The influence of self-reported noise exposure on 2ƒ₁-ƒ₂ distortion product otoacoustic emission level, fine structure, and components in a normal-hearing population

Abstract

Distortion product otoacoustic emissions (DPOAEs) offer an outcome measure to consider for clinical detection and monitoring outer hair cell dysfunction as a result of noise exposure. This investigation detailed DPOAE characteristics and behavioral hearing thresholds up to 20 kHz to identify promising metrics for early detection of cochlear dysfunction. In a sample of normal-hearing individuals with and without self-reported noise exposure, the DPOAE and hearing threshold measures, as assessed by two questions, were examined. The effects on various auditory measures in individuals aged 10–65 years old with clinically normal/near-normal hearing through 4 kHz were evaluated. Individuals reporting occupational noise exposures (n = 84) and recreational noise exposures (n = 46) were compared to age-matched nonexposed individuals. The hearing thresholds and DPOAE level, fine structure, and component characteristics for the full frequency bandwidth were examined. The data suggest that the DPOAE levels measured using a range of stimulus levels hold clinical utility while fine structure characteristics offer limited use. Under carefully calibrated conditions, the extension to frequencies beyond 8 kHz in combination with various stimulus levels holds clinical utility. Moreover, this work supports the potential utility of the distortion product place component level for revealing differences in cochlear function due to self-reported, casual noise exposure that are not observable in behavioral hearing thresholds.

I. INTRODUCTION

Noise-induced hearing loss (NIHL) is a growing health concern globally (WHO, 2021) and nationally as more than 30 × 10⁶ Americans are exposed to hazardous sound levels on a regular basis (NIDCD, 2002). NIHL is well-established in the literature (reviewed by Hong et al., 2013; WHO, 2021) and has a common place in the clinical practice across the lifespan. Significant research has focused on a deeper understanding of the auditory and nonauditory effects of recreational (e.g., Neitzel and Fligor, 2019), occupational (e.g., Themann and Masterson, 2019), and military (e.g., Moore, 2020) noise exposures. For this evaluation, we are highlighting self-reported “casual” noise exposure from our research population to reflect the results of noise exposures (defined as yes/no on a two-item questionnaire) in our auditory measures to explore potential clinical utility. Although there are several opportunities for overexposure to noise in everyday situations, these potentially damaging exposures may go unnoticed as they are not often painful and may not result in immediate or long-term measurable changes in auditory function, especially behavioral hearing thresholds. Opportunities to leverage noninvasive, objective measures across the lifespan are important for the earliest detection and monitoring of NIHL.

Distortion product otoacoustic emissions (DPOAEs) are often reduced or absent as a result of manipulations that reduce cochlear sensitivity, which is why DPOAEs are in wide clinical use as a noninvasive, objective indicator of cochlear health (e.g., Lonsbury-Martin and Martin, 1990). The 2ƒ₁-ƒ₂ DPOAEs can be recorded in the ear canal at frequencies mathematically related to the stimulus frequencies (ƒ₁ and ƒ₂, ƒ₁ < ƒ₂; Kemp, 1979). The nonlinearity responsible for this originates in the auditory periphery, which is evidenced by the response components at 2ƒ₁-ƒ₂ being measurable in the auditory nerve discharge (Goldstein and Kiang, 1968), ear-canal sound pressure (Kemp, 1979; Kim, 1980), and basilar membrane motion (Robles et al., 1997). DPOAE generation depends on the physiological status of the outer hair cells (OHCs), which are among the cochlear structures initially damaged by noise (e.g., Nordmann et al., 2000).

Exposure to high levels of noise and ototoxic medications, solvents, and heavy metals damages the OHCs, although the specific mechanisms of damage differ across these insults and extend to other aspects of the sensorineural apparatus (Hirose and Liberman, 2003; Kujawa and Liberman, 2009; Bielefeld and Henderson, 2011). DPOAEs offer a promising outcome measure for clinical detection and monitoring as OHC function cannot be directly measured in humans (Kramer et al., 2006; Kopke et al., 2015; Konrad-Martin et al., 2016). The evidence suggests that DPOAEs might be used as an early warning sign of NIHL as they appear to be more sensitive to cochlear damage compared to pure-tone hearing thresholds up to 8 kHz (reviewed by Marshall et al., 2001). Moreover, DPOAEs may be more sensitive than pure-tone audiometric thresholds in detecting the early stages of noise-induced inner-ear damage in humans. Typical results for noise-exposed groups followed longitudinally show a decrease in the DPOAE level without a change in the audiometric thresholds (e.g., Engdahl and Kemp, 1996; Lapsley Miller et al., 2006). Therefore, low-level or absent emissions in noise-exposed individual ears may be a predictor of hearing loss for continuous noise overlaid with impact noise (Lapsley Miller et al., 2006). However, limited information is available when exploring DPOAE measures for frequencies beyond 6–8 kHz; therefore, this investigation examined detailed DPOAE characteristics and behavioral hearing thresholds up to 20 kHz in individuals grouped based on self-reported noise exposure (occupational or recreation) as a focus.

