Wideband Acoustic Immittance in Cochlear Implant Recipients: Reflectance and Stapedial Reflexes.

Abstract

Objectives.

To characterize differences in wideband power reflectance for ears with and without cochlear implants (CI), to describe electrically evoked stapedial reflex (eSR) -induced changes in reflectance, and to evaluate the benefit of a broadband probe for reflex threshold determination for CI recipients. It was hypothesized that reflectance patterns in ears with CI would be consistent with increased middle ear stiffness and that reflex thresholds measured with a broadband probe would be lower compared to thresholds obtained with single-frequency probe.

Design.

Eleven CI recipients participated in both wideband reflectance and eSR testing. Ipsilateral reflexes were measured with three probes: a broadband chirp (swept from 200 to 8000 Hz), a 226 Hz tone, and a 678 Hz tone. Wideband reflectance measures acquired from 28 adults without CIs and with normal middle ear function served as a normative data set for comparison.

Results.

Considering the group data, average reflectance was significantly higher for individuals with CI across 250–891 Hz and 4238–4490 Hz compared to the normative data set, though individual reflectance curves were variable. Some CI recipients also had low 226 Hz admittance, which contributed to the group finding, considering the control group had clinically normal 226 Hz admittance by design. Electrically evoked stapedial reflexes were measurable in 9 of 14 ears (64.3%), and in 24 of 46 electrodes (52.5%) tested. Reflex-induced changes in reflectance patterns were unique to the participant/ear, but similar across activator (electrode) within a given ear. Additionally, reflectance values at or above 1000 Hz were affected most by activating the stapedial reflex, even in ears with clinically normal 226 Hz admittance. This is a higher frequency range than has been reported for acoustically evoked reflex induced reflectance changes, and is consistent with increased middle ear stiffness at rest. Electrically evoked reflexes could be measured more often with 678 or the broadband probe compared to the 226 Hz probe tone. Although reflex thresholds were lower with the broadband probe compared to the 678 Hz probe in 16 of 24 conditions, this was not a statistically significant finding (Wilcoxon signed-rank test; p = 0.072).

Conclusions.

The applications of wideband acoustic immittance measurements (reflectance and reflexes) should also be considered for ears with CI. Further work is needed to describe changes across time in ears with CI to more fully understand the reflectance pattern indicating increased middle ear stiffness, and to optimize measuring eSRs with a broadband probe.

INTRODUCTION

The ability to preserve acoustic hearing following cochlear implantation has increased audiologists’ attentiveness to the function of the cochlear implant (CI) recipient’s entire auditory pathway, including the conductive mechanism. Even for traditional CI users without acoustic hearing, the status of the conductive pathway is relevant to consider with regards to measuring and interpreting electrically evoked stapedial reflexes (eSRs), but also as a general best practice of monitoring overall structure, function, and health of the system. The present investigation explores the use of wideband immittance to (1) characterize acoustic transmission in ears with CI and (2) measure eSRs.

Conductive Mechanism in Cochlear Implant Recipients

Some etiologies of hearing loss that lead to cochlear implantation, such as otosclerosis and CHARGE^* syndrome, involve the conductive mechanism in addition to the sensorineural system. Transient or chronic middle-ear dysfunction unrelated to the primary etiology of hearing loss can also be an issue for CI recipients. In addition, the CI surgery/surgical technique, iatrogenic injury or post-operative complications, and/or the body’s reaction to the device all have the potential to directly impact and change the structure and function of the conductive mechanism.

The middle ear is the entry point for inserting the electrode array into the cochlea, and the typical surgical approach to gain entry to the middle-ear space is a mastoidectomy (Roland & Sabatini, 2017). Although care is taken to avoid disturbing middle ear structures, such as the tympanic annulus and ossicles (Cohen, 1997; Balkany et al. 1999), there is inherent risk, and sometimes unavoidable impingement or injury, especially in cases of unusual anatomy. In some ears, a cochleostomy is drilled to allow for electrode insertion (creating, in effect, a “third window”, though it is closed). In other ears, the round window is used for cochlear access. Once the electrode array has been inserted into the cochlea, the lead is stabilized on the side of the middle ear. Examples include packing around various landmarks (e.g., cochleostomy or round window insertion site, facial recess and mastoid cavity), suturing, and sometimes clipping the electrode lead to the incus (Roland & Sabatini, 2017).

Although a standard surgery with implantation of an electrode array in the cochlea of an isolated cadaveric temporal bone has not been shown to impact energy transfer of the conductive mechanism, (Greene et al. 2015; Pazen et al. 2017), these procedures have the potential to trigger reactive processes in vivo. Postoperative complications are rare; however, retraction pockets or perforated tympanic membranes and cholesteatoma have been reported (e.g., Farinetti et al. 2014). Moreover, adhesions and fibrotic encapsulation of the electrode leads both within the cochlea and also extending into the middle ear space, have been observed (e.g., Roland & Sabatini, 2017). Any changes in the structure and anatomy of this mechanical system has the potential to impact its impedance (stiffness, mass and/or resistance), which in turn can affect the function of the conductive mechanism.

