High-Frequency Distortion-Product Otoacoustic Emission Repeatability in A Patient Population

High-Frequency Distortion-Product Otoacoustic Emission
Repeatability in a Patient Population

Laura Dreisbach1, Erika Zettner2, Margaret Chang Liu1, Caitlin Meuel Fernhoff1,
Imola MacPhee1, and Arthur Boothroyd1,2
Objectives: Distortion-product otoacoustic emissions (DPOAEs) are the frequency of 2f1−f2. DPOAE generation involves the active
repeatable over time at lower frequencies (≤8 kHz) and higher frequen- motile responses of outer hair cells (OHCs), which are respon-
cies (>8 kHz) in healthy, normal-hearing subjects. The purpose of this sible for basilar membrane–tuning properties (Ruggero & Rich
study was to examine the repeatability of DPOAEs measured with high- 1991) and the sensitivity and frequency selectivity observed in
frequency (HF) stimuli in a patient population. It was hypothesized that
the tips of neural tuning curves (Liberman & Dodds 1984). The
HF DPOAEs would be repeatable over four trials.
reduction in level or elimination of emissions in a particular
Design: DPOAEs were measured in 40 cystic fibrosis (CF) patients (17 frequency range has been shown to occur with OHC damage in
females and 23 males) with measurable behavioral thresholds and pres- that cochlear region (Brown et al. 1989).
ent DPOAEs for at least 2 of the high frequencies tested (8 to 16 kHz). DPOAEs can be elicited by varying the frequencies (f1 and
A depth-compensated simulator sound pressure level (SPL) method
f2), frequency ratio (f2/f1), or levels (L1 and L2) of the evoking
of calibration was utilized. Each patient attended four trials, in which a
complete set of data were collected. At each trial, three different DPOAE
stimuli, or primary tones. Traditionally, discrete frequency
paradigms were completed. First, a discrete frequency sweep was mea- sweeps, where the frequencies of the primary tones are varied
sured between 8 and 16 kHz with a ratio (f2/f1) of 1.2 and levels of 65/50 with a fixed ratio and levels, are measured in clinical settings.
dB SPL for L1/L2. Next, ratio and level sweeps were obtained at the two However, the ratio and levels of the primary tones also can
highest frequencies with a present DPOAE determined from the discrete be altered at various frequency regions. When the ratio of the
frequency sweep, and the results were used to calculate DPOAE group primary tones is varied over a small-frequency region and the
delay and DPOAE detection thresholds, respectively. Ratio sweeps were stimulus levels are fixed, a group delay value can be calculated
collected with f2/f1 varied from 1.1 to 1.3 and stimulus levels of 60/45 dB from the resultant DPOAE phase. Group delay is related to the
SPL (L1/L2). Level sweeps were collected with an f2/f1 of 1.22 and L2 = 50
travel time along the basilar membrane, and it is expected that
and L1 varied between 20 and 70 dB SPL. Differences and correlations
between trials, SE of the measurement, and confidence intervals were
higher frequency primary tones will yield shorter group delay
calculated, as well as a repeated-measures analysis of variance. values than lower-frequency primary tones, due to the tonotopic
nature of the basilar membrane. If the levels of the primary
Results: DPOAE response and behavioral threshold variability in CF tones are varied, DPOAE growth can be examined, and a detec-
patients were not significantly different across four trials. It can be tion threshold can be determined. The primary tone levels can
expected in 95% of CF patients that differences between trials of DPOAE
vary together equally (L1 = L2), at a fixed difference (e.g., L1 = L2
levels, group delay, and detection thresholds and behavioral thresholds
are less than 6.26 dB, 0.87 msec, 9.34 dB, and 9.60 dB, respectively. − 10 dB), or one level can be fixed while the other level is varied
(Gaskill & Brown 1990; Stover et al. 1996). Threshold can be
Conclusions: HF DPOAEs were repeatable across four test trials for all defined as the lowest level of the primary tones that evoke an
three paradigms measured in a group of CF patients. These results are emission that meets specified criterion values.
encouraging for the measurement of HF DPOAEs to be monitored in
Regardless of which stimulus parameters are varied to elicit
those exposed to ototoxic agents.
a DPOAE, the result provides an objective measure of cochlear
Key words: Cystic fibrosis, Detection thresholds, Distortion-product oto- status without requiring a patient response, thus allowing for
acoustic emissions, Group delay, High-frequency stimuli, Monitoring, efficiency, elimination of tester or response bias, and bedside
Ototoxicity, Repeatability. administration in unhealthy patients (Ress et al. 1999). Common
(Ear & Hearing 2017;XX;00–00) clinical uses for DPOAEs include hearing screenings, differen-
tial diagnosis, and serial monitoring of cochlear damage due to
INTRODUCTION ototoxic agents (noise or medications). Because many damag-
ing agents initially target OHCs, monitoring DPOAEs appear
When two continuous acoustic pure tones, close in fre- ideal for assessing ototoxicity (Roland 2004; Reavis et al. 2008,
quency (f1 and f2), are simultaneously presented to the cochlea, 2011; Dille et al. 2010). Because a physiological change may be
acoustic distortion products at frequencies not present in the detected before a perceptual difference is noticed by the patient,
stimuli are produced due to the nonlinear properties of the basi- monitoring DPOAEs may allow for earlier detection of cochlear
lar membrane and can be measured in the ear canal, namely, dis- damage (Katbamna et al. 1999a,b; Ress et al. 1999).
tortion-product otoacoustic emissions (DPOAEs; Kemp 1978). Patients who require audiologic monitoring are usually
The frequencies of DPOAEs are arithmetic combinations of the exposed to ototoxic agents for a prolonged period of time,
lower (f1) and higher frequency (f2) stimulus tones, and the most and monitoring should be carried out routinely throughout the
prominent and widely studied DPOAE in humans occurs at duration of exposure (American Speech-Language-Hearing
School of Speech, Language, and Hearing Sciences, San Diego State
1 Association 1994; American Academy of Audiology 2009). An
University, San Diego, California, USA; and 2School of Medicine, effective monitoring tool, therefore, must be reliable across test
University of California, San Diego, California, USA. sessions over short and long periods of time, to ensure that a
0196/0202/17/XXXX-0000/0 • Ear & Hearing • Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved • Printed in the U.S.A.
1
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
<zdoi; 10.1097/AUD.0000000000000465>
2 DREISBACH ET AL. / EAR & HEARING, VOL. XX, NO. XX, 00–00
change in results is due to actual cochlear alterations rather than depth can change the SPL at the eardrum by up to 20 dB
measurement error or test–retest variability. Beattie and Bleech (Neely & Gorga 1998). Thus, alternate calibration methods
(2000) examined DPOAEs in an immediate test–retest trial (no have been developed to compensate for these potential stand-
probe removal) at 0.5, 1, 2, and 4 kHz and found that changes ing wave influences. One such method is the depth-compen-
must exceed approximately 6 dB to be statistically significant. sated simulator SPL calibration technique, which utilizes
Beattie et al. (2003) reported repeatability of DPOAEs over short corrected stimulus levels to account for half-wave resonances
periods of time (same day or within 5 to 10 days) and suggested of individual ear canals (Gilman & Dirks 1986). This method
that a change must exceed 14 dB at 0.55 kHz and 7 dB between has been successfully employed in a large cohort of subjects
1 and 4 kHz to be considered significant. Wagner et al. (2008) for the measurement of behavioral hearing thresholds up to
reported that changes of 0.7 to 1.25 or 1.8 to 3.1 dB between 20 kHz (Lee et al. 2012) and when characterizing DPOAEs
DPOAE measurements would be considered significant for the up to (f2) 16 kHz (Poling et al. 2014). If alternate calibration
1 to 5 kHz range for testing occurring on the same day without methods minimize differences in probe placement across test
probe removal or across 5 to 10 days, respectively. These values sessions, test–retest repeatability may be expected to improve
were slightly higher when examining 6 kHz, with significant (Souza et al. 2014).
differences of 2.3 dB for same day testing or 3.7 dB for multiple Another potential contributing factor to test–retest variabil-
days of testing. Ng and McPerson (2005) also reported greater ity is the condition of the patient. Given audiologic monitoring
variability for higher frequency measurements made 20 minutes is often performed in unhealthy patients undergoing aminogly-
or 15 days apart. They reported that differences of or greater coside or chemotherapeutic treatment, repeatability of emis-
than 5.3 or 6 to 8 dB would constitute a significant change for sions in this population also needs to be established to ensure
frequencies of 1 to 6 and 6.5 to 7 kHz, respectively. Reavis et al. that changes in emission levels are of clinical significance rather
(2015) completed a meta-analysis, increasing the sample size, than attributed to the patient’s general health status.
with data from 10 published studies with the goal of increasing The goal of ototoxic monitoring is to optimize the earli-
the accuracy for estimating DPOAE test–retest change expecta- est detection of damage in the hearing and balance organs. In
tions. They used the SEM, which is given by s·√1 − r, where s is general, ototoxic medications initially affect the basal end of
the standard deviation (SD) of combined baseline and follow-up the cochlea and progress apically, thus emphasizing audiologic
measurements and r is Pearson correlation coefficient between assessment in the ultra-high-frequency (HF) range (>8 kHz)
baseline and follow-up measurements (Demorest & Walden may allow an alteration of the course of treatment (Fausti et al.
1984) to define a reference limit for DPOAE test–retest change. 1984, 1999; Rizzi & Hirose 2007) before there is an ototoxic
The 90% reference limit for DPOAE test–retest changes was effect on frequencies crucial to speech perception. It may also
defined as ±1.645·√2·SEM (McMillan & Hanson 2014) and allow discussion of the options for aural rehabilitation and tin-
resulted in reference limits ranging from ±3.76 to ±5.63 dB for nitus management, should the need arise, discussion of noise
f2 frequencies spanning 1 to 6 kHz (Reavis et al. 2015). In addi- protection to minimize the possibility of synergistic effects of
tion, it was confirmed that SEM values and associated reference noise and medication, and stress the need to monitor hearing,
limits increased slightly for the higher frequencies analyzed (4 tinnitus, and balance.
and 6 kHz) and increased as the time between tests increased up Current audiologic monitoring guidelines recommend mini-
to 20 days from baseline at all test frequencies. mally testing bilateral pure-tone air-conduction behavioral
Furthermore, repeatability has been demonstrated over suc- thresholds at the standard octave and interoctave frequencies, as
cessive weeks. Franklin et al. (1992) determined that a change well as ultra-high frequencies (>8 kHz), when possible (Ameri-
in DPOAE level greater than 4 dB between 1 and 8 kHz will be can Speech-Language-Hearing Association 1994; American
exhibited in only 5% of the population over a 4-week period, Academy of Audiology 2009). These guidelines acknowledge
while Roede et al. (1993) suggested that a change of greater that pure-tone audiometry has several limitations that may
than 6 to 9 dB in emission level in the 1 to 8 kHz frequency affect its usefulness as a monitoring tool. First, many patients
range constitutes a significant change over a 6-week period. undergoing treatment may not be willing or able to toler-
Overall, a 4 to 9 dB change in DPOAE level would be consid- ate extensive testing, and the reliability may be affected by
ered statistically significant over short and long periods of time the patient’s ability to provide consistent responses (Fausti et
when monitoring DPOAEs at and below 8 kHz. al. 1991). Thus, those patients with limited responses should
Audiologic monitoring focuses on detecting subtle changes be monitored at the highest frequencies, and those who are
within the cochlea; thus, it is important to identify sources of unresponsive should be monitored with objective tests (e.g.,
test–retest variability in DPOAE testing and minimize these evoked OAEs, auditory brainstem response; American Speech-
factors as much as possible. Zhao and Stephens (1999) quan- Language-Hearing Association 1994; American Academy of
titatively assessed various causes of test–retest variance and Audiology 2009). Second, behavioral thresholds are a subjec-
found that altering the position and fit of the probe greatly tive measure capable of detecting changes in hearing sensitivity
contributes to the variance across trials. These researchers sug- (Jacob et al. 2006), while objective measures can provide infor-
gested that refitting the probe affects the interaction between mation regarding specific auditory structures that are known to
acoustic stimuli and ear-canal resonances, as well as the level sustain damage, possibly before changes in hearing sensitivity
of measured background noise. Dreisbach and Siegel (2001) (Ress et al. 1999; Konrad-Martin et al. 2016; Dreisbach et al.
reported that ear-canal standing waves can cause 15 to 20 dB 2017). This is important information to know, so audiologists
greater sound pressure level (SPL) delivered to the eardrum can have the conversation with their patients that auditory dam-
than that measured by the microphone at the entrance of the age has occurred despite lack of changes in hearing sensitivity
ear canal for low (3 to 7 kHz) and high (≥10 kHz) frequencies. and counsel them as to what to watch for in terms of hearing
In addition, the variation in probe insertion and reinsertion damage and when to seek assistance.
DREISBACH ET AL. / EAR & HEARING, VOL. XX, NO. XX, 00–00 3
HF (>8 kHz) DPOAEs have the potential to be an efficient, showed identical audiometric thresholds at 4 kHz. Katbamna et
objective monitoring tool. Previously, the examination of these al. (1999) extended these findings reporting elevated DPOAE
measures was hindered by equipment limitations; however, if thresholds at 4 and 8 kHz for young CF patients (n = 18) treated
these measures are proven to be repeatable and reliable, they with tobramycin compared with healthy controls (n = 15). In
may be considered for clinical use in monitoring protocols. HF addition, Stavroulaki et al. (2002) reported elevated detection
DPOAEs were explored by Dreisbach and Siegel (2001, 2005) thresholds between 4 and 8 kHz (f2) in children with CF (n =
using customized equipment and various DPOAE paradigms 12) exposed to gentamicin compared with children with CF (n
(varying stimulus frequency, ratio, and levels), and the results = 8) not exposed to gentamicin and healthy volunteers (n = 11).
were similar to those elicited with lower-frequency stimuli. They suggested that the elevated detection thresholds may be
They suggested that the same cochlear processes involved in attributed to the effects of aminoglycoside exposure rather than
producing emissions in the conventional range are involved in the characteristics of the disease. DPOAE measures other than
producing HF DPOAEs. Dreisbach et al. (2006) examined the the traditional discrete frequency sweeps could be considered