DPOAEs, commonly measured at a limited number of stimulus frequency pairs per octave, can be recorded using more closely spaced stimulus frequencies to reveal a pattern of alternating maxima and minima in the DPOAE level and phase, known as fine structure (e.g., Heitmann et al., 1998; Talmadge et al., 1999; Dhar et al., 2002). This fine structure is largely understood to be an interference pattern between multiple source components of the DPOAE at 2ƒ₁-ƒ₂ (e.g., Talmadge et al., 1999). DPOAE fine structure has been at the center of clinical and research interests (refer to Poling et al., 2014, for an overview). Changes in the fine structure have been documented after acute noise exposure (Engdahl and Kemp, 1996), excessive salicylate consumption (Rao and Long, 2011), and cisplatin exposure (McMillan et al., 2012). Although the sensitivity of the fine structure characteristics to noise exposure has been mixed (e.g., Reuter and Hammershøi, 2007), there appears to be limited evidence of the clinical utility of the fine structure in a sizeable normal-hearing population capturing age-related changes in hearing at the highest frequencies of hearing (Poling et al., 2014). Furthermore, the current research focus appears to have shifted from the general fine structure to DPOAE components to eliminate the uncertainty related to the DPOAE fine structure (e.g., Abdala and Dhar, 2010, 2012; Rao and Long, 2011).

Two primary regions of the cochlea contribute to the 2ƒ₁-ƒ₂ DPOAE recorded in the ear canal in humans (e.g., Kim, 1980; Siegel and Kim, 1982) with the fine structure believed to arise from the interaction between these two components: one (overlap component; other terms include generator, nonlinear, and wave-fixed) from the region where the activity patterns of the stimulus tones overlap on the basilar membrane (near the f₂ region) and the other distortion product (DP) place component; other terms include reflection, linear, and place-fixed] from the characteristic frequency (CF) region of the 2ƒ₁-ƒ₂ DPOAE (e.g., Talmadge et al., 1999). Additional contributions to the DPOAE recorded in the ear canal come from areas basal to the f₂ CF region have also been indicated; however, the proportion of these basal contributions appears to be species specific, and notable contributions in humans are observed at high stimulus levels and low ( f₂) frequencies (Martin et al., 2011). Contributions from multiple internal reflections between the the DPOAE CF region and the basal boundary of the cochlea under certain stimulus conditions may also occur (Dhar et al., 2002). Hence, the DPOAE recorded in the ear canal is a complex product of various sources with assorted magnitude and phase characteristics.

The overlap component and DP place component exhibit significantly different phase behaviors as a function of the frequency. This difference in phase behavior has led to the development of models that hypothesize different generation mechanisms for the two components (Talmadge et al., 1998, 1999; Mauermann et al., 1999; Shera and Guinan, 1999). The phase of the overlap component is relatively invariant as a function of the frequency, whereas the phase of the DP place component shows a steep phase gradient with the frequency. This difference in phase characteristics between the two components is ultimately responsible for the observation of the fine structure as the two components cycle in and out of phase as a function of the frequency, resulting in alternating peaks and valleys. The diverse phase behavior of the two components is also exploited to separate the two DPOAE components from the ear-canal recorded signal (Talmadge et al., 1999; Kalluri and Shera, 2001; Dhar et al., 2002; Shaffer and Dhar, 2006). Source separation permits closer examination of clinical utility, and in prior investigations, the DP place component proved successful in exposing age effects and potential sensitivity to detection of small changes in cochlear function (e.g., Poling et al., 2014).

The purpose of this investigation was to explore the effects of self-reported noise exposure on various measures of auditory function in individuals between the ages of 10 and 65 years old with clinically normal/near-normal hearing through 4 kHz. In this sample and based on self-reported noise exposure (i.e., occupational or recreational exposure), the hearing thresholds and DPOAE level, fine structure, and component characteristics up to 20 kHz were examined.

II. METHODS

A. Subjects

From a sample of 356 individuals ranging in age from 10 to 65 years old, two subgroups were identified based on the self-reported questionnaire measures of occupational or recreational noise exposure. Specifically, the occupational noise-exposed group (n = 84) comprised those individuals that answered “yes” to the following question: Do you now or have you ever worked in a noisy environment (e.g., machine shop, firearms, musicians, using power equipment such as lawn mowers)? The recreational noise-exposed group (n = 46) comprised those individuals that answered “yes” to the following question: Do you have any hobbies or are you engaged in activities which involve being in noisy environments or using loud machinery (e.g., power saws, lawn mowers, firearms)? Matched pairs were created for the noise-exposed group from the (control) individuals reporting no history of occupational and recreational noise exposure (i.e., answered “no” to both study questions). The control subjects were sex- and age-matched (most cases were exact age-matched; two cases within 5–10 years) to the noise-exposed individuals, resulting in the breakdown detailed in Table I. From the two-item questionnaire, we identified a subgroup of “combined” noise-exposed groups in which the individuals that answered “yes” to both questions were excluded from the two defined groups here (n = 26). The results presented here were part of a larger population study described previously (see Lee et al., 2012, for the behavioral hearing thresholds and Poling et al., 2014, for the DPOAEs).

