninvasive and produces reliable results that are consistent with other more direct tests of basilar membrane motion and VIIIth-nerve response characteristics. The simple DP-gram might not have identified these subtle immaturities in neonatal cochlear function, but the suppression paradigm seems to provide a more comprehensive window into some of the complexities of cochlear coding and processing.

Summary

The DPOAE is a versatile and effective tool in the assessment of hearing and in the hearing scientist’s laboratory. Like any hearing assessment tool, it must be used correctly to ensure appropriate evaluation and interpretation of test outcome. When conducted in the optimal manner, with a clear understanding of interpretation, DPOAEs add significantly to the auditory assessment battery and provide a means of detecting hearing loss in infants and children that cannot be tested using standard behavioral techniques. An equally beneficial application of the DPOAE is in its use as a scientific tool to explore the cochlea. The DPOAE provides a noninvasive window into the cochlea and allows for frequency-specific exploration of CA function. By applying this scientific probe of the cochlea to infants during development, it also allows for the study of cochlear maturation in humans. We hope that a more comprehensive understanding of this maturational process will provide a normative framework for diagnosis and eventually lead to enhanced screening and detection of hearing loss in newborns.