Implications for Electrically Evoked Stapedial Reflexes

Electrical stimulation from a CI is capable of exciting the afferent auditory pathway to the degree necessary to elicit stapedial muscle contraction (e.g., Jerger et al. 1986, 1988). The muscle contraction can be observed visually at the time of the surgical procedure (e.g., Lindstrom & Bredberg, 1997; Gordon et al. 2004; Caner et al. 2007; van den Abbeele et al. 2012), but the more common approach is the use of an immittance bridge post-operatively as the measurement tool (e.g., Jerger et al. 1986, 1988; Spivak & Chute, 1994; Spivak et al. 1994; Hodges et al. 1997; Stephan & Welz-Müller, 2000; Bresnihan et al. 2001; Polak et al. 2005; Brickley et al. 2005; Wolfe & Kasulis, 2008; Kosaner et al. 2009; Gordon et al. 2012; Asal et al. 2016). Despite a relatively large body of evidence supporting the incorporation of eSR measures to assist with programming the electrical dynamic range (e.g., Spivak & Chute; Spivak et al. 1994; Hodges et al. 1997; Bresnihan et al. 2001; Gordon et al. 2005; 2012; Polak et al. 2005; Wolfe & Kasulis, 2008; Kosaner et al. 2009), eSRs are rarely performed (Vaerenberg et al. 2014). One limitation is that eSRs are not present in all CI recipients. The reported proportion of successful post-operative, immittance-based eSR recordings ranges from 62 to 84% of CI users (Battmer et al. 1990; Spivak & Chute, 1994; Hodges et al. 1997; Bresnihan et al. 2001; Gordon et al. 2004; Brickley et al. 2005; Wolfe & Kasulis, 2008), with lower success rates reported closer to the time of surgery (Gordon et al. 2004).

The reason for absent reflexes in some CI recipients is not always obvious (e.g., Spivak & Chute, 1994; Brickley et al. 2005; Kosaner et al. 2009; Asal et al. 2016). In some cases, insufficient afferent neural excitation is suspected (Battmer et al. 1990; Asal et al. 2016). A history of middle-ear disease or evidence of middle-ear dysfunction is also a likely contributor to absent reflexes (e.g., Bresnihan et al. 2001; Brickley et al. 2005; Wolfe & Kasulis, 2008; Kosaner et al. 2009; Asal et al. 2016). But in general, there is a suspicion that the conductive mechanism in the implanted ear might be altered. Therefore, for eSR measurements in unilateral recipients, the immittance probe has traditionally been placed in the contralateral, non-implanted ear to avoid a surgically modified middle ear system.

Recently it has been demonstrated that eSRs are observed more often and at lower activator levels when using a higher frequency probe (678 or 1000 Hz) compared to the standard 226 Hz probe tone (adults: Wolfe et al. 2017; children: Hernandez et al. 2018). In cases where the probe was placed in the contralateral, non-implanted ear, the finding suggests a general advantage of the higher-frequency probe (Wolfe et al. 2017). However, Hernandez and colleagues (2018) observed the advantage only in implanted ears, which suggests that adopting this methodology might partially compensate for suspected changes in the middle-ear system following cochlear implantation.

Measuring Conductive Changes Post Implantation

For individuals with sufficient acoustic hearing preservation following cochlear implantation, a comparison of air- and bone-conduction thresholds permits identification of changes in the conductive mechanism. In some individuals, transient changes in the size of the air-bone gap have been observed immediately following the surgery, presumably due to fluid in the middle-ear space (Lenarz et al. 2006). However, in others, the increased air-bone gap persists in the absence of any evidence of middle-ear pathology (e.g., Chole et al. 2014; Mattingly et al. 2016). Although it is important for clinics to monitor air- and bone-conduction thresholds for individuals with hearing preservation, once bone-conduction thresholds exceed a severe degree (i.e. the output limits of the audiometer), it may not be possible to identify or quantify an air-bone gap.

Laser Doppler vibrometry is another technique that can be used to assess the conductive mechanism and does not rely on acoustic hearing. Wasson and colleagues (2018) demonstrated that on average, changes in air- and bone-conduction umbo velocity pre- and 3-months post-operatively were not statistically significant at any frequency (160 – 16,000 Hz) in newly implanted ears compared to the subjects’ contralateral ears (some were previously implanted). However, trends in the mean data from newly implanted ears were consistent with stiffening of the stapes and increased ossicular chain mass (Wasson et al. 2018). Individual stapes velocity data revealed a more complicated picture. In some ears, pre-and post-operative measurements were essentially equivalent, but in others stapes mobility decreased, and in others, stapes mobility increased (Donnelly et al. 2009; 250 – 2000 Hz).