TABLE I.

The breakdown of subjects in the noise-exposed and age-matched control (nonexposed) groups for occupational exposure and recreational exposure. The sex distribution of the subjects is presented in parentheses. SD, standard deviation.

Exposure	Noise-exposed	Mean age (SD)	Age-matched controls	Mean age (SD)
type	(M/F)	(yr)	(M/F)	(yr)
Occupational	84	37.6 (13.3)	84	37.7 (13.7)
	(50/34)	(18–63)	(50/34)	(18–64)
Recreational	46	37.0 (14.8)	46	37.0 (14.7)
	(28/18)	(16–65)	(28/18)	(18–65)

All of the subjects had hearing thresholds equal to or less than 20 dB HL (hearing level; ANSI, 1996) from 0.25 to 4 kHz (refer to Fig. 1). Immittance measures and clinical audiometry (0.25–8 kHz in each ear) were performed using an Interacoustics AA220 Audiometer and Middle Ear Analyzer (Interacoustics USA, Eden Prairie, MN). All of the testing was conducted in one of two sound-treated audiometric booths anf each met the current maximum allowable ambient noise standard (ANSI, 2008). The measurements were made in one ear chosen at random with a visible and healthy tympanic membrane as evaluated by otoscopy. The subjects were paid for their participation and all of the measurements were conducted in accordance with the guidelines of the Institutional Review Board at Northwestern University.

FIG. 1.

The mean hearing thresholds as a function of the frequency for the occupational (A) and recreational (B) noise-exposed (filled symbols) groups compared to the age-matched control (nonexposed) groups (open symbols). The conventional audiometric findings (0.25–8 kHz in dB HL; right axis) as well as the extended high-frequency tracking thresholds (0.125–20 kHz in dB SPL; left axis) are displayed with the inset centered on the 3–6 kHz region. The error bars represent 95% CI.

B. Calibration

A depth-compensated ear simulator calibration technique (described in Lee et al., 2012; Poling et al., 2014) was used to control the stimulus level at the eardrum. Specifically, stimuli used for the behavioral hearing threshold tracking technique and swept-frequency DPOAE measurements were calibrated using an ear simulator (i.e., IEC 60318-4, IEC, 2020) calibration procedure that allowed for the approximate compensation of the depth of insertion in the ear canal. Detailed comparisons of the calibration methods are described in Souza et al. (2014).

C. Hearing threshold tracking measurements

Threshold values were obtained for 21 standard audiometric frequencies between 0.125 and 20 kHz using a modified Békésy technique (detailed in Lee et al., 2012). In short, stimulus delivery was achieved using custom software, built on Max/MSP,¹ with pulsed tones, 250 ms in duration with 25 ms rise and fall times, and presented twice per second. The adaptive threshold task required the subject to press and hold a computer mouse button to indicate that the pulsed tone was heard. As the tone decreased in presentation level, the subject was asked to release the button when the tone was no longer heard, which initiated the presentation level to increase. The frequencies were presented in a fixed order, starting at 1 kHz, proceeding to the highest frequency (20 kHz), repeating the measurement at 1 kHz, and then proceeding downward to the lowest test frequency (0.125 kHz). If a subject did not respond to the tone at the output limit of the equipment, the threshold was recorded as the maximum output for that frequency.

D. DPOAE measurements

The signal generation and recordings for the DPOAE measurements were performed using custom software (developed by C. Talmadge) on an Apple Macintosh computer (Cupertino, CA). A MOTU 828 MkII input/output FireWire device (Cambridge, MA) was used for analog-to-digital (44.1 kHz, 24-bits) and digital-to-analog conversion. Generated signals were sent through a custom headphone amplifier to custom-built sound sources containing modified MB Quart 13.01HX tweeters (Libertyville, IL). The sound sources were coupled to the subjects' ear canal via 16 gauge plastic tubing connected to an Etymotic Research ER10B+ (Etymōtic Research Inc., Elk Grove Village, IL) probe and was sealed into the canal using a 13 mm foam ear tip. Signals from the test ear were recorded using the ER10B+ microphone and preamplifier combination, digitized using the MOTU, and stored on disk for analysis.