repeatability of HF DPOAEs in normal-hearing adults using when assessing or monitoring cochlear sensitivity.
discrete frequency and ratio sweeps over four trials employing The assessment of HF DPOAEs may prove to be an effec-
traditional calibration procedures. Average DPOAE level differ- tive monitoring tool given the sensitivity of traditional DPOAE
ences between trials for discrete frequency sweeps at higher and measures to ototoxic effects. Monitoring HF DPOAE detection
lower frequencies were 5.15 dB (SD 4.40 dB) and 2.80 dB (SD thresholds and group delay values may provide even earlier
2.70 dB), respectively. Average group delay differences between indicators of ototoxicity; however, the repeatability of these
trials, calculated from the ratio sweeps, for higher (10 through paradigms must be established in a patient population before
16 kHz) frequencies were 0.22 msec (SD 0.20) and 0.28 msec it can be recommended for clinical use. Thus, the present study
(SD 0.24 msec) for lower (<8 kHz) frequencies. Although HF examined the repeatability of DPOAEs measured with various
DPOAE levels elicited with a discrete frequency sweep were paradigms at frequencies ≥8 kHz in normal-hearing CF patients
more variable compared with the low frequencies, the responses utilizing a depth-compensated calibration approach. CF patients
were not significantly different across trials for either paradigm were chosen as our patient population because they often suffer
tested, demonstrating repeatability over a minimum of 4 weeks. from frequent coughing and shortness of breath, leading to high
The question remains whether HF DPOAEs are repeatable in a physiologic noise levels (Sawicki et al. 2008) but can be tested
patient population those targeted for this type of testing. in between treatments reducing possible ototoxic effects on the
As mentioned previously, the standard method of examining results. It is hypothesized that HF DPOAEs will be repeatable
DPOAEs is to measure the emission level across a frequency in CF patients, demonstrating the use of DPOAE discrete fre-
range (discrete frequency sweep), to determine the status of quency, ratio, and level sweeps as reliable paradigms for use in
OHCs along the tonotopically organized basilar membrane. ototoxicity monitoring programs.
However, others have suggested that additional DPOAE para-
digms, allowing group delay calculations and growth detection MATERIALS AND METHODS
threshold determination, may be more sensitive measures of
ototoxicity. Both Engdahl & Kemp (1996) and Namysłowski et Patients
al. (2004) reported differences in group delay values measured Forty adult patients with CF (17 females and 23 males) rang-
before and after exposure to either brief or long durations of ing in age from 19 to 55 years (average age: 29 years) participated
noise in humans. Telischi et al. (1998) found that after the inter- in this study. These patients were recruited from the Adult Cystic
ruption of cochlear blood flow, changes in group delay values Fibrosis Clinic at the University of California, San Diego Medi-
occurred sooner than DPOAE level changes measured using cal Center. Patients were not receiving intravenous aminoglyco-
a discrete frequency sweep paradigm. Katbamna et al. (1999) side treatments at the time of their participation, although inhaled
reported that changes in DPOAE group delay values in a group tobramycin usage was permitted because it has not been shown
of cystic fibrosis (CF) patients receiving chronic tobramycin to be ototoxic (Tan et al. 2003). Each patient provided informed
treatments despite identical hearing sensitivity and DPOAE consent before participation in the study. The Institutional
levels measured with a discrete frequency sweep paradigm to Review Boards at San Diego State University and University of
their healthy counterparts. Thus, there is evidence that DPOAE California, San Diego, approved this study before data were col-
group delay can be a sensitive measure of cochlear damage. lected. Patients needed at least one clear ear canal, HF behavioral
In addition, Dreisbach and Siegel (2001) and Dreisbach et al. thresholds for at least 2 of the HF test frequencies (8 to 16 kHz),
(2006) reported that mean test–retest group delay values over 2 and present emissions at 2 of the HF DPOAE frequencies tested
trials result in very high correlations (R values of 0.98) and were (8 to 16 kHz) to participate in the study. Upon completion of the
not significantly different over 4 trials in normal-hearing adults, study, patients were compensated for their time.
respectively. Together, these findings suggest that DPOAE
group delay values are repeatable over time in healthy adults Equipment, Calibration, and Software
and may provide an earlier indication of cochlear damage com- An Interacoustics AC40 audiometer with Sennheiser HDA
pared with more traditional DPOAE measures. 200 headphones was used to collect HF behavioral thresholds.
A DPOAE detection threshold, derived from a primary tone An emission probe system with all the capabilities necessary
level sweep, is the lowest stimulus level at which an emission to reliably measure DPOAEs up to 16 kHz was used. The sig-
is considered present (Whitehead et al. 1993). Mulheran and nals for both the f2 and the f1 channels were generated at the
Degg (1997) found that detection thresholds in young, normal- digital-to-analog converter of a MOTU 96 kHz Audio Firewire
hearing CF individuals (n = 14) were higher at 4 kHz (f2) com- Interface, which was connected to a laptop computer, and pre-
pared with healthy controls (n = 36), even though both groups sented through Sennheiser CX 300 ultra-HF transducers. The
transducers were connected to a low-noise DPOAE probe assem- monitoring programs (American Speech-Language-Hearing
bly positioned in the subject’s ear canal. The probe assembly Association 1994; American Academy of Audiology 2009).
also housed the emission probe microphone (ER-10B+), which After HF audiometry, DPOAE testing was completed in the
transduced the sound pressure in the ear canal. The signal mea- ear with the better HF behavioral thresholds, and if both ears
sured by the ER-10B+ microphone was amplified (20 dB) by had equivalent HF pure-tone thresholds, the right ear was cho-
the ER-10B+ preamplifier before being digitized by the analog- sen for testing. In total, 29 right ears and 11 left ears were tested.
to-digital converter of the MOTU Audio Firewire Interface. The The DPOAE probe was placed into the subject’s ear canal, and
stimulus and DPOAE levels were measured at the plane of the the half-wave resonance was measured using SysRes. The value
emission probe microphone near the entrance to the ear canal. of the half-wave resonance was chosen based on the first larg-
Real-time simultaneous response computing software, Sys- est amplitude peak in the ear-canal frequency response, then
Res (Neely & Stevenson 2002), was used to record the half-wave that chosen value was used to determine which stimulus file to
resonance of the patient’s ear canal. The recorded half-wave res- present to the patient. The half-wave resonance of the patient’s
onance of the ear canal was used for calibration of the stimulus ear canal was determined before each trial to ensure the probe
levels and replacement of the probe assembly in the ear at the was placed appropriately compared with the half-wave reso-
start of each trial. Thus, for calibration purposes, the patient’s nance values of the previous trial(s). An acceptable half-wave
half-wave resonance was incorporated into the stimulus file in resonance value was within 0.1 Hz of previous recordings. If
an attempt to better control the stimulus levels presented to the an acceptable half-wave resonance value was not obtained, the
ear, which can be adversely affected by standing waves in the probe was removed and repositioned in the patient’s ear canal.
ear canal (Siegel 1994; Siegel & Hirohata 1994). Before data Patients were screened for the presence of 2f1−f2 DPOAEs
collection, the emission probe sound sources were calibrated using a discrete frequency sweep with f2 swept from 16 to 8 kHz,
in a Brüel & Kjær (B&K) 4157 ear simulator. The probe was resulting in 16 points. The frequency ratio (f2/f1) was 1.2, and
inserted in increments of ~1 mm, and the pressure response the levels of the lower (L1) and higher (L2) frequency primary
was measured with the emission probe and a simulator micro- tones were 65 and 50 dB SPL, respectively. DPOAEs were con-
phone. The frequency of the half-wave resonance was recorded sidered to be present when the signal to noise ratio was 6 dB or
for each insertion depth. After the insertion depth measure- more, and the level of the emission was greater than −20 and
ments, stimulus files were created using the recorded pressure lower than 15 dB SPL. To participate in the study, patients had
responses for the different half-wave resonances. At each of the to have present DPOAEs minimally at 8 and 10 kHz obtained
four trials, the patient’s half-wave resonance was measured, and from the frequency sweep.
the ear simulator microphone pressure response with the closest The highest f2 frequency producing a present DPOAE was
half-wave resonance frequency was selected as the estimate of identified from the discrete frequency sweep for each patient.
the patient’s eardrum pressure (Siegel 2009). Once this frequency was determined, the 2 highest f2 frequen-
EMAV (Neely & Liu 1993) software was used to collect cies out of 5 possibilities (8, 10, 12, 14, or 16 kHz), at which
DPOAE data. The EMAV software samples the ear-canal sound DPOAEs were present were determined. On the basis of this
pressure in 46 ms time windows, accumulates these samples determination, a DPOAE ratio and level sweep were completed
into 2 interleaved buffers, and averages each buffer in the time at each of these two highest frequencies for each patient. For
domain. The two subaverages are then subtracted (A − B) and example, if a patient had present DPOAEs from 13 kHz and
Fast Fourier transformed to provide an estimate of the back- below in the DPOAE discrete frequency sweep, a DPOAE ratio
ground noise floor. The grand average ([A + B]/2) is transformed and level sweep would be completed at 12 and 10 kHz. Only
to estimate the levels and phases of both stimuli and distortion the two highest frequencies were examined with these DPOAE
products present in the ear canal. The level of the noise at the ratio and level sweeps due to time constraints of the patients
distortion-product frequencies is also estimated from the A − B (total testing time of 1 hour).
spectrum. Sampling occurred, for a minimum of 4 seconds or Ratio sweeps were collected to determine DPOAE group
longer, until 1 of 2 stopping rules were met: the noise floor at delay values, which were calculated from the slope of the distor-
the distortion-product frequency was less than −20 dB SPL or tion-product phase versus distortion-product frequency curve,
until 4 seconds of artifact-free sampling had been averaged. at the two highest frequencies where DPOAEs were present
To estimate system distortion, probe microphone measure- in the frequency sweep. Group delay is related to the time it
ments were made in an acoustic cavity (B&K 4157 ear simulator) takes between the presentations of the stimuli and the time the
with impedance characteristics similar to the human ear canal. microphone measures the DPOAE in the ear canal. To calcu-
System distortion (energy recorded in the test cavity at 2f1–f2) was late group delay, DPOAE level and phase were measured with
at least 80 dB below the stimulus levels used in this study. f2 held constant and f1 varied in increments of 43 Hz, resulting
in frequency ratios (f2/f1) of 1.1 to 1.3. This method resulted in
Procedures fewer data points for ratio sweeps at 8 kHz (41) and the greatest
All testing was performed in a quiet examination room where number of data points at 16 kHz (82). Fixing f2 and varying f1
the patient was positioned in a reclining chair. Otoscopy was was chosen to estimate delays associated with the f2 region and
performed to determine that patients had clear ear canals. HF because this same procedure has been used to monitor ototoxic
behavioral thresholds were obtained using the modified Hugh- effects in humans (Katbamna et al. 1999) and examine group
son–Westlake (Carhart & Jerger 1959) procedure at discrete delay repeatability (Dreisbach et al. 2006). Closely spaced dis-
frequencies from 8 to 16 kHz in each ear because it is known tortion-product frequencies, covering a small-frequency range,
that HF behavioral thresholds are repeatable over time (Frank & were used so that when calculating the DPOAE group delay for
Dreisbach 1991) and are used as the gold standard for ototoxic each frequency region, there were enough points per cycle to
prevent 2 Π phase ambiguity. The stimulus levels (L1/L2) were Estimating the Lowest L1 Required to Elicit a DPOAE
60/45 dB SPL. DPOAE thresholds were estimated subjectively and
DPOAE level sweeps were collected with L1 varied from 20 objectively.
to 70 dB SPL in 3 dB increments, resulting in 17 data points, a. Subjective approach: For subjective estimation, input–
whereas L2 was fixed at 50 dB SPL (f2/f1 = 1.22) at the 2 high- output functions were plotted with DPOAE level on the y
est frequencies where DPOAEs were present in the frequency axis and L1 on the x axis. DPOAE threshold was defined
sweep. DPOAE detection thresholds were estimated subjec- as the lowest L1 at which the DPOAE level exceeded the
tively and objectively. These procedures will be discussed in estimated noise floor by 6 dB, reached a value greater than
data analyses. −20 dB SPL, and demonstrated continued growth with
subsequent increases in L1. A single example is shown in
Assessment Schedule the upper panel of Figure 1 where the subjective thresh-
Overall, a total of one behavioral assessment (audiogram) old was determined to be at an L1 of 45 dB SPL. One
and five DPOAE paradigms (discrete frequency sweep, two limitation with the subjective method is the difficulty of
ratio sweeps, and two level sweeps) were run at each trial estimating the level of the noise floor and deciding when
for each patient. Each trial took approximately 30 minutes the threshold criterion has been reached, especially when
to complete. This protocol was repeated four times to assess the noise is close to −20 dB SPL. Another limitation
repeatability of these measurements. The four trials were is the 3 dB precision with which threshold can be esti-
completed in one or two sessions dependent on how often mated (L1 varied in 3 dB increments). A more objective
the patient came to the clinic. If they did not come regularly approach was developed to address these limitations.
to the clinic, typically due to greater distances to travel, they b. Objective approach: The objective approach involved
were tested in one session to avoid the possibility of receiv- fitting a nonlinear function to the data and using the fit-
ing antibiotic treatments before returning to the clinic for ting parameters to compute the L1 at which the DPOAE
audiologic testing. Twenty-three patients completed all four level reaches a threshold criterion. Because the input–
trials in one session. The remaining 17 patients completed 2 output function for many patients clearly followed a cur-
trials at each session, and the sessions were scheduled from vilinear path, a quadratic equation was used to represent
1 to 45 days apart, with an average and median time between the change in DPOAE level with increasing L1:
sessions of 12 and 7 days, respectively. It should be noted
that after each trial (n = 4), the probe was removed, and the y = a + bx + cx 2 (1)