Although a promising tool, laser Doppler vibrometry is not practical in a clinical setting. Standard 226 Hz tympanometry is a widely used measurement in diagnostic audiology for identifying conductive pathologies; however, it is not the most sensitive (e.g., Keefe & Simmons, 2003; Beers et al. 2010; Prieve et al. 2013; Mattingly et al. 2016). An alternative technique is wideband acoustic immittance. Wideband immittance can be performed with clinically available instruments (though not widely used to date), does not require residual acoustic hearing, and is more sensitive to conductive pathology than standard 226 Hz tympanometry.

Wideband Acoustic Immittance

Wideband Reflectance •

Wideband reflectance has been explored extensively in non-implanted ears.^† One motivation driving this work is the desire to improve upon pediatric screening and diagnostic assessments (reviewed by Hunter et al. 2013). Wideband reflectance is also useful for differential diagnoses among conductive pathologies, as the reflectance pattern is affected in predictable ways by various middle ear disorders. Patterns have been described for pathologies involving the tympanic membrane (perforation, flaccidity, or scarring), middle-ear cavity (positive or negative pressure or effusion), ossicles (stapes fixation or ossicular discontinuity) and cochlea (superior semicircular canal dehiscence or increased intracranial pressure; reviewed by Nakajima et al. 2013). These patterns are potentially useful for exploring the various changes to the conductive mechanism that may occur following cochlear implantation.

Stapedial Reflexes •

Wideband acoustic immittance also includes measuring stapedial reflexes (reviewed by Schairer et al. 2013). Stapedial muscle contraction affects reflectance across a broad frequency range (e.g., 200–8000 Hz); the effect is greater at some frequencies and differs across individuals. A primary advantage of using a broadband probe compared to the conventional single-frequency probe is that the maximum change in immittance may occur at a frequency other than conventional probe frequencies (226 or 678 Hz). Acoustic reflexes measured with a broadband probe tend to be observed at lower activator levels (see Table 1 in Schairer et al. 2013 for a summary). The wideband acoustic immittance technique has recently been adopted for measuring eSRs (Scheperle et al. 2018, Reference Note 1; Wolfe et al. 2018). Wolfe and colleagues (2018) observed lower eSR thresholds (eSRTs) with the broadband probe compared to a 226 Hz probe tone; however, it is unclear whether these results reflect a general benefit of broadband probes (also observed in non-implanted ears) or whether the benefit is greater in ears surgically modified by cochlear implantation with a potentially altered middle ear system.

Table 1.

Demographics and Characteristics for Participants with Cochlear Implants

ID	Sex	Race/ Ethnicity	Age (yr)	Etiology	Onset, Progression	Ear(s) Implanted	Right Device	Left Device
C001	M	White	55	Unknown	Childhood, Progressive	Bilateral	CI24RE-CA	CI24RE-CA
C002	F	White	26	Unknown	Childhood, Progressive	Unilateral Left	Hearing Aid	CI24RE-CA
C005	F	White	67	Unknown	Adult, Progressive	Unilateral Left	Hearing Aid	CI522
C006	F	White	68	Unknown	Adult, Sudden	Unilateral Left	Hearing Aid	CI24R-CS
C007	F	White	59	Unknown	Childhood, Progressive	Bilateral	CI512	CI24RE-CA
C008	F	White	77	Meningitis	Adult, Sudden	Bilateral	CI24RE-CA	CI24RE-CA
C009	M	White/ Jewish	72	Genetic	Childhood	Unilateral Left	--	CI22M
C010	M	Asian	13	Unknown	Childhood, Progressive	Bilateral	CI24RE-CA	CI24RE-CA
A001	M	White	69	Unknown	Adult, Progressive	Unilateral Right	HiRes90K MidScala	Hearing Aid
A002	M	Black	22	Meningitis	Childhood, Sudden	Unilateral Right	C1.2	Hearing Aid
A003	F	White	78	Otosclerosis	Adult, Progressive	Unilateral Left	CROS	CII-HiFocus1J

Identifier (ID), Male (M), Female (F), Contralateral routing of signal (CROS). Notes: C001: Multiple explants/re-implants in both ears due to infection and device failure. C002: Exploratory left ear surgery for a potential fistula. A003: Trial stapedectomy in the right ear (unsuccessful).

Purpose

This study has 2 aims: (1) to characterize reflectance patterns in CI recipients and (2) to test the repeatability and robustness of previous findings of increased eSR incidence and lower eSRTs with a broadband probe (Wolfe et al. 2018) when using the ipsilateral, implanted ear for recording. It is predicted that wideband energy reflectance curves will be more variable in the ears with CI than the non-implanted controls but also that the average pattern will be consistent of a system with increased middle ear stiffness (higher reflectance at low frequencies than in non-implanted ears). We also predict that the eSRT findings of Wolfe and colleagues (2018) will be replicable when using an ipsilateral probe, and expect the same amount of benefit, if not greater, given that the probe ear will always be an ear that has undergone CI surgery.