The DPOAEs were recorded for a fixed stimulus frequency ratio (ƒ₂/ƒ₁) of 1.22 for 2ƒ₁-ƒ₂ frequencies between 0.5 and 10.24 kHz (i.e., ƒ₂ between ∼0.75 and 16 kHz). The DPOAE recordings were obtained at three stimulus-level combinations [L₁ and L₂ in dB SPL (sound pressure level)]: 55/40, 65/55, and 75/75. Stimulus tones were swept at 8- and 24-s/octave for the 2ƒ₁-ƒ₂ frequencies below and above 6 kHz with a 2-s silent interval between sweeps, which is consistent with previous work from this laboratory. At least six sweeps were averaged before using a least-squares-fit (LSF) procedure (Long and Talmadge, 1997; Talmadge et al., 1999; Dhar et al., 2002, 2005; Long et al., 2008) to estimate the level and phase of the DPOAE at the frequency 2ƒ₁-ƒ₂. The initial analyses resulted in a frequency resolution between 0.002 and 0.02 kHz at the lowest and highest frequencies. The noise floor was estimated from an average of the sweeps with every alternate sweep inverted in phase. The median DPOAE level was computed for every three consecutive frequencies and compared against the noise floor estimate for the corresponding center frequency to determine the signal-to-noise ratio (SNR). All data points that did not meet a 6 dB SNR criterion for the three-point median reference were eliminated from further analyses to ensure accurate quantification of the DPOAE fine structure (detailed in Poling et al., 2014).

The DPOAE fine structure features were defined as the number of fine structure periods per 1/3-octave, depth, and spacing. The maxima and minima in the DPOAE level were identified by the zero crossings of the first derivative of the level function. The distinction between the maxima and minima was made based on the sign of the second derivative of the level function at the zero crossing of the first derivative (see Dhar et al., 2002, for details). The fine structure depth for each period bounded by two maxima or minima was computed as the emission amplitude at the maximum divided by the average amplitude of the two surrounding minima. Only fine structure periods where the depth was greater than 2 dB were included in further analyses. The fine structure spacing was quantified by computing the geometric mean ( f) between two adjacent minima and dividing by the frequency separation between them (Δf), which is referred to as the spacing ratio—f/Δf (Shera, 2003). Only spacing ratios between 4 and 32 (1/3- to 1/20-octave) were included in further analyses. The number of fine structure periods per 1/3-octave was also calculated. The fine structure parameters were averaged by 1/3-octave bands centered at audiometric test frequencies (0.75–16 kHz) with the detailed population statistics available in Poling et al. (2014).

E. Separation of DPOAE components

The components (overlap and DP place) contributing to the DPOAE signal in the ear canal were separated based on the group delay from the composite estimate (without additional data processing for the SNR) using an inverse fast Fourier transform (IFFT) in a custom matlab analysis program (The MathWorks, Natick, MA; see Abdala and Dhar, 2012, for details). Once isolated, these individual components were processed through a regular fast Fourier transform (FFT) algorithm to obtain independent estimates of the magnitude. The DPOAE complex pressure in the frequency domain was multiplied by a moving Hann window in overlapping 0.05 kHz steps. The Hann window length was adjusted on a logarithmic scale to account for the cochlear frequency-place map (Greenwood, 1990) and ranged from 0.4 kHz at the lowest to 1.623 kHz at the highest DPOAE frequency. The impulse response (IPR) functions were derived for each window and rectangular time-domain filters were applied to each IPR to extract the DPOAE components with different delays. Specifically, the short-latency (overlap) component was identified as a peak in the amplitude function with two minima within a time window between −2 and 10 ms. Similarly, the long-latency (DP place) component was identified in a time window between 8 and 15 ms. The filtered windows of data were then transformed back to the frequency domain by the FFT, and the levels and phases of the overlap and DP place components were reconstructed.

Additionally, the data at the extreme low and high frequencies were eliminated because of artifactual edge effects inherent to the time-windowing process. The IFFT-derived magnitude estimates for the overlap and DP place components were averaged into 1/3-octave bins with these qualification criteria resulting in the data being limited to an ƒ₂ range of 1–15 kHz (as opposed to 0.75 and 16 kHz for the fine structure analysis). Additionally, 1/8-octave bins were examined and reported for the difference scores calculated by subtracting the estimates for the control group from the noise-exposed groups for each component (overlap and DP place) as well as the composite DPOAE estimates.

F. Statistical procedures

All analyses were performed using sas version 9.2 (Cary, NC). In Figs. 1–4, the 95% confidence intervals are reported in lieu of the standard deviation (SD) or standard error as they are more amenable to clinical comparisons across groups. The esults were averaged by 1/3-octave bands centered around audiometric test frequencies (0.75–20 kHz) for comparison across the test measures unless otherwise noted. A mixed effects model with the group × frequency interaction term was used to compare the mean hearing thresholds and DPOAE measures (level and fine structure characteristics) between the noise-exposed and control matched pairs (not adjusted for multiple comparisons). The mixed effects model takes into account the correlations among the repeated measures across the frequency while accounting for matched pairs (at fixed stimulus level conditions). If the interaction term was not significant, the results from the main effects model are reported. The overall significance level was held at 0.05.