patient was given a minimum break of 5 minutes, or as long
as requested by the patient. In addition, if testing was com- This function, however, needed a lower limit at which y remains
pleted all in 1 day, the patient was given a minimum of a constant—representing the noise level:
10-minute break after trial 2 before starting trial 3. The ses- y = d (2)
sions did not occur across longer intervals of time (days) as
the patients often receive intravenous antibiotic treatments, A single equation was obtained by logarithmic addition:
which would have excluded the patients from participating. ∧
Zhao and Stephens (1999) reported that a large portion of y = log10 (10 d + 10( a + bx + cx
2)
) (3)

the variability in DPOAE measures occurs when the probe
The least-squares fit of this function to the data in the top panel
assembly is removed from the ear. In monitoring programs,
of Figure 1 is shown in the lower panel, along with the fitting
patients are tested multiple times over weeks and months,
parameters. By inserting the values of the four parameters and
not in one session. Thus, to assess variability due to probe
solving for x, the L1 input level providing a DPOAE output of
removal and to better approximate what would occur in mon-
−20 dB SPL (or any other criterion, if desired) is obtained*:
itoring programs, the probe assembly was removed from the
patients’ ear, and they were given a break from testing after
each trial during the session. The patients watched closed-

xth = (−b + (b – 4c (a − log (10
2 th
− 10d ))))
/ ( 2c ) , (4)
caption videos or read quietly to prevent excessive move-
ment due to boredom during testing. where th indicates the selected DPOAE level (in this case −20),
d the level of the noise floor, and a, b, and c are the fitting
parameters of the curvilinear portion of the curve.
Data Analyses In the example of Figure 1, the resulting L1 threshold for a
All data were imported to a spreadsheet for analyses and to DPOAE level of −20 dB SPL is 40.4 dB SPL. The purpose of
calculate DPOAE group delay values and determine DPOAE this more complicated approach was to compare the resulting
detection thresholds. For the DPOAE discrete frequency confidence limits with those obtained by the subjective approach.
sweep data, the recorded DPOAE level had to be 6 dB above
the noise floor and greater than −20 dB SPL to be included in
the average. As a result, not every data point is the average of Repeatability Determination
all 40 subjects. For DPOAE ratio and level sweeps, only the A 2-factor repeated-measures analysis of variance (ANOVA),
two highest frequencies where DPOAEs were present in the using a general linear model, with frequency and trial as vari-
discrete frequency sweep were tested. Thus, for group delay ables, was completed for behavioral threshold and all DPOAE
calculations and DPOAE threshold determination, the num- * The quadratic equation has a second solution. This, however, is the L1
ber of patients included in the average varied and is indicated at which the curvilinear function would go back to the noise floor and is
in figures 5, 6, and 8. ignored here.
frequencies. First, the correlation of the absolute DPOAE levels