MATERIALS AND METHODS

This study was approved by the Montclair State University Institutional Review Board (IRB16–17-479: CI recipients, IRB17–18-1042: normative data collection). Otoscopy, tympanometry, and wideband reflectance measurements were performed and ear impressions obtained on all participants (those with and without CIs). The ear impressions were used to adjust the reflectance measurement for ear canal size, which contributes to the variability attributed to age, sex, and ethnicity (Voss, Reference Note 2). Electrically evoked SRs (single-frequency and broadband probe) were measured in ears with CIs (probe on the ipsilateral, implanted side). All participants spoke English.

Participants

Cochlear Implant Recipients •

Thirteen CI recipients responded to recruitment efforts for this study. History of chronic middle ear infections or ear surgeries (other than cochlear implantation) in the test ear was negative for most except C001, C002 and A003 (see Table 1 notes). Cochlear implant participants were not excluded based on tympanometric results. This approach was more relaxed than criteria used by Wolfe et al. (2018), and more relaxed than the criteria used for the normative, non-implanted group in an effort to capture the variability within the population of interest and to gain insight into the potential clinical impact of modifying the eSR methodology. One CI participant was excluded because of upper respiratory symptoms at the time of the appointment. Another participant was tested during the pilot stage but opted out of ear impressions, so her data were also excluded. Of the remaining eleven participants, four presented with occluding cerumen in one or both ears at the initial appointment. Two rescheduled and were tested following management. For the other two, the cerumen affected their non-implanted, contralateral ear, so only the implanted ear was tested.

Demographic data for the included participants with CI are provided in Table 1. Four were bilaterally implanted and seven were unilaterally implanted with devices from Cochlear Ltd (N=8; identifier begins with “C”) or Advanced Bionics (N=3; identifier begins with “A”). Ages ranged from 13 to 78 years (mean: 55 years). One participant (ID: C005) was seen 2 months following initial activation. All other participants were long-term users (ranging from 3 to 25 years). Participant C005 also had residual hearing in the implanted ear of a severe to profound degree. The presence of an air-bone gap could not be determined due to the degree of the hearing loss and vibrotactile responses at low frequencies.

Normative Data Set •

Thirty-four individuals without CIs (the control group) were recruited for the collection of system-specific normative wideband reflectance data to use for comparison with data from implanted ears. Although the differences Shahnaz and colleagues (2013) observed when comparing measurements obtained with the Mimosa Acoustics HearID or the Interacoustics systems were not considered clinically significant (i.e. smaller than differences caused by middle ear pathologies), this study used a custom system, and the goal was to quantify potentially subtler effects on the middle ear than pathology induced changes. Inclusion criteria required clear ear canals, bone-conduction thresholds within 10 dB of air-conduction thresholds for 250 to 4000 Hz, 226 Hz tympanometric peak pressure between −83 and +50 daPa, and peak compensated static acoustic admittance between 0.3 and 1.4 mmho (Liu et al. 2008). Otologic history was negative for previous ear surgery or recurrent ear infections (multiple sets of tubes). Normal hearing sensitivity was not required. Six individuals were excluded because neither ear met one or more inclusion criteria (e.g., high admittance, notched tympanograms, and air-bone-gaps > 10 dB at 1 or more frequencies). For the remaining 28 participants (21 females; 7 males), one ear was selected for subsequent testing. If one ear did not meet the inclusion criteria, the opposite ear that did was selected. If both ears met the inclusion criteria, ears were selected to avoid potential discomfort or issues with ear impressions (e.g., tragal piercings or excessive cerumen) and to provide equal numbers of left and right ears.

The average age of the participants in control group was 40 years (ranging from 21 to 67 years), which is younger than the experimental group of CI recipients. Similar to the participants with CI, the majority indicated White for race/ethnicity; only seven indicated a race/ethnicity other than white (Middle Eastern: N=1; Black: N=2; Asian, N=1; White/Hispanic: N=1; Other: N=1; No response: N=1).

Tympanometry

The Titan (Interacoustics, Eden Prairie, MN) immittance system was used for 226 Hz tympanometric measurements. TympStar (Grason-Stadler, Eden Prairie, MN) results are reported for two subjects due to difficulty obtaining a pressure seal with the Titan probe. Estimates of ear canal volume (ECV; mL), peak-compensated static acoustic admittance (Ya; mmho), and tympanometric peak pressure (TPP; daPa) were recorded for reference (Table 2). The positive tail was used for ECV estimates.

Table 2.