FIG. 2.

(Color online) The DPOAE fine structure characteristics for the occupational noise-exposed (first column) and recreational noise-exposed (second column) groups. The (A; row 1) average number of fine structure periods per 1/3-octave as a function of the frequency for the noise-exposed group (filled symbols) and control group (unfilled symbols), (B; row 2) average fine structure spacing as a function of the frequency, and (C; row 3) average fine structure depth as a function of the frequency are shown. The error bars represent 95% CI. The missing error bars signify cases where there is only one data point.

FIG. 3.

The mean DPOAE composite and component levels as a function of the frequency for the occupational noise-exposed (gray) and age-matched controls (black). The three stimulus level conditions of 75/75, 65/55, and 55/40 dB SPL are displayed in the three rows. The error bars represent 95% CI. The average noise floor estimates were similar across the age groups (never exceeding -15 dB SPL) and are not included here for clarity.

FIG. 4.

(Color online) The difference scores (occupational noise-exposed group − control DPOAE level) as a function of the frequency for the composite DPOAE composite, overlap, and DP place component (same subset as that represented in Fig. 3). The three stimulus level conditions of 75/75, 65/55, and 55/40 dB SPL are displayed in the three rows. The error bars represent 95% CI.

III. RESULTS

The focus of this investigation was to examine the effects of self-reported noise exposure on various measures of auditory function in individuals between the ages of 10 and 65 years old with clinically normal/near-normal hearing through 4 kHz (defined as hearing thresholds ≤20 dB HL through 4 kHz). The hearing thresholds and 2ƒ₁-ƒ₂ DPOAE level, fine structure, and component characteristics up to 20 kHz were examined in a normal-hearing population that was grouped based on self-report questionnaire measures related to occupational or recreational noise exposure (see Sec. II, Methods). Specifically, conventional audiometry (up to 8 kHz), extended high-frequency audiometry using a threshold tracking procedure (up to 20 kHz), DPOAE level, DPOAE fine structure (number of periods per 1/3-octave, depth, spacing), and the magnitude and phase of individual DPOAE source components were explored.

A. Hearing thresholds

Hearing thresholds were assessed using two methods in this study. First, conventional pure-tone hearing thresholds were obtained through 8 kHz with the inclusion criteria that each participant have thresholds of 20 dB HL or better from 0.25 to 4 kHz (results obtained with standard clinical audiometry approaches versus a threshold tracking procedure for direct frequency comparisons). Second, extended high-frequency hearing thresholds up to 20 kHz were obtained in the same test ear using a threshold tracking procedure only (see Sec. II, Methods). Figure 1 displays the mean hearing thresholds (dB HL, right axis; dB SPL, left axis) for the occupational noise-exposed group (gray; filled circles) compared to the control group (black; open circles) [Fig. 1(A)] and the recreational noise-exposed versus the control group [Fig. 1(B)]. The box insets center on the 3–6 kHz test region capturing the typical “noise-notch” region of the audiogram.

Overall, similar hearing thresholds were observed for each noise-exposed (occupational or recreational) and control group comparison. For the conventional hearing thresholds obtained with standard clinical audiometry (Fig. 1; 0.25–8 kHz only), a statistically significant difference in the hearing thresholds between the occupational noise (poorer) and control group was observed at 3 kHz (p = 0.0204) and 4 kHz (p = 0.0117). A similar trend was noted in the recreational noise group as well as for the threshold tracking data as observed in Fig. 1; however, this did not reach significance. No difference was observed at the corresponding frequencies when a threshold tracking approach (for the full bandwidth of 0.125–20 kHz) was used. A statistically significant difference between the occupational noise-exposed and control groups was only observed at 11.2 kHz (p = 0.0016) and 12.5 kHz (p = 0.0249). Specifically, the exposed group demonstrated lower (better) thresholds than the control group at these frequencies.

B. DPOAE fine structure characteristics

The DPOAE fine structure characteristics of the number of periods, depth, and spacing were examined as a function of the ƒ₂ frequency by group (occupational, recreational, or control) at each stimulus level combination (75/75, 65/55, and 55/40 dB SPL). In general, the number of periods observed per 1/3-octave was similar across all frequencies for all of the stimulus level combinations. Moreover, there was a similar number of periods in the noise-exposed groups compared to the control group. For the occupational noise group, the average number of periods for 55/40 dB SPL was 1.63 (1.62 for the control group), whereas for 65/55 dB SPL, it was 1.60 (1.55 for the control group), and for 75/75 dB SPL, it was 1.53 (1.50 for the control group). For the recreational noise group, the average number of periods for 55/40 dB SPL was 1.61 (1.62 for the control group), whereas for 65/55 dB SPL, it was 1.55 (1.56 for the control group), and for 75/75 dB SPL, it was 1.53 (1.52 for the control group).