for the different trials (trials 1 & 2, trials 1 & 3, trials 2 & 3,
trials 2 & 4, trials 3 & 4, and trials 4 & 1; 6 correlations) was
determined using a Pearson product moment correlation. Next,
the SD of the DPOAE levels between the different trial combi-
nations was determined. The SE of the measurement was cal-
culated as the SD of the DPOAE levels times the square root of
1 -correlation. Confidence intervals or repeatability coefficients
were also determined by multiplying 1.96 by the square root
of 2 times the SE of the measurement. The confidence interval
quantifies the absolute reliability of the measurement error in the
same units as the measurement tool (dB SPL; Vaz et al. 2013).
RESULTS
Behavioral Thresholds
The average behavioral thresholds (in dB SPL) across the
four trials for each tested frequency can be seen in Figure 2.
The triangles, squares, diamonds, and circles correspond to
trials 1, 2, 3 and 4, respectively, and error bars are SDs. As
expected, threshold values worsened as frequency increased.
A total of 40 ears had measurable thresholds at 8, 10, 11.2,
and 12.5 kHz, 38 ears at 14 kHz, and 34 ears at 16 kHz. To
examine the issue of repeatability of behavioral thresholds
across trial, a 2-factor (Frequency × Trial) repeated-measures
ANOVA was completed. The main effect of trial and interac-
tion for frequency by trial were not significant (p > 0.05).
However, a significant main effect for frequency [F(5, 150)
= 9.23; p < 0.01] was found, indicating that thresholds varied
Fig. 1. The upper panel shows the input–output data for a single trial of 1 across frequency, as expected. To define the extent of vari-
patient at 10 kHz. From this plot, a threshold was determined subjectively ability to be expected for behavioral thresholds, differences
by determining the L1 that produced a 6 dB signal to noise ratio, DPOAE between trials (6 calculations) were determined. Average
level > −20 dB SPL and continued growth of the DPOAE level with further
absolute differences between trials for behavioral thresh-
increases in L1. The lower panel shows the same data with a least-squares
fit to Eq. (3). The fitting parameters are shown in the box and were used in
olds were 2.79 dB (SD = 3.48), with the smallest differences
Eq. (4) to obtain the estimate of the L1 producing a DPOAE level of −20 between trials occurring at 12.5 kHz (2.05 dB) and the great-
dB SPL (objective threshold determination). DPOAE indicates distortion- est differences at 16 kHz (3.81 dB). The 95% range of data for
product otoacoustic emission. the absolute differences between trials for all frequencies and
trial combinations was 9.60 dB.
data. For DPOAE group delay and detection threshold measure-
ments, where only the two highest frequencies with present DPOAE Discrete Frequency Sweep Measurement
DPOAEs were tested for each patient, which resulted in uneven Average DPOAE levels compared with the average noise
numbers of patients for each frequency, frequencies were des- floor from 8 through 16 kHz across the 4 trials are shown in Fig-
ignated as frequency 1 or 2 with four entries for each trial. This ure 3 for patients with present emissions. Recall, to participate in
eliminated the need to run separate statistical analyses at each this study, patients had to have at least 2 present DPOAEs from
frequency where not all patients were represented. The signifi- the 16 data points collected. Thus, not every data point includes
cance level for this study was defined as p < 0.05. data from all 40 ears. As frequency increased, the number of
For the behavioral threshold data and the data from each patients with present DPOAEs declined. In Figure 3, the tri-
DPOAE paradigm, an average difference between trials calcula- angles, squares, diamonds, and circles correspond to trials 1, 2,
tion was made. To make this calculation, each absolute trial differ- 3, and 4, respectively, and error bars are SEM. The thin, dashed
ence (trial 1–2, trial 1–3, trial 2–3, trial 2–4, trial 3–4, trial 4–1; 6 line denotes the average noise floor across the four test trials
calculations) was determined for each subject and then averaged. for those who had present DPOAEs. DPOAE levels generally
The average DPOAE absolute trial difference for all frequencies appeared to be lower and more variable at higher frequencies,
plus the SD of the DPOAE level differences for all frequency but differences across trials appeared to be small. Repeatability
times 1.96 was calculated to define the 95% range of the differ- of DPOAE discrete frequency sweeps across trial was exam-
ence between trials data, so expectations for differences between ined using a 2-factor (Frequency × Trial) repeated-measures
trials could be determined for HF DPOAEs and allow for com- ANOVA. The main effect of trial and interaction of frequency
parisons with conventional DPOAE data. by trial were not significant (p > 0.05). However, a significant
In addition, for the DPOAE data, correlations and SE of main effect for frequency [F(15, 570) = 53.49; p < 0.01] was
the measurement were calculated to allow for comparisons found, indicating that DPOAE levels varied across frequency,
with repeatability studies completed at traditional DPOAE which is seen in Figure 3.
Fig. 2. Average behavioral thresholds for the cystic fibrosis patients are shown for the 4 different trials for frequencies between 8 and 16 kHz. The triangles,
squares, diamonds, and circles correspond to trials 1, 2, 3, and 4, respectively, and error bars are SDs.
To examine the extent of variability to be expected for frequencies and trial combinations was 6.26 dB, with the tight-
DPOAE levels obtained via discrete frequency sweeps, abso- est range of differences across trial comparisons occurring at
lute differences between trials (6 calculations) for each of the 8 kHz (5.12 dB) and the widest range of differences across trial
16 frequencies tested were determined for those with present comparisons occurring at 14.5 kHz (9.28 dB).
DPOAEs. The average absolute difference between trials for Table 1 presents the average and SD for the absolute level
DPOAE levels collapsed across frequency and trial compari- difference between trials for each frequency and all trial com-
sons was 1.96 dB (SD = 2.19), with the smallest differences parisons. The trial comparison yielding the greatest level differ-
occurring at 8 kHz (1.52 dB) and the greatest differences at ence for each frequency is bolded. Fifty percent of the time, the
15 kHz (2.80 dB). The absolute difference between trials across difference between trials 1 and 4 yielded the largest level dif-
frequency for all patients ranged from 0.0 to 21.0 dB. The 95% ferences for the frequencies tested. SE of measurement calcula-
range of data for the absolute differences between trials for all tions and correlations for each frequency and trial comparison
Fig. 3. Average distortion-product otoacoustic emission levels for f2 frequencies from 8 to 16 kHz are shown for the 4 different trials. The triangles, squares, dia-
monds, and circles correspond to trials 1, 2, 3, and 4, respectively, and error bars are SEs. The thin line denotes the average noise floor across the four test trials.
TABLE 1. Average and SD for the absolute level difference between trials for each frequency and all trial comparisons for the patients
with present emissions in the discrete DPOAE frequency sweep measurement
Average Level Difference Between Trials Average SD of Level Difference Between Trials
Trial
Comparison T1–T2 T1–T3 T1–T4 T2–T3 T2–T4 T3–T4 T1–T2 T1–T3 T1–T4 T2–T3 T2–T4 T3–T4
F2 (kHz)
16 1.80 2.90 2.87 3.01 2.13 1.94 1.32 2.16 1.81 2.38 1.62 2.03
15 1.61 4.20 2.20 2.08 2.04 1.27 0.77 4.91 1.84 1.85 1.31 0.94
14.6 3.33 2.76 2.80 3.23 2.64 1.99 4.33 1.99 2.44 5.38 2.04 1.56
14 2.09 1.88 3.50 1.08 2.40 2.65 1.42 1.40 2.69 1.30 2.33 2.49
13 1.72 2.16 2.98 1.89 2.42 1.52 1.30 2.30 2.21 2.08 1.79 1.36
12.7 1.49 2.47 2.28 2.54 1.78 2.24 1.64 2.49 2.19 3.49 1.75 3.33
12 2.28 2.88 3.40 1.72 2.02 2.18 1.75 2.84 2.79 1.69 2.03 1.78
11.6 1.97 2.73 2.79 2.16 2.47 1.94 1.96 3.17 2.46 3.22 2.52 1.98
11 1.53 2.71 2.52 1.90 2.07 1.48 1.38 2.36 2.13 2.40 2.22 1.37
10.6 1.24 1.92 2.09 1.50 2.10 1.62 1.09 1.87 2.27 1.89 2.02 1.35
10 1.00 1.86 2.22 1.50 2.12 1.52 0.86 2.06 2.17 1.93 1.85 1.88
9.6 1.28 1.86 2.33 1.45 1.94 1.61 0.99 2.17 2.66 1.86 2.58 1.81
9 1.41 2.23 2.26 1.86 2.07 1.74 1.08 1.97 2.86 1.84 2.92 2.12
8.8 1.20 1.67 1.97 1.49 1.78 1.60 0.86 1.70 2.53 1.51 2.75 2.02
8.4 1.40 1.54 1.79 1.81 2.12 1.65 1.24 1.87 2.42 1.61 2.73 2.53
8 1.17 1.49 1.69 1.51 1.71 1.52 1.07 1.53 1.95 1.62 2.57 2.02
The trial comparison yielding the greatest level difference for each frequency is bolded.
are presented in Table 2. The SE of measurement ranged from between these trials. Examination of the data indicates that
0.94 to 3.94, correlations ranged from 0.76 to 0.98, and confi- extreme DPOAE level differences between trials occur for only
dence intervals ranged from 2.60 to 10.92. a few patients. Using the 95% range of data for absolute differ-
Figure 4 represents DPOAE level differences between trials ences between trials for all frequencies and trial combinations
1 and 4 as a function of both the number (histograms referred (6.26 dB) as a guide, inspection of individual data revealed 3
to the count axis) and cumulative percentage of patients (con- patients (sessions occurring 0, 2, or 28 days apart) with DPOAE
nected points referred to the right vertical axis). The 4 panels in level differences between trials 1 and 4 at 2 consecutive fre-
each figure correspond to different f2 frequencies (8, 10, 12, and quencies and 2 patients (both sessions occurring on the same
14 kHz). The sign of the differences was ignored, and the total day) with level differences at 5 or 8 consecutive frequencies.
number of patients with present DPOAEs for a particular fre- These last 2 patients with greater than 6.26 dB DPOAE level
quency is indicated in each panel. Trials 1 and 4 were chosen, as differences between trials 1 and 4 were due to trial 4 DPOAE
the longest amount of time (minimally 105 minutes) had passed levels varying compared with the other trials, as other trial
TABLE 2. SE of measurement calculations and correlations for each frequency and trial comparison for the discrete DPOAE frequency
sweep measurement for all patients with present emissions
SE of Measurement Correlation
Trial
Comparison T1–T2 T1–T3 T1–T4 T2–T3 T2–T4 T3–T4 T1–T2 T1–T3 T1–T4 T2–T3 T2–T4 T3–T4
F2 (kHz)
16 1.57 2.13 2.21 2.00 1.77 1.62 0.95 0.91 0.89 0.92 0.93 0.95
15 1.40 2.10 1.72 1.99 1.76 0.94 0.96 0.90 0.93 0.91 0.93 0.98
14.6 3.94 2.54 2.85 3.44 2.52 1.90 0.76 0.89 0.86 0.81 0.90 0.94
14 1.90 1.74 3.29 1.29 2.41 2.91 0.95 0.96 0.86 0.98 0.92 0.89