Stimulation Parameters for Electrically Evoked Stapedial Reflexes

ID	Ear	Software	Rate (pps)	Mode	PPD (μsec)	Electrodes
C001	L	CSEP	900	MP	50, 50, 37, 37, 37, 37	2, 5, 8, 12, 15, 19
C001	R	CSEP	900	MP	100, 100, 75, 75, 50, 50	2, 5, 8, 12, 15, 19
C002	L	CSEP	900	MP	37, 37, 37	5, 12, 19
C005	L	CSEP	900	MP	25, 25, 25	5, 12, 19
C006	L	CSEP	900	MP	25, 25, 25	5, 12, 19
C007	L	CSEP	900	MP	37, 37, 37	5, 12, 19
C007	R	CSEP	900	MP	37	19
C008	L	CSEP	900	MP	25, 25, 25	5, 12, 19
C008	R	CSEP	900	MP	25, 25, 25	5, 12, 19
C009	L	CSEP	900	BP+5	50, 50, 50	5, 12, 16
C010	R	CSEP	900	MP	25, 25, 25	5, 12, 19
A001	R	SoundWave	829	MP	75.4, 75.4, 75.4	15, 9, 3
A002	R	SoundWave	1444	Radial pairs	75.4, 75.4, 75.4	L8, L4, L1
A003	L	BEDCS2	1000	MP	25, 25, 25	14, 8, 3

Custom Sound EP (CSEP), Bionic Ear Data Collection System (BEDCS), Monopolar (MP), Bipolar (BP), Pulses per second (pps), Pulse Phase Duration (PPD). Bold formatting indicates that eSRs were observed in at least one condition.

Wideband Acoustic Immittance

Wideband acoustic immittance measurements (reflectance and broadband reflexes) were performed at ambient pressure using MATLAB-based software package (Auditory Research Lab audio software [ARLas] provided by Shawn S. Goodman, University of Iowa; MathWorks 2015b, Natick, MA). Hardware included an RME BabyFace Pro soundcard (RME, Germany) and an ER10C probe microphone system (Etymōtic Research, Elk Grove Village, IL). A chirp stimulus was used for all measurements (0.2-sec linear frequency sweep from 200 to 8000 Hz). Sampling rate was 44.1 kHz for all recordings. An artifact rejection criterion of 2.25 times the interquartile range was used to reject individual recordings that were collected together. For reflectance measurements, up to 256 sweeps were averaged. For reflex measurements, up to 32 sweeps were averaged (two sets of 16) with the exception of one subject: four sets of 16 sweeps were obtained for C009 due to noisiness of the recordings. The averaged waveforms were smoothed by taking the mean of across non-overlapping 1/24^th-octave bands starting at 250 Hz.

Thévenin-Equivalent Source Characteristics •

A 5-tube acoustic calibration procedure was used to estimate the probe (source) impedance and pressure (e.g., Scheperle et al. 2008). A single brass tube (inside diameter; i.d. = 0.79 cm) with an internal, moveable steel rod was used to create test cavities with lengths of 1.85, 2.56, 4.0, 5.4, and 8.3 cm (Scheperle et al. 2011). System calibration was performed each day an appointment was scheduled and was repeated/replaced if the load pressure error was ≥ 1.0 (Eq. 6 in Scheperle et al. 2011). Of 63 calibrations made during this time period, 51 met the criterion. However, the variability across acceptable calibrations revealed clear outliers. Much of the variability is attributed to variations in the coupling of the probe tip to the test cavity (e.g., Burke et al. 2010). Currently, there is no standard procedure for identifying the “true” Thévenin-equivalent source parameters.

In an attempt to improve upon the estimated Thévenin-equivalent source parameters and decrease the variability introduced by the daily calibration procedure, a robust-averaging technique was employed that rejected outliers, resulting in a single source calibration (inspired by Mishra and Talmadge, 2018). Phase-detrended complex pressure amplitudes >1.5 times the interquartile range (IQR) around the median were rejected, along with the corresponding complex impedance. Details can be found in the Supplement. This single source calibration file was used for all reflectance estimates.

Characteristic Impedance Estimation •

Wideband acoustic immittance calculations require an estimate of the characteristic impedance (Z₀) of the acoustic cavity. In perfectly cylindrical tubes, Z₀ can be calculated from

$$Z_0=\rho c/A$$ Eq 1.

where ρ is density of air and c is the speed of sound, which are both constants (Keefe, 1984), and A is the cross-sectional area. For the calibration tubes used in this study, Z₀ =83.1 acoustic ohms. In some wideband acoustic immittance studies, the Z₀ associated with the calibration tubes is used as the estimate for all participants (e.g., Scheperle et al. 2008), but this can lead to errors (Scheperle et al. 2011). In some studies, the size of the eartip has been used for a more individualized estimate (e.g., Withnell et al. 2009). An acoustically derived individualized estimate for non-uniform cavities has also been proposed (Rasetshwane & Neely, 2011). For this study, silicone impressions of the ear canal were used for individualized estimates of the cross-sectional area at 12 mm (the length of the ER10C foam tip) from the tragus (Voss, Reference Note 2). These areas ranged from 25.5 to 123.2 mm² (mean = 70.5). Individualized estimates of Z₀ calculated from these areas (Eq. 1) ranged from 33.27 to 160.75 acoustic ohms (mean = 63.84).

Wideband Reflectance

Wideband reflectance calculations were made using the robust-averaged Thévenin-equivalent source impedance and pressure, and the individualized estimates of Z₀. For 8 CI recipients (4 bilateral and 4 unilateral), wideband reflectance was measured in both ears.