Overall, similar spacing ratios ( f/Δf) were observed across groups and stimulus levels; however, there was a trend for the spacing ratio to increase with an increasing frequency. For the occupational noise group, the average spacing ratio for 55/40 dB SPL was 18 (18 for the control group), whereas for 65/55 dB SPL, it was 18 (18 for the control group), and for 75/75 dB SPL, it was 18 (17 for the control group). For the recreational noise group, the average spacing ratio for 55/40 dB SPL was 18 (18 for the control group), whereas for 65/55 dB SPL, it was 18 (18 for the control group), and for 75/75 dB SPL, it was 18 (17 for the control group). As observed in prior work, sharper (higher) spacing estimates were observed in the higher frequencies compared to those in the lower frequencies.

The depth estimates were similar across groups and stimulus levels with a trend for an increasing depth with an increasing frequency as seen in prior work. For the occupational noise group, the average fine structure depth (dB) for 55/40 dB SPL was 5.30 (5.10 for the control group), whereas for 65/55 dB SPL, it was 4.77 (4.44 for the control group), and for 75/75 dB SPL, it was 4.86 (4.72 for the control group). For the recreational noise group, the average fine structure depth (dB) for 55/40 dB SPL was 5.16 (5.20 for the control group), whereas for 65/55 dB SPL, it was 4.61 (4.31 for the control group), and for 75/75 dB SPL, it was 4.87 (4.55 for the control group).

Similar DPOAE fine structure features were observed for the noise-exposed and control groups, which were further confirmed by no significant difference between the groups when the results were averaged over the frequency. Figure 2 is provided as an overview of the DPOAE fine structure characteristics for the exposed (occupational, left panels; recreational, right panels) and matched control groups. Of note, this exploration is a subset of our larger results shared in Poling et al. (2014).

C. DPOAE component analysis

The ear-canal composite DPOAE was separated into the components from the overlap and DP place regions, and the magnitude and phase estimates of each component were examined separately. To ensure the validity of the results of the IFFT procedures, only the data sets where information was available at least every 0.006 kHz were included. The larger sample of the occupational noise-exposed group was used for component analysis only. The average DPOAE level as a function of the frequency is shown in Fig. 3 for the three stimulus level combinations (75/75, 65/55, and 55/40 dB SPL in separate rows) for each group. Specifically, the level of the composite DPOAE and two separated components (overlap and DP place) for the occupational noise-exposed group are shown in Fig. 3, and the difference scores between the exposed and control groups are addressed below. Although the phase characteristics were examined, the results are not presented here due to space limitations and the fact that there was no difference between the exposed and control groups on preliminary analysis of the individual phase data (the results are similar to those in the full population presented in Poling et al., 2014).

Figure 3 displays the mean composite DPOAE and separated component levels as a function of the ƒ₂ frequency for the occupational noise-exposed (gray) and control (black) groups. The measurements made using different stimulus level combinations are presented in the three rows. A mixed effects model, accounting for the matched pairs outcome by each stimulus level at each frequency, was used to examine the composite and component DPOAE levels. Pairwise comparisons for the composite and component characteristics at each stimulus level were examined to explore the differences between groups at each frequency. Collapsed across the frequency, significant differences between the control and noise-exposed groups were observed for all conditions: composite (75/75 dB SPL, p = 0.0005; 65/55 dB SPL, p = 0.003; 55/40 dB SPL, p ≤ 0.0001), overlap component (75/75 dB SPL, p = 0.0022; 65/55 dB SPL, p = 0.0135; 55/40 dB SPL, p ≤ 0.0001), and DP place component (75/75 dB SPL, p = 0.0003; 65/55 dB SPL, p = 0.0221; 55/40 dB SPL, p = 0.0014). The composite and individual component DPOAE levels increased with an increasing stimulus level; however, the lowest stimulus level (55/40 dB SPL) demonstrated the most sensitivity to noise exposure for the composite and individual components.

The pairwise comparisons revealed statistically significant differences between the groups at traditionally noise-notch frequencies (i.e., 3–6 kHz) with the exposed groups demonstrating reduced amplitudes compared to those in the control group. Statistically significant differences at 3 kHz (p = 0.0207) for 75/75 dB SPL and 6 kHz (p = 0.0039) for 55/40 dB SPL were observed for the composite DPOAE. For the overlap component, differences were observed at 3 kHz (p = 0.0249) for 75/75 dB SPL and 6 kHz (p = 0.0031) for 55/40 dB SPL. In the case of the DP place component, which by definition is believed to reflect the CF place, statistically significant differences were obtained at 4 kHz for 65/55 dB SPL (p = 0.029) and 55/40 dB SPL (p = 0.0014).