13 1.68 2.38 2.93 1.96 2.09 1.52 0.97 0.93 0.91 0.96 0.96 0.98
12.7 1.75 2.89 2.76 2.27 1.80 1.67 0.95 0.89 0.90 0.93 0.95 0.97
12 2.06 3.05 3.37 1.76 2.12 1.72 0.94 0.87 0.84 0.96 0.94 0.96
11.6 2.09 3.40 2.93 2.52 2.57 1.66 0.92 0.80 0.86 0.89 0.89 0.95
11 1.57 2.65 2.71 2.35 2.46 1.45 0.96 0.87 0.89 0.90 0.91 0.96
10.6 1.28 2.01 2.24 1.82 2.21 1.63 0.97 0.93 0.91 0.95 0.93 0.96
10 1.02 1.81 2.27 1.52 2.13 1.21 0.98 0.94 0.91 0.95 0.90 0.97
9.6 1.19 2.00 2.51 1.57 2.46 1.73 0.97 0.92 0.88 0.94 0.88 0.94
9 1.26 2.16 2.69 1.84 2.58 1.93 0.97 0.88 0.82 0.92 0.84 0.89
8.8 1.08 1.65 2.21 1.53 2.46 1.77 0.97 0.93 0.88 0.93 0.84 0.93
8.4 1.33 1.81 2.24 1.90 2.74 2.04 0.95 0.91 0.87 0.90 0.81 0.88
8 1.25 1.54 1.76 1.57 2.40 1.80 0.96 0.93 0.91 0.94 0.85 0.92
Fig. 4. Distortion-product otoacoustic emission level differences between trials 1 and 4 for both the number (represented by histograms referred to the count
axis) and cumulative percentage of patients (represented by connected points referred to the right vertical axis) with present emissions for various f2 frequencies
are shown. The total number of patients with present emissions for a particular frequency is indicated in each panel.
comparisons resulted in differences that were within the 6.26 between trials across frequency for all patients ranged from 0.00
dB range. to 2.25 msec. The 95% range of data for the absolute differ-
ences between trials for all frequencies and trial combinations
DPOAE Group Delay Measurements was 0.87 msec, with the tightest range of differences across trial
The average group delay values across the four trials for comparisons occurring at 14 kHz (0.70 msec) and the widest
each tested frequency can be seen in Figure 5. The triangles, range of differences across trial comparisons occurring at 8 kHz
squares, diamonds, and circles correspond to trials 1, 2, 3 and (1.21 msec). In addition, the SE of measurement ranged from
4, respectively, and error bars are SEM. The number of patients 0.08 to 0.56, correlations ranged from 0.37 to 0.95, and confi-
tested at each frequency is indicated in Figure 5. Recall, dence intervals ranged from 0.22 to 1.55.
only the two highest frequencies where patients had present Figure 6 represents DPOAE group delay value differences
DPOAEs determined from the discrete frequency sweep were between trial 1 and trial 4 as a function of both the number (his-
tested. Group delay values generally decreased as frequency tograms referred to the count axis) and cumulative percentage
increased, as expected. The greatest variability in group delay of patients (connected points referred to the right vertical axis).
values occurred at 8 and 16 kHz, the frequencies with the few- The four panels in each figure correspond to different f2 fre-
est patients. A 2-factor (Frequency × Trial) repeated-measures quencies (8, 10, 12, and 14 kHz). The sign of the differences
ANOVA was completed, where each patient’s lower and higher was ignored, and the number of patients examined at each fre-
frequency group delay value was referred to as frequency 1 quency is indicated. Trials 1 and 4 were chosen, as the longest
or 2, respectively. This analysis revealed that the main effect amount of time had passed between these trials. Examination
of trial and the interaction of frequency by trial were not sig- of these data indicate that extreme DPOAE group delay value
nificant (p > 0.05). A significant main effect of frequency differences between trials occur for only a few patients. Using
[F(1, 40) = 9.73; p < 0.01] was found, indicating that group the 95% range of data for absolute differences between trials for
delay values varied at different frequencies. all frequencies and trial combinations as a guide, inspection of
To define the expected variability for DPOAE group delay individual data revealed 4 patients (sessions 0 days apart for 3
calculations, absolute differences between trials (6 calculations) patients and 6 days apart for other patients) with DPOAE group
for each of the 5 frequencies tested were determined for those delay differences greater than 0.87 msec between trials 1 and
with present DPOAEs. The average absolute difference between 4. However, these patients who had an extreme DPOAE group
trials for group delay values collapsed across frequency and delay value differences between trials for one frequency did not
trial comparisons was 0.23 msec (SD = 0.33), with the small- have an extreme difference value for the other frequency tested,
est differences occurring at 14 kHz (0.19 msec) and the great- nor were they the same patients who had extreme differences
est differences at 16 kHz (0.29 msec). The absolute differences for the discrete frequency sweep data.
Fig. 5. Group averages for distortion-product otoacoustic emission group delay are shown for the four different trials at the frequencies tested. The triangles,
squares, diamonds, and circles correspond to trials 1, 2, 3, and 4, respectively, and error bars are SEs. The total number of patients included for a particular
frequency is indicated below the frequency label on the x axis.
DPOAE Level Sweep Measurements noise floor by 6 dB, was greater than −20 dB SPL, and was fol-
Recall, DPOAE detection thresholds were determined sub- lowed by subsequent growth. An example of a DPOAE detec-
jectively and objectively. Subjective determination of DPOAE tion threshold determined subjectively can be seen in Figure 1,
detection thresholds were estimated at the point on the resul- where threshold is indicated for this patient by the vertical arrow
tant growth curve when the emission level first exceeded the (45 dB SPL). Objective DPOAE detection thresholds were
Fig. 6. The same as Figure 4 except that group delay differences between trials 1 and 4 are shown for various f2 frequencies.
determined using Eq. 3. Table 3 displays the data for the patient those with present DPOAEs. The average absolute difference
shown in Figure 1 obtained during four trials. Also shown are between trials for subjective detection threshold values collapsed
the values of a, b, c, and d obtained from least-squares fits of the across frequency and trial comparisons was 3.33 dB (SD = 3.07),
data to Eq. 3. The resulting fitted curves were shown in Figure with the smallest differences occurring at 10 kHz (2.84 dB) and
1. A detection threshold criterion of −20 dB SPL was used for the greatest differences at 16 kHz (3.74 dB). While the absolute
these fits. The detection threshold determined from these fits is difference between trials for objectively determined thresholds
shown at the bottom of Table 3 (Threshold at −20 dB SPL), and was 3.47 dB (SD = 2.62), with the smallest differences occurring
for this individual, the thresholds are very similar across trials. at 16 kHz (2.77 dB) and the greatest differences at 12 kHz (3.66
The average subjective and objective DPOAE detection thresh- dB). The absolute difference between trials across frequency for
old values for each trial and frequency are plotted in Figures 7A, B, all patients ranged from 0.00 to 16 dB and 0.04 to 15.6 dB for the
respectively. Overall, DPOAE thresholds appeared similar across subjective and objective determination of threshold, respectively.
trials, with slightly higher threshold values at higher frequencies. A The 95% range of data for the absolute differences between
2-factor (Frequency × Trial) repeated-measures ANOVA was com- trials for all frequencies and trial combinations was 9.34 and 8.6
pleted, where each patient’s lower and higher frequency threshold dB for subjectively and objectively determined thresholds, respec-
was referred to as frequency 1 and 2, respectively. For both the tively. The tightest range of differences across trial comparisons
subjective and objective determination of threshold, the main effect for subjectively and objectively determined thresholds occurred at
of trial was not significant (p > 0.05). However, a significant main 12 kHz (7.78 dB) and 16 kHz (6.44 dB), correspondingly. While
effect of frequency for both the subjective and objective determi- the widest range of differences across trial comparisons occurred
nation of threshold was revealed, [F(1, 40) = 26.59; p < 0.01] and at 14 kHz (10.65 dB) and 10 kHz (9.73 dB) for subjectively and
[F(1, 39) = 18.65; p < 0.01], respectively, indicating that DPOAE objectively determined thresholds, respectively. In addition, for
detection threshold varied across frequency. The interaction of subjective and objective determination of thresholds, the SE of
frequency by trial was not significant (p > 0.05) for the objective measurement ranged from 1.66 to 4.23 or 1.37 to 4.01, correla-
determination but was significant for the subjective determination tions ranged from 0.57 to 0.96 or 0.63 to 0.97, and confidence
of threshold, [F(3, 120) = 3.65; p < 0.02]. A Bonferroni post hoc intervals ranged from 4.6 to 11.7 or 3.8 to 11.1, correspondingly.
analysis revealed no significant differences between subjectively Figure 8 represents subjective DPOAE detection threshold
determined detection thresholds for any trial for either frequency. value differences between trial 1 and trial 4 as a function of both
To examine the expected variability for DPOAE detection the number (histograms referred to the count axis) and cumula-
threshold values, absolute differences between trials (6 calcula- tive percentage of patients (connected points referred to the right
tions) for each of the 5 frequencies tested were determined for vertical axis). The four panels in each figure correspond to differ-
ent f2 frequencies (8, 10, 12, and 14 kHz). The sign of the differ-
TABLE 3. Level sweep data for the patient shown in Figure 1 ences was ignored, and the number of patients examined at each
obtained during four trials are shown in the top portion frequency is indicated. Trials 1 and 4 were chosen, as the longest
amount of time had passed between these trials. As reported for
Trial
the other measures, extreme DPOAE detection threshold value
dB (SPL) 1 2 3 4 differences between trials occur for only a few patients. Using
20 −27.1 −34.97 −29.01 −26.46 the 95% range of data for absolute differences between trials for
23 −19.03 −23.13 −24.86 −32.37 all frequencies and trial combinations as a guide, inspection of
26 −33.31 −35.4 −18.97 −28.62 individual data revealed 6 patients (sessions 0 days apart for 4
29 −29 −30.52 −27.56 −26.73 patients; 13 and 28 days apart for the other patients) with DPOAE
32 −21.07 −33.03 −26.79 −23.52 threshold differences greater than 9.34 dB between trials 1 and 4.
35 −24.23 −41.62 −33.47 −34.02 Yet, these patients who had an extreme DPOAE threshold value
38 −20.72 −20.74 −31.09 −19.75 differences between trials for one frequency did not have an