Electrically Evoked Stapedial Reflexes in Ears with Cochlear Implants

The probe was placed in the ear ipsilateral to the side of CI stimulation, for both single-frequency and broadband recordings. Depending upon time and availability for return appointments, 1 to 6 electrodes (spaced across the array) were tested per ear (Table 2).

Stimulation •

Single-electrode electrical stimulation consisted of cathodic-leading, charge-balanced, biphasic pulse trains. For participants with devices manufactured by Cochlear Ltd. (Sydney, NSW, Australia), a laboratory Freedom processor was used for stimulation. The processor was connected to a personal computer using a programming pod. CustomSound EP 5.0 software (Cochlear Ltd., Sydney, NSW, Australia) was used to control single-electrode stimulation. For participants with devices manufactured by Advanced Bionics (Valencia, CA), a laboratory Naida Q70 processor controlled with Soundwave 3.0 or BEDCS2 (2.0.6058.27146; a MATLAB interface) was used for stimulation. Amplitude and duration of the current pulses were adjusted for the individual ear and electrode to provide sufficient loudness growth within the compliance limits of the device.

Electrical Loudness Ratings •

Loudness ratings were obtained on each electrode using a clinical chart provided by Advanced Bionics, which spans from 0 (off) to 10 (too loud). The purpose was to determine the electrical dynamic range and avoid presenting stimuli that would be uncomfortably loud. Pulse trains were delivered as 3 sets of 500-ms bursts (when using CustomSound EP and BEDCS2) or using the tone burst selection in SoundWave 3.0. Pulse amplitude was adjusted in an ascending order. For Cochlear devices, the pulse amplitude was increased using a step size of 5 clinical level (CL) units (approximately 0.78 dB). For Advanced Bionics devices, pulse amplitude is expressed in charge units ($CU=\mu A\times PPD\times 0.013$). Pulse phase duration (PPD) was held constant and amplitude (μA) was adjusted using a step size of approximately 0.78 dB.

Reflexes Measured with Single-Frequency Probe Tones •

Reflex thresholds were measured with a 226 Hz probe tone and again with a 678 Hz probe tone using a TympStar immittance bridge. The system was set to the reflex decay mode. The level of the acoustic activator could not be disabled, so it was set to the lowest output possible (45 dB HL) and routed to the contralateral transducer, which was physically disconnected. The ear canal was pressurized to the TPP.

The tester manually coordinated the initiation of recording (TympStar) and stimulation (computer with CustomSound EP, SoundWave, or BEDCS2). The recording window was started first, followed by activation of the electrical stimulus. The electrical activator for the first presentation was set to the CL or CU associated with a loudness rating of 6 (most comfortable). Immittance recordings were evaluated in real time for deflections that were time-locked with stimulation from the implant (three 500-ms pulse trains with 500-ms of silence in between for CSEP and BEDCS2). Pulse amplitude was increased or decreased (in the CL or CU corresponding to approximately 0.78 dB) for subsequent measurements depending upon whether a reflex-induced change in immittance was observed or suspected. Reflex thresholds were considered the lowest level at which a response could be visually detected at least twice. A deflection criterion of 0.02 mmho was used as a reference, but the size of the response was not measured.

Reflexes Measured with a Broadband Probe •

The ER10C probe-microphone system was used for the broadband reflex recordings. A bursting activator is conventional to assist with time-domain visual detection of a response when using a single-frequency probe tone. Because averaging was required to reduce the variance for frequency domain analysis, a 5-second, ongoing pulse train was used as the activator to enable a more stable muscle contraction across multiple recordings of the broadband probe signal.

As before, the tester manually coordinated initiation of recording (via a computer with MATLAB) and stimulation (via a separate computer with CustomSound EP, SoundWave, or BEDCS2). Activator stimulation was initiated first to induce reflexive muscle contraction prior to starting the signal averaging of the broadband probe, which was triggered manually by the tester. Sixteen recordings were obtained during each 5-sec presentation of the electrical activator. Each activator condition (electrode and stimulus level) was tested twice and combined offline to double the total number of sweeps to be averaged.

Because reflectance-based eSRs could not be determined in real time, activator levels were pre-determined by the single-frequency probe reflex measurements and the participant’s electrical dynamic range for a given electrode. Because acoustically evoked reflex thresholds measured with a broadband probe are often lower than observed with single-frequency probes (e.g., Schairer et al. 2013) testing included a greater number of stimulation levels below the reflex threshold than above. If a reflex had been observed with either the 226 or 678 Hz probe, activator levels were selected to range from the equivalent of 18 CL below to 6 CL above the lowest threshold (3-CL step size). If any of these activator levels would exceed the upper comfort limit of the individual, or if a reflex had not been observed with the single-frequency probe, then the highest CL that was not uncomfortable was selected as the maximum, and the remaining levels were set lower (step-size of 3 CL down to 24 CL below the maximum, or equivalent CU values). A “no stimulus” condition was interleaved between each condition with an activator. In a few instances, a response was observed at the lowest level tested. When possible, participants were rescheduled for additional testing with lower stimulus levels.