D. Difference scores

To explore whether the contribution of noise exposure to the composite DPOAE as well as the two components changed as a function of the stimulus level or frequency, the difference between the occupational noise and control groups was calculated in 1/8-octave bin frequencies. The mean difference scores as a function of the frequency for the composite and individual components for the three stimulus levels are shown in Fig. 4. Difference scores of zero indicate no difference in the level between the two groups. Positive and negative difference scores indicate larger amplitudes and reduced amplitudes for the occupational noise-exposed group, respectively. For the most part, reduced amplitudes for the noise-exposed group relative to the control group are noted for all conditions, yet they do not reach statistically significant differences. The difference in the level between the two groups was most pronounced in the noise-sensitive frequency region between approximately 3 and 6 kHz and at the lowest stimulus level (55/40 dB SPL). From the mixed effects model, there were no significant effects of the stimulus level or frequency on the difference score except between 3 and 4 kHz (p = 0.0205) for the DP place component within the 55/40 dB SPL condition. Although there is a trend for a similar noise-notch with an increasing stimulus level as well as within the composite and overlap component data, this result does not denote a clinically significant indicator between the two groups.

IV. DISCUSSION

NIHL is a growing health concern accompanied by a global call to action for limiting noise exposures across the life course (WHO, 2021). Potentially damaging exposures to noise may go unnoticed by the individual as they are not often painful and may not result in immediate or long-term measurable changes in auditory function, especially for behavioral hearing thresholds screened for in the clinic. The conventional (up to 8 kHz) behavioral hearing threshold assessment following overexposure to noise may reveal threshold shifts that return to pre-exposure levels (referred to as “temporary” threshold shift; TTS) or those that remain at an elevated level relative to pre-exposure (referred to as “permanent” threshold shift; PTS). Physiological evidence of overexposure is characterized by hair cell death that can occur immediately following exposure and continue for days (e.g., Wang et al., 2002) in conjunction with damage to spiral ganglion cells that can be delayed by months and continue for years (Kujawa and Liberman, 2006, 2009). Moreover, growing physiological evidence (Kujawa and Liberman, 2006, 2009; Lin et al., 2011) suggests that experimentally induced TTS may result in permanent, degenerative changes to the auditory nerve even when auditory thresholds and DPOAEs return to pre-exposure levels. For a recent review of the translating noise exposure risk from animal models into human domains, refer to Le Prell et al. (2019).The utility of DPOAEs has long been of interest for scientific and clinical purposes in measuring and monitoring noise exposure as changes in the level (e.g., Marshall et al., 2001) and fine structure (e.g., Engdahl and Kemp, 1996) have been documented after acute noise exposure. More recent focus has been given to the vulnerability of individual DPOAE components to noise, although limited (e.g., Torre et al., 2003). The argument in favor of focusing on DPOAE components is to minimize the uncertainty related to the DPOAE fine structure and potentially maximize the clinical utility. The presence of the fine structure and its depth are dependent on the relative levels of the two components. Therefore, changes in this level relationship could result in either an increase or decrease in the fine structure depth, making it an unreliable clinical metric (e.g., Poling et al., 2014). Similarly, the phase characteristics between the two components as they cycle in and out of phase as a function of the frequency result in alternating peaks and valleys (fine structure) with cancellations in the middle. Observations in our results of a possibly increased DPOAE level and fine structure around the noise-notch region may reflect a release from this cancellation due to the fact that some of the contributions in the “notch frequency region” are not elicited at those stimulus levels as a result of possible noise-injury (loss) or contribute less. Moreover, alternate methods of component segregation or segregating components over more limited frequency ranges are worth consideration. The clinical utility of the DPOAE components will improve with the discovery and refinement of these strategies in vulnerable populations such as noise-exposed populations.

Among all of the features of the DPOAE level, fine structure, and components that were evaluated, the one with the most potential as an indicator of noise exposure beyond what is established for the DPOAE level was the DP place component level difference scores, particularly at 55/40 dB SPL. This lower stimulus level combination, thought to be ideal for detecting small changes in cochlear function, also demonstrated prominent DPOAE level differences. This adds to the growing evidence that the DPOAE level and DP place component level [i.e., stimulus frequency otoacoustic emissions (SFOAEs) by extension] may be viable tools which are sensitive to early changes in cochlear function resulting from noise exposure.

Although not fully understood in the literature, “notched audiograms,” defined as conventional behavioral audiograms with elevated air and bone thresholds at 3, 4, and 6 kHz than the adjacent 2 and 8 kHz, have long been associated as a clinical indicator of NIHL (see the detailed review by Wilson, 2011, and Wilson and McArdle, 2013). However, the relationship between the notched audiogram and noise exposure has been confounded by the complex interrelationship between noise exposure and aging (e.g., Dobie, 2008) as well as the complexities of defining the self-reporting of noise exposure (e.g., Gates et al., 1990; Nondahl et al., 2009). It can be difficult to establish a causal relationship between noise exposures and hearing threshold outcomes as the self-report histories are often only one of many variables that may be contributing to dysfunction. Furthermore, there is growing evidence that identical noise exposures have differential effects on individual humans and animal models (Kujawa and Liberman 2006, 2009; Lin et al., 2011). The full clinical utility of this is a focus of ongoing investigations.