42 −19.48 −23.58 −18.04 −22.83 extreme difference value for the other frequency tested. In addi-
45 −16.61 −15.62 −19.42 −20.54 tion, none of the patients exhibiting an extreme DPOAE detec-
48 −10.76 −14.44 −16.55 −16.69 tion threshold value presented with an extreme DPOAE group
51 −9.42 −9.44 −11.38 −10.07
delay value at either of the frequencies tested. One patient with
54 −6.13 −8.41 −4.04 −6.2
an extreme DPOAE detection threshold value also displayed an
57 −3.46 −4.52 −2.78 −2.96
61 −3.01 −2.82 −1.7 −3.39
extreme DPOAE discrete frequency sweep value; however, these
64 −3.96 −4.21 −2.24 −3.07 extreme values occurred for different frequencies.
67 −6.02 −5.95 −5.15 −7.11 Objective DPOAE detection threshold value differences
70 −12.03 −10.75 −8.19 −6.9 between trial 1 and trial 4 were examined, and extreme DPOAE
Fitting parameters: −20 dB threshold detection threshold value differences between trials occur for only
a −150.6 −158.8 −190 −123.9 a few patients. Using the 95% range of data for absolute differences
b 4.88 5.04 6.05 3.66 between trials for all frequencies and trial combinations as a guide,
c −0.0408 −0.0412 −0.049 −0.0281 inspection of individual data revealed 3 patients (sessions 0, 7, and
d −25.6 −32.8 −27.4 −28.6 28 days apart) with DPOAE threshold differences greater than 8.60
Threshold at −20 40.4 41.9 43.2 41.8 dB between trials 1 and 4. These patients with extreme DPOAE
dB SPL threshold value differences between trials for one frequency did not
have an extreme difference value for the other frequency tested. In
The values of a, b, c, and d obtained from the least-squares fits of the data to Eq. (3) for this
patient are shown in the bottom portion, followed by the objectively determined threshold addition, none of the patients exhibiting an extreme DPOAE detec-
at −20 dB SPL for each trial. tion threshold value presented with an extreme DPOAE group
A B
Fig. 7. Average L1 threshold (dB SPL) values determined either subjectively (A) or objectively (B) are shown for the four different trials for the different frequen-
cies tested. The various trials are depicted with either solid or hashed bars, and error bars are SEs.
delay value at either of the frequencies tested. One patient with test sessions in young, healthy adults with normal low-frequency
an extreme DPOAE detection threshold value also displayed an hearing, the HF behavioral threshold data for the CF patients
extreme DPOAE discrete frequency sweep value; however, these were repeatable over time. The average difference between tri-
extreme values occurred for different frequencies. als for behavioral thresholds was 2.79 dB (SD = 3.48) for the
CF patients compared with ±1.1 dB (SD = 3.0 to 4.4 dB) for the
young, normal-hearing adults. The 95% range of data for the
DISCUSSION
absolute differences between trials for all frequencies and trial
Repeatability of HF Behavioral Thresholds combinations was 9.60 dB for the CF patients and within ±10
In agreement with Frank and Dreisbach (1991) who exam- dB for 94% of the young adults. These data indicate that the CF
ined the repeatability of HF behavioral thresholds across four patients who were not receiving ototoxic treatments during the
Fig. 8. Same as Figures 4 and 6 except that subjective distortion-product otoacoustic emission threshold differences between trials 1 and 4 are shown for
various f2 frequencies.
testing period had repeatable HF hearing thresholds, as has been were noted for the various level combinations (Dreisbach et
reported for nonpatient populations. In addition, these results al. 2006). More CF patients participated than healthy volun-
imply no significant changes in their hearing status during this teers but not all CF patients had recordable data at each fre-
study (American Speech-Language-Hearing Association 1994; quency tested. The amount of time between trials was longer
American Academy of Audiology 2009). for the healthy volunteers, tested 4 times separated by 1 week
and no more than 2 weeks apart (tested over 4 to 7 weeks in
Repeatability of Discrete HF DPOAE Levels total), whereas 57% of our CF patients were tested in 1 day
As has been reported previously in normal-hearing, healthy (with probe removal after each trial), and the remaining 43%
adults for DPOAE levels less than 8 kHz (Franklin et al. 1992; of patients were tested on 2 different days, separated by 1 to 45
Rhode & Cooper 1993; Beattie et al. 2003; Ng & McPerson days (median value of 7 days apart). It is known that DPOAE
2005; Reavis et al. 2015), discrete HF DPOAE levels were variability increases as time between testing sessions increases
repeatable over time in CF patients. Differences between tri- (Reavis et al. 2015), but given the patient population, we had
als less than 6.26 dB can be expected in 95% of CF patients access to study who routinely receive ototoxic therapies and
for frequencies ≥8 kHz compared with differences between tri- have to travel great distances for their health care, the timing of
als of 5.3 or 6 to 8 dB for frequencies 1 to 6 or 6.5 to 7 kHz, the testing was dictated by the patients schedules.
respectively, in healthy adults (Ng & McPerson 2005). Average One major difference between the present study and the
differences across trial comparisons for CF patients increased healthy participants (Dreisbach et al. 2006) was the calibration
as frequency increased with differences of approximately 5 procedure. Traditional calibration techniques, which are sub-
dB expected at 8 kHz versus 9 dB at 14.5 kHz. This trend of ject to the deleterious effects of standing waves from 3 to 7 and
increasing differences between trials as frequency is increased ≥10 kHz (Dreisbach & Siegel 2001), were used by Dreisbach et
has been reported previously (Ng & McPerson 2005; Dreisbach al. (2006), whereas depth-compensated simulator SPL calibra-
et al. 2006; Wagner et al. 2008; Reavis et al. 2015). The SE tion procedures were used in the present study. In a pilot study
of measurement ranged from 0.94 to 3.94 in the CF patients (Major & Dreisbach, Reference Note 1), 10 normal-hearing,
versus 1.31 to 3.45 for 1 to 6 kHz and 2.59 to 3.04 for 6.5 to young adults with an age range of 22 to 26 years (mean age:
7 kHz in healthy adults (Ng & McPerson 2005). Overall, the 24 years) underwent the exact same testing procedures used in
variability of HF DPOAE levels across trials in CF patients is the present study with CF patients, including the use of depth-
slightly greater than what has been reported at lower frequen- compensated calibration techniques and the timing of the test
cies (≤6 kHz) but comparable to results reported for frequencies sessions. The pilot data revealed average absolute differences
greater than 6 and less than 8 kHz for healthy adults. between trials, collapsed across all frequencies and trial com-
The DPOAE variability across trials in CF patients at fre- parisons, of 1.82 (SD = 1.87 dB), and differences between trials
quencies ≥8 kHz were more aligned with data reported at con- less than 5.48 dB can be expected in 95% of normal-hearing
ventional frequencies, <8 kHz, than those previously reported young adults. The values obtained from the pilot study are
for young, normal-hearing adult subjects at higher frequen- more similar to those obtained from CF patients in the pres-
cies, >8 kHz (Dreisbach et al. 2006). The average DPOAE level ent study versus values obtained from normal-hearing young
difference between trials was 1.96 dB (SD = 2.19 dB) for CF adults in Dreisbach et al. (2006), as seen in Table 4. Thus, these
patients and 5.15 dB (SD = 4.40 dB) for young adults at fre- results suggest that improved calibration techniques can lead to
quencies ≥8 kHz. This finding was unexpected, and a few expla- reduced variability between serial measurements, as posited by
nations were explored. It is possible, although unlikely, that CF Souza et al. (2014), when using HF stimuli.
patients present with less variable results than young, normal-
hearing, healthy volunteers. Generally speaking, the overall Repeatability of HF DPOAE Group Delay Values
DPOAE levels were similar between the groups with the high- As has been reported for normal-hearing, young adults
est levels occurring at 8 kHz and a gradual decline in emission (Dreisbach et al. 2006), HF DPOAE group delay values were
level as frequency increased. The noise floors also were similar repeatable over four trials in CF patients. The greatest vari-
between the CF patients and the young, normal-hearing adult ability between trials was found at the lowest and highest fre-
group. Although only 1 level combination was used in the pres- quencies examined (8 and 16 kHz), those with the fewest ears
ent study (L1/L2 = 65/50 dB SPL), 4 level combinations were tested. The average absolute difference between trials for group
used with the healthy volunteers, but no significant differences delay values collapsed across frequency and trial comparisons
TABLE 4. Average and SD absolute difference between trials collapsed across frequency and trial comparison and 95% range of the
data for DPOAE discrete frequency sweeps for 25 NH young adults, CF patients in the present study, and pilot data for 10 NH young
adults
NH Young Adults (n = 25; CF Patients (n = 40; NH Young Adults (n = 10; Pilot Data by
DPOAE Discrete Frequency Sweep Dreisbach et al. 2006) Present Study) Major and Dreisbach (Reference Note 1)
Calibration method Traditional Depth compensated Depth compensated
Average difference (dB) 5.15 1.96 1.82
SD (dB) 4.40 2.19 1.87
95% range of data (dB) 13.78 6.26 5.48
Traditional calibration methods were used with the 25 normal-hearing young adults compared with a depth-compensated simulator SPL method of calibration for the CF patients and the pilot
data from 10 normal-hearing young adults.
was 0.23 msec (SD = 0.33) for the CF patients compared with same day with probe removal after each trial. This reinforces
0.22 msec (SD = 0.20) and 0.28 msec (SD = 0.24) for high that probe removal and reinsertion contribute to test–retest vari-
(10 to 16 kHz) and low (2 to 8 kHz) frequencies, respectively, ability, as reported by Zhao and Stephens (1999). Dobie (1983)
in young, healthy adults (Dreisbach et al. 2006). Differences in reported that significant differences at consecutive test frequen-
group delay values less than 0.7 msec are expected in 96.5% of cies would indicate a more systematic change and increase the
healthy adults across trials at high frequencies compared with likelihood of a true decrease in sensitivity, and Dreisbach et al.
differences less than 0.87 msec for 95% of CF patients. If only (2006) suggested that a significant change for DPOAE levels
the frequencies (10, 12, and 14 kHz) with the number of ears and group delay values should occur at two consecutive fre-