Offline Calculations •

Differences were calculated between reflectance measurements obtained during activator stimulation and those obtained during the previously measured “no stimulus” conditions (baseline). See Figure 1 for an example pair of reflectance curves (panel A) and the difference curve (${\mathcal{R}}_{activator}-{\mathcal{R}}_{baseline}$; panel B).

Figure 1.

Panel A: Pair of reflectance curves for participant C001L during the baseline (light gray) recording and when activating the stapedial reflex with stimulation from electrode 5 (black). Dark gray shaded region is the 5^th-95^th percentile from the control ears. Panel B: Difference curve $({\mathcal{R}}_{activatoron}–{\mathcal{R}}_{baseline})$.

Differences were also calculated across interleaved “no stimulus” conditions to estimate a noise floor.

The root-mean-square (RMS) amplitude of a 2-octave frequency range was calculated for both sets of difference curves: (1) activator-on versus baseline and (2) pairs of baseline curves for noise-floor estimation. For each series of recordings, the analysis range was determined by the difference curve calculated from the highest activator level. The maximum difference for frequencies ≥ 500 Hz was identified. The frequency restriction was necessary to avoid poor signal-to-noise ratios at lower frequencies. The analysis range extended 1-octave below and above the frequency location of the maximum. Although initial selection was automated, the range was reviewed and, in some cases, manually adjusted to capture a more robust response region. The RMS amplitude was considered along with visual inspection of the difference curves for threshold determination (Fig 2).

Figure 2.

Example eSR data set for C001L, electrode 5. Left panel: Series of reflex-induced changes in reflectance for the range of activator levels used for threshold estimation. Gray shaded region is the 2-octave frequency range for the RMS calculations. Right panel: Associated RMS amplitudes for each difference curve. Black circles: calculated from the activator-on difference curves. Gray asterisks: noise floor estimate from the interleaved baseline conditions. Horizontal dotted line is at the mean and dashed line is +1 SD of the noise-floor estimates.

Statistical Analyses

Grand mean reflectance measurements for the normative, non-implanted ears and the ears with CI were compared at each 1/24 octave band using a two-tailed Wilcoxon rank-sum test. Reflex thresholds for the 678 Hz and broadband probe conditions were compared first by calculating dB differences and then testing for significance using the Wilcoxon signed-rank test. The significance criterion for both tests was 0.05.

RESULTS

Tympanometry

Table 3 provides tympanometric results for the individuals with CI, separated by implanted and non-implanted ears. The grand means and standard deviations for the individuals without CI (i.e. controls) are also provided for comparison in the last 2 rows. Some caution is warranted in the comparison because the control ears were not chosen at random but met the strict inclusion criteria described previously; whereas, inclusion for implanted ears was less strict.

Table 3.

Tympanometric Results

		IMPLANTED				NON-IMPLANTED
ID	Ear	ECV (mmho)	Pressure (daPa)	Ya (mmho)	Ear	ECV (mmho)	Pressure (daPa)	Ya (mmho)
C001	R	1.26	−14	0.2
C001	L	1.41	38	0.17
C007	R	1.76	−7	0.48
C007	L	1.55	−11	0.72
C008	R	1.4	10	0.3
C008	L	1.7	20	1.5
C010	R	0.87	0	0.3
C010	L	0.85	5	0.19
C002	L	1.13	−19	0.62	R	1.31	−1	0.73
C005	L	1.81	−47	0.83	R	2.1	−14	1.23
C006	L	1.62	−71	0.88	R	DNT	DNT	DNT
C009	L	2.70	40	2.60	R	2.50	20	1.00
A001	R	1.25	9	0.48	L	1.57	−1	0.63
A002	R	1.26	7	0.52	L	1.35	−55	0.91
A003	L	1.37	−6	0.23	R	DNT	DNT	DNT
	Mean (SD)	1.46 (0.45)	−3.07 (28.75)	0.67 (0.65)		1.77 (0.52)	−10.20 (27.85)	0.90 (0.23)
				Controls N=28	Mean (SD)	1.25 (0.28)	−2.36 (10.8)	0.64 (0.23)

Note: Standard deviation (SD). Millimhos (mmho). decaPascals (daPa). Number (N). Left (L). Right (R). Did not test (DNT). Two ears were not tested due to excessive cerumen.

Although average compensated static acoustic admittance (Ya at 226 Hz) of the 15 implanted ears is lower than the average of the non-implanted ears of CI recipients (0.67 compared to 0.90 mmho), it is similar to the average of the control ears (0.64 mmho). However, the variability of Ya observed across ears following implantation is greater than the variability observed across non-implanted ears of the CI recipients and the normative data set. Focusing on individual data, Ya was <0.3 mmho (i.e., stiff) in 4 implanted ears and >1.4 mmho (i.e., hypercompliant) in 2 implanted ears. To summarize, 6 of 15 (40%) implanted ears had clinically significant immittance results while none of the immittance results in non-implanted ears of the CI recipients were clinically significant.

Wideband Reflectance

Wideband power reflectance measurements are displayed in Figure 3. The shaded region in each panel is the 5^th – 95^th percentiles calculated from the normative data set (excluding N005; explained below). Individual reflectance curves, shown as solid lines, are separated into the 3 panels by group: normative control (left) and experimental (middle: implanted ears; right: non-implanted ears). The smallest eartip could not be fully inserted into N005’s ear canal and the largest eartip was too small for C009. These data were excluded from the remaining wideband reflectance analyses. Due to the limited amount of reflectance data in the non-implanted ears of CI recipients, these data (Fig 3, right panel) are solely provided for illustration purposes and will not be discussed further.

Figure 3.

Wideband power reflectance for 3 groups: controls (left panel), CI recipients (implanted ear; middle panel), and CI recipients (non-implanted ear; right panel). The gray shaded region in all 3 panels is the 5^th – 95^th percentile from the control group. Dotted lines are data that were excluded due to poor probe fit.

A trend toward higher low-frequency reflectance is observed for the implanted ears compared to the normative data set. One exception is the reflectance curve with a minimum around 660 Hz from an ear with a hypercompliant tympanogram (C008L). Low-frequency notches have been observed in ears with ossicular discontinuity (Nakajima et al. 2013).

Figure 4 displays the group means and standard deviations of the reflectance curves for the ears in the normative group (solid line: mean; black outline: ±1 SD) and ears with implants (dashed line: mean; light gray outline: ±1 SD). Note the higher mean reflectance and larger standard deviation specifically at low frequencies for the implanted ears compared to the mean reflectance for the normative data set. At high frequencies, the means and standard deviations are similar. The Wilcoxon rank-sum test was significant for frequencies 250 to 891 Hz (p < 0.007) and 4238 to 4490 Hz (p < 0.05; indicated by asterisks in Fig 4).

Figure 4.

Cross-group comparison of wideband power reflectance (means ± 1 SD). Normative data are shown in black and data from ears with CIs in light gray. Asterisks indicate the frequencies that were statistically different across groups (p < 0.05).

Figure 5 compares reflectance curves in implanted ears with and without abnormal Ya (dotted and solid lines, respectively). All reflectance measurements in ears with abnormal Ya (dotted) were outside of the 5^th to 95^th percentile of the normative data at some frequencies (primarily low frequencies). More notable is that half of the reflectance curves from ears with normal Ya (solid) are above the 95^th percentile of the normative data at low frequencies.

Figure 5.

Wideband reflectance for ears with normal (solid) compared to abnormal (dotted) 226 Hz admittance overlaid on the 5^th – 95^th percentile from the control group (gray shaded region).

It is expected that wideband acoustic immittance measures in ears with abnormal 226 Hz immittance measurements will deviate from the typical pattern. It is the ears with normal 226 Hz Ya that are of particular interest to assess whether reflectance measures are providing additional information about the conductive mechanism. A limitation of the clinical interpretation of Ya (normal versus abnormal) is that with this categorical approach, “normal” includes a large range of values (0.3 to 1.4 mmhos, inclusive). Therefore, 226 Hz Ya was treated as a continuous variable and correlated with low-frequency reflectance (averaged across 250 to 1000 Hz). Figure 6 is a scatterplot of these 2 variables.

Figure 6.

Scatterplot of average low-frequency power reflectance as a function of 226 Hz Ya for all ears: gray circles: normative data; white squares: CI recipients with abnormal Ya; black squares: CI recipients with normal Ya. Linear fit to all data is provided. Asterisks denote data with studentized residuals > 2.0.

Ears with CIs are plotted as squares; shading denotes whether the 226 Hz Ya was considered clinically abnormal (white) or normal (black). Data from the normative group (226 Hz Ya was normal by design) are shown as gray circles. The expected negative relationship (greater reflectance for ears with low 226 Hz Ya and lower reflectance for ears with greater 226 Hz Ya) is observed. Also, the data from the CI users with normal 226 Hz Ya largely overlap with the data of the normative group. Two potential exceptions are the data for C008L and C006L; studentized residuals were −2.07 and 2.71, respectively. Using a criterion value of >±2.0, 1 subject in the normative data set (N007) would also be considered an outlier (−2.83). The statistically higher low-frequency reflectance in ears with CIs (observed in the group mean data; Fig 4) is largely explained by the 226 Hz Ya when it is considered as a continuous variable rather than a categorical variable.

Electrically Evoked Stapedial Reflexes

Reflexes were measurable in 9 of 14 ears tested (64.3%) and in 24 of 46 electrodes tested (52.2%) with at least one probe condition (Table 2). Ears with abnormal 226 Hz Ya were not necessarily the ears with unmeasurable reflexes. For example, the individual with the lowest 226 Hz Ya had robust reflexes across the electrode array.

Table 4 provides a comparison