Conventional hearing thresholds (0.25–8 kHz) used to represent a standard clinical assessment revealed a statistically significant difference at 3 and 4 kHz, aligning with a noise-induced influence in the outcome. This is surprising given the relatively strict inclusion criteria for the large study population, which was defined as having clinically normal/near-normal hearing thresholds (defined as ≤20 dB HL) through 4 kHz, as well as the fact that the study design was not aimed at characterizing noise exposure (e.g., based on two yes/no questions about noise exposure). However, no significant differences in the hearing thresholds between the noise-exposed and control groups at the same frequencies were observed with the threshold tracking method used. This result supports the importance of advanced in-ear calibrations to minimize variability, which may lead to interpretations of NIHL that resolve with improved calibration (i.e., not a true threshold shift but a calibration side effect).

Extended high-frequency (>8 kHz) hearing threshold assessments hold much clinical value for age-related changes in hearing thresholds as they first appear at frequencies above 8 kHz (Bunch, 1929; Rosen et al., 1964; Fletcher, 1965; Northern et al., 1972; Ahmed et al., 2001). However, the value of using extended high-frequency hearing thresholds to monitor noise-induced changes to the auditory periphery is not clear. Some reports have found no change in hearing thresholds above 8 kHz, concurrent with measurable shifts in thresholds between 2 and 8 kHz (Osterhammel and Osterhammel, 1979; Laukli and Mair, 1985). In contrast, others have reported thresholds >8 kHz to be equally or more sensitive to noise damage compared with thresholds below 8 kHz (Sataloff et al., 1967; Corliss et al., 1970). Fausti et al. (1979) found thresholds >8 kHz to be altered before any observable changes in thresholds below 8 kHz in cases of chronic exposure to hazardous levels of noise. The additive effects of NIHL and age-related hearing loss have also been documented in the same age range as reported in this study. Moreover, growing evidence suggests that the specific type of noise exposure may guide interpretation with attention given on military noise exposures, which may also include impulse noise (reviewed by Moore, 2020; Büchler et al., 2012; Wilson and McArdle, 2013). Although not extensively studied, the consensus of published reports seems to be that age-related mechanisms dominate the changes >8 kHz, whereas noise-induced changes are more prominent at frequencies below 8 kHz (Morton and Reynolds, 1991; Ahmed et al., 2001; Moore, 2020). The current findings are consistent with the associated noise-induced changes observed in standard clinical audiometry being more prominent below 8 kHz. However, there were statistically significant differences between the data observed with a tracking threshold at 11.2 and 12.5 kHz such that the exposed group demonstrated lower (better) thresholds than the control group at these frequencies. Given the limited and mixed findings in the literature to date, it is difficult to assign meaning to this interesting finding. The advanced calibration methods do not eliminate all of the influences of standing wave errors, which would be more prominent at higher frequencies (e.g., Souza et al., 2014). Although it is difficult to assess in this study, it may be of value to consider if serial measurements with these controlled approaches focused on this frequency region may have some clinical utility. For this evaluation, there was no direct comparison at the same extended high frequencies between those hearing thresholds obtained with standard clinical approaches versus threshold tracking approaches.

V. CONCLUSIONS

Limited information is available when exploring DPOAE measures (including fine structure features and components) for frequencies beyond 6–8 kHz. This investigation examined detailed DPOAE characteristics and behavioral hearing thresholds up to 20 kHz in individuals grouped based on self-reported noise exposure. By documenting the DPOAE level, the fine structure and component characteristics in a sample of clinically normal/near-normal hearing individuals, we aimed to provide a baseline for future studies exploring the auditory effects of ototoxic noise exposure relative to those without exposure. The extension to frequencies beyond 8 kHz and use of several stimulus levels under carefully calibrated conditions (e.g., Souza et al., 2014; Poling et al., 2014) support the clinical utility of hearing thresholds and DPOAE levels when measured at multiple stimulus levels, whereas the fine structure characteristics offer limited use in a specialized risk population (aligning with Poling et al., 2014, and Poling et al., 2019). Moreover, this work supports the potential clinical utility of the DP place component level for revealing differences in cochlear function due to noise exposure that is not observable in behavioral hearing thresholds. That said, this warrants further investigation in populations specifically targeted for specialized exposures to fully explore the clinical utility of such a measure where specific considerations for ototoxic noise effects on cochlear function in the region represented by the DP place component can be explored.