similar to Dreisbach et al. (n = 25) are included in the calcula- quencies to be alarming. If an individual demonstrated a sig-
tion of the 95% range of data, then differences in group delay nificant change between trials 1 and 4 for either a group delay
values less than 0.75 msec between trials would be expected in (n = 4) or DPOAE threshold (subjective, n = 6, or objective,
CF patients. Overall, group delay values in CF patients are not n = 3), only one frequency was affected, and the other para-
significantly different across trials, despite being slightly more digm did not result in significant differences at either frequency
variable than those reported for healthy adults. Also, improved tested. For the discrete frequency sweep paradigm, a total of 16
calibration techniques do not appear to decrease the variability individuals had a significant difference between trials 1 and 4.
of group delay values as it did for DPOAE levels obtained from Five of the 16 individuals had differences at 2, or more, con-
discrete frequency sweeps. secutive frequencies. Two of those individuals with significant
differences at consecutive frequencies in the frequency sweep
Repeatability of HF DPOAE Thresholds—Subjective paradigm did not have significant differences in either of the
and Objective Determination other two paradigms. One individual with significant differ-
DPOAE detection thresholds were slightly greater when ences for two consecutive frequencies in the frequency sweep
determined using a subjective compared with objective approach. paradigm also had a significant difference in a DPOAE thresh-
This is most likely a result of methodology, where objective old but at a different frequency region than that in the frequency
thresholds were determined at the point where the fit intersected sweep paradigm. Finally, the last two individuals had significant
−20 dB SPL and subjective determination occurred at the first differences for more than two consecutive frequencies but did
value of L1 greater than −20 dB SPL meeting the 6 dB signal to not have significant differences in the other paradigms. Over-
noise criterion. Also, with the subjective determination, we were all, there were 16 individuals with significant differences in
constrained by the 3 dB step size used for data collection, which the frequency sweep paradigm, versus 6 individuals with a sig-
is not an obstacle with the objective determination of DPOAE nificant threshold difference at 1 frequency when determined
threshold. Thresholds were slightly higher at 16 kHz compared subjectively and 3 individuals when determined objectively, and
with the other frequencies tested, no matter which method was 4 individuals had significant group delay value differences at
used to determine threshold. This frequency had the fewest 1 frequency. Thus, the least amount of significant differences
patients with measurable responses. The average absolute dif- between trials 1 and 4 was found for group delay values and the
ference between trials for subjective detection threshold values objective determination of threshold. Out of those with extreme
collapsed across frequency and trial comparisons was 3.33 dB values, only one patient had differences occurring for more
(SD = 3.07), compared with differences between trials of 3.47 than one type of DPOAE paradigm, but the frequencies affected
dB (SD = 2.62) for objectively determined detection thresholds. were not the same for the two paradigms. In summary, for a
Similar results were reported for subjectively determined thresh- significant change between trials to be concerning in a DPOAE
old repeatability in young, healthy volunteers in the pilot study frequency sweep, a change should be observed at consecutive
by Major and Dreisbach (Reference Note 1), where the aver- frequencies, and the test should be repeated to ensure the results
age differences between trials was 3.00 dB (SD = 3.46). Differ- were not erroneous. In addition, alternate paradigms, DPOAE
ences between trials of less than 9.34 and 8.6 dB for subjectively ratio or level sweeps, which appear more resilient to significant
and objectively determined thresholds, respectively, would be changes across trials in CF patients, should be considered when
expected in 95% of CF patients. Overall, no matter which meth- monitoring patients over time for ototoxic insults.
odology was used to determine DPOAE threshold, the results
for the CF patients across four trials were repeatable. In future Effects of Calibration on DPOAE Variability
studies monitoring HF DPOAE thresholds in those undergoing When using traditional calibration techniques, test–retest
ototoxic treatments, both methodologies for determining thresh- variability appears increased for DPOAE levels obtained using
old should be employed to determine whether one is more sensi- a frequency sweep paradigm compared with the variability
tive than the other to changes, given that the subjective method observed when using an improved calibration technique (depth-
has some limitations, as outlined in the Materials and Methods compensated simulator SPL). It has been postulated that hav-
section, and can be influenced by tester bias. ing better control over the stimulus levels presented to the ear
could decrease DPOAE level variability across trials (Souza et
Individual Patient Results al. 2014). The results from the present study lend support to this
The data for the difference between trials 1 and 4 were exam- notion at least for the DPOAE frequency sweep paradigm. The
ined to determine how many individuals exhibited a significant improved calibration technique used in this study did not appear
difference based on the 95% expected range of the data for each to decrease the variability between trials for the ratio and level
paradigm tested. A total of 25 individuals had a significant dif- sweep paradigms. There are other improved calibration tech-
ference between trials 1 and 4 for at least one DPOAE para- niques being utilized, forward pressure level, which should be
digm tested. Sixty percent of these patients were tested on the examined to determine whether DPOAE variability for various
paradigms could be reduced compared with traditional calibra- HF audiometry and DPOAE measures at frequencies ≤8 kHz
tion techniques. Reducing the variability of DPOAE measures are sensitive to ototoxic damage (Ress et al. 1999). Given that
across trials is beneficial so that a determination of a significant the damage initially starts at the basal end of the cochlea and
physiologic change can be made with more confidence. HF DPOAE levels, group delay values, and thresholds are
repeatable in young, normal-hearing adults and CF patients,
Sensitivity of Various DPOAE Paradigms to Ototoxic these measures should be employed in a monitoring program
to determine which paradigm or combination of paradigms are
Insults
most sensitive to ototoxic insults. Those with the best hearing
Recall, some have suggested that DPOAE group delay mea-
sensitivity can suffer the most deleterious effects of ototoxicity
sures could be a more sensitive indicator of cochlear damage
(Reavis et al. 2011). Thus, the repeatability of HF DPOAE mea-
than DPOAE discrete frequency sweep measures (Telischi et sures also should be determined in children, a population that
al. 1998; Katbamna et al. 1999; Namysłowski et al. 2004). The should not have presbycusic hearing loss and could benefit from
effect on group delay values, shorter or longer compared with an objective method of monitoring cochlear status at the highest
baseline or healthy controls, was length of exposure and dose frequencies of human hearing.
dependent. Katbamna et al. (1999) posed that factors influenc-
ing basilar membrane stiffness could alter group delay measure-
ments. Comparing group delay results from the present study ACKNOWLEDGMENTS
and that of Dreisbach et al. (2006), which used almost identical The authors thank Garnett McMillan for his statistical guidance and
stimulus parameters except for calibration, it generally appears expertise.
as though the group delay values are slightly decreased in the The authors have no conflicts of interest to disclose.
CF patients compared with the healthy, young adults. Katbamna
et al. (1999) reported that CF patients receiving higher doses of Address for correspondence: Laura Dreisbach, School of Speech, Language,
and Hearing Sciences, San Diego State University, 5500 Campanile Drive,
tobramycin (older patients: 15 to 23 years) exhibited reduced San Diego, CA 92182, USA. E-mail: ldreisba@mail.sdsu.edu
DPOAE group delay values compared with healthy controls. It
Received December 16, 2016; accepted May 8, 2017.
is possible that the slight reduction in group delay values noted
for our CF patients (19 to 55 years) collected between ototoxic The San Diego Foundation Blasker-Rose-Miah Fund supported this work.
treatments compared with healthy, young adults is a result of
longer term aminoglycoside use. However, patients were not
matched in terms of length of exposure to ototoxic drugs, and REFERENCES
the different calibration methods used in the two studies could Abdala, C., Dhar, S., Kalluri, R. (2011). Level dependence of distortion
be influencing these differences as well. While the improved product otoacoustic emission phase is attributed to component mixing.
J Acoust Soc Am, 129, 3123–3133.
calibration technique used in the present study did not necessar-
American Academy of Audiology. (2009). American Academy of Audiol-
ily improve the variability between group delay values over time ogy’s Position Statement and Practice Guidelines on Ototoxicity Moni-
compared with traditional calibration methods, DPOAE phase toring. American Academy of Audiology, 1–25.
has been shown to be level dependent (Abdala et al. 2011). If American Speech-Language-Hearing Association. (1994). Guidelines for
improved calibration techniques do provide better control over the audiologic management of individuals receiving cochleotoxic drug
therapy. ASHA, 36 (Suppl 12), 11–19.
stimulus levels presented to the ear for regions affected by Beattie, R. C., & Bleech, J. (2000). Effects of sample size on the reliability
standing waves (3 to 7 and ≥10 kHz), then inferences about dif- of noise floor and DPOAE. Br J Audiol, 34, 305–309.
ferences in group delay values from healthy adults (traditional Beattie, R. C., Kenworthy, O. T., Luna, C. A. (2003). Immediate and short-
calibration) and CF patients (improved calibration) should be term reliability of distortion-product otoacoustic emissions. Int J Audiol,
42, 348–354.
made with caution. Nevertheless, when monitoring cochlear Brown, A. M., McDowell, B., Forge, A. (1989). Acoustic distortion prod-
status for damage, group delay measures should be considered ucts can be used to monitor the effects of chronic gentamicin treatment.
given that group delay values are repeatable over time and have Hear Res, 42, 143–156.
been shown to vary compared with healthy ears. Carhart, R., & Jerger, J. F. (1959). Preferred method for clinical determina-
tion of pure-tone thresholds. J Speech Hear Disord, 24, 330–345.
Several studies have reported that DPOAE thresholds at fre- Demorest, M. E., & Walden, B. E. (1984). Psychometric principles in the
quencies between 4 and 8 kHz in CF patients exposed to ami- selection, interpretation, and evaluation of communication self-assess-
noglycosides were worse than those in CF patients not exposed ment inventories. J Speech Hear Disord, 49, 226–240.
or healthy controls despite similar DPOAE levels obtained from Dille, M. F., McMillan, G. P., Reavis, K. M., et al. (2010). Ototoxicity risk
assessment combining distortion product otoacoustic emissions with a
frequency sweeps and behavioral thresholds (Mulheran & Degg
cisplatin dose model. J Acoust Soc Am, 128, 1163–1174.
1997; Katbamna et al. 1999; Stavroulaki et al. 2002). These Dobie, R. A. (1983). Time for action in hearing conservation. Otolaryngol
findings suggest that DPOAE thresholds may be more sensitive Head Neck Surg, 91, 347–349.
to cochlear damage compared with more traditional DPOAE Dreisbach, L. E., & Siegel, J. H. (2001). Distortion-product otoacoustic
and behavioral measures. Due to methodological differences, emissions measured at high frequencies in humans. J Acoust Soc Am,
110(5 Pt 1), 2456–2469.
comparisons between the DPOAE thresholds reported in pre- Dreisbach, L. E., & Siegel, J. H. (2005). Level dependence of distortion-
vious studies and the present study for CF patients cannot be product otoacoustic emissions measured at high frequencies in humans.
made. However, based on the presumed sensitivity to cochlear J Acoust Soc Am, 117, 2980–2988.
damage and the repeatability of DPOAE thresholds over time Dreisbach, L. E., Long, K. M., Lees, S. E. (2006). Repeatability of high-
frequency distortion-product otoacoustic emissions in normal-hearing
in CF patients, this methodology should be explored further for adults. Ear Hear, 27, 466–479.
use in monitoring programs and for possible use in clinical trials Dreisbach, L., Ho, M., Reid, E. et al. (2017). Effects of oxaliplatin, carbo-
(Konrad-Martin et al. 2016). platin, and cisplatin across treatment on high-frequency objective and
subjective auditory measures in adults. Perspect ASHA Spec Interest Reavis, K. M., McMillan, G. P., Dille, M. F., et al. (2015). Meta-analysis of
Groups, 2 (Part 1), 17–36. distortion product otoacoustic emission retest variability for serial moni-
Engdahl, B., & Kemp, D. T. (1996). The effect of noise exposure on the toring of cochlear function in adults. Ear Hear, 36, e251–e260.
details of distortion product otoacoustic emissions in humans. J Acoust Ress, B. D., Sridhar, K. S., Balkany, T. J., et al. (1999). Effects of cis-
Soc Am, 99, 1573–1587. platinum chemotherapy on otoacoustic emissions: The development of
Fausti, S. A., Schechter, M. A., Rappaport, B. Z., et al. (1984). Early detec- an objective screening protocol. Third place–Resident Clinical Science
tion of cisplatin ototoxicity. Selected case reports. Cancer, 53, 224–231. Award 1998. Otolaryngol Head Neck Surg, 121, 693–701.
Fausti, S. A., Rappaport, B. Z., Frey, R. H., et al. (1991). Reliability of Rhode, W. S., & Cooper, N. P. (1993). Two-tone suppression and distortion
evoked responses to high-frequency (8–14 kHz) tone bursts. J Am Acad production on the basilar membrane in the hook region of cat and guinea
Audiol, 2, 105–114. pig cochleae. Hear Res, 66, 31–45.
Fausti, S. A., Henry, J. A., Helt, W. J., et al. (1999). An individualized, sen- Rizzi, M. D., & Hirose, K. (2007). Aminoglycoside ototoxicity. Curr Opin
sitive frequency range for early detection of ototoxicity. Ear Hear, 20, Otolaryngol Head Neck Surg, 15, 352–357.
497–505. Roede, J., Harris, F. P., Probst, R., et al. (1993). Repeatability of distortion
Frank, T., & Dreisbach, L. E. (1991). Repeatability of high-frequency product otoacoustic emissions in normally hearing humans. Audiology,
thresholds. Ear Hear, 12, 294–295. 32, 273–281.
Franklin, D. J., McCoy, M. J., Martin, G. K., et al. (1992). Test/retest reli- Roland, P. S. (2004). New developments in our understanding of ototoxicity.
ability of distortion-product and transiently evoked otoacoustic emis- Ear Nose Throat J, 83(9 Suppl 4), 15–16; discussion 16.
sions. Ear Hear, 13, 417–429. Ruggero, M. A., & Rich, N. C. (1991). Application of a commercially-man-
Gaskill, S. A., & Brown, A. M. (1990). The behavior of the acoustic dis- ufactured Doppler-shift laser velocimeter to the measurement of basilar-
tortion product, 2f1-f2, from the human ear and its relation to auditory membrane vibration. Hear Res, 51, 215–230.
sensitivity. J Acoust Soc Am, 88, 821–839. Sawicki, G. S., Sellers, D. E., Robinson, W. M. (2008). Self-reported physi-
Gilman, S., & Dirks, D. D. (1986). Acoustics of ear canal measurement of cal and psychological symptom burden in adults with cystic fibrosis.
eardrum SPL in simulators. J Acoust Soc Am, 80, 783–793. J Pain Symptom Manage, 35, 372–380.
Jacob, L. C., Aguiar, F. P., Tomiasi, A. A., et al. (2006). Auditory monitoring Siegel, J. H. (1994). Ear-canal standing waves and high-frequency sound
in ototoxicity. Braz J Otorhinolaryngol, 72, 836–844. calibration using otoacoustic emission probes. J Acoust Soc Am, 95(5),
Katbamna, B., Homnick, D. N., Marks, J. H. (1999). Effects of chronic 2589–2597.
tobramycin treatment on distortion product otoacoustic emissions. Ear Siegel, J. H. (2009). Estimating wideband sound levels in humans. J Assoc
Hear, 20, 393–402. Res Otolaryngol, 32, 11–12.
Kemp, D. T. (1978). Stimulated acoustic emissions from within the human Siegel, J. H., & Hirohata, E. T. (1994). Sound calibration and distor-
auditory system. J Acoust Soc Am, 64, 1386–1391. tion product otoacoustic emissions at high frequencies. Hear Res, 80,
Konrad-Martin, D., Poling, G. L., Dreisbach, L. E., et al. (2016). Serial 146–152.
monitoring of otoacoustic emissions in clinical trials. Otol Neurotol, 37, Souza, N. N., Dhar, S., Neely, S. T., et al. (2014). Comparison of nine methods
e286–e294. to estimate ear-canal stimulus levels. J Acoust Soc Am, 136, 1768–1787.
Lee, J., Dhar, S., Abel, R., et al. (2012). Behavioral hearing thresholds Stavroulaki, P., Vossinakis, I. C., Dinopoulou, D., et al. (2002). Otoacoustic
between 0.125 and 20 kHz using depth-compensated ear simulator cali- emissions for monitoring aminoglycoside-induced ototoxicity in children
bration. Ear Hear, 33, 315–329. with cystic fibrosis. Arch Otolaryngol Head Neck Surg, 128, 150–155.
Liberman, M. C., & Dodds, L. W. (1984). Single-neuron labeling and Stover, L. J., Gorga, M. P., Nelly, S. T., et al. (1996). Toward optimizing the
chronic cochlear pathology. III. Stereocilia damage and alterations of clinical utility of distortion product otoacoustic emission measurements.
threshold tuning curves. Hear Res, 16, 55–74. J Acoust Soc Am, 100(2 Pt 1), 956–967.
McMillan, G. P., & Hanson, T. E. (2014). Sample size requirements for Tan, K. H., Mulheran, M., Knox, A. J., et al. (2003). Aminoglycoside pre-
establishing clinical test-retest standards. Ear Hear, 35, 283–286. scribing and surveillance in cystic fibrosis. Am J Respir Crit Care Med,
Mulheran, M., & Degg, C. (1997). Comparison of distortion product OAE 167, 819–823.
generation between a patient group requiring frequent gentamicin ther- Telischi, F. F., Stagner, B., Widick, M. P., et al. (1998). Distortion-product
apy and control subjects. Br J Audiol, 31, 5–9. otoacoustic emission monitoring of cochlear blood flow. Laryngoscope,
Namysłowski, G., Morawski, K., Trybalska, G., et al. (2004). The latencies of 108, 837–842.
the 2f1-f2 DPOAE measured using phase gradient method in young adults Vaz, S., Falkmer, T., Passmore, A. E., et al. (2013). The case for using the
and in workers chronically exposed to noise. Otolaryngol Pol, 58, 131–138. repeatability coefficient when calculating test-retest reliability. PLoS
Neely, S., & Gorga, M. P. (1998). Comparison between intensity and pressure as One, 8, e73990.
measures of sound level in the ear canal. J Acoust Soc Am, 104, 2925–2934. Wagner, W., Heppelmann, G., Vonthein, R., et al. (2008). Test-retest repeat-
Neely, S. T., & Liu, Z. (1993). EMAV: Otoacoustic emission averager. Tech ability of distortion product otoacoustic emissions. Ear Hear, 29, 378–391.
Memo No. 17. Boys Town National Research Hospital, Omaha. Whitehead, M. L., Lonsbury-Martin, B. L., Martin, G. K. (1993). The influ-
Neely, S., & Stevenson, R. (2002). SysRes. Tech Memo No. 1. Boys Town ence of noise on the measured amplitudes of distortion-product oto-
National Research Hospital, Omaha, 1–10. acoustic emissions. J Speech Hear Res, 36, 1097–1102.
Ng, I.-Y., & McPerson, B. (2005). Test-retest reliability of distortion product Zhao, F., & Stephens, D. (1999). Test-retest variability of distortion-product
otoacoustic emissions in the 1 to 7 kHz range. Audiol Med, 3, 108–115.
otoacoustic emissions in human ears with normal hearing. Scand Audiol,
Poling, G. L., Siegel, J. H., Lee, J., et al. (2014). Characteristics of the 2f(1)-
28, 171–178.
f(2) distortion product otoacoustic emission in a normal hearing popula-
tion. J Acoust Soc Am, 135, 287–299.
Reavis, K. M., Phillips, D. S., Fausti, S. A., et al. (2008). Factors affecting REFERENCE NOTE
sensitivity of distortion-product otoacoustic emissions to ototoxic hear-
ing loss. Ear Hear, 29, 875–893. Major, I., & Dreisbach, L. (2010). Repeatability of ultra-high frequency
Reavis, K. M., McMillan, G., Austin, D., et al. (2011). Distortion-product DPOAEs in young normal-hearing listeners using an alternate calibra-
otoacoustic emission test performance for ototoxicity monitoring. Ear tion method: Bridging the gap from lab to clinic. Paper presented at the
Hear, 32, 61–74. California Academy of Audiology, San Francisco, CA.

High-Frequency Distortion-Product Otoacoustic Emission Repeatability in A Patient Population

Uploaded by

Copyright:

Available Formats

High-Frequency Distortion-Product Otoacoustic Emission Repeatability in A Patient Population

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

High-Frequency Distortion-Product Otoacoustic Emission Repeatability in A Patient Population

Uploaded by

Copyright:

Available Formats

High-Frequency Distortion-Product Otoacoustic Emission

Repeatability in a Patient Population

frequencies. First, the correlation of the absolute DPOAE levels

You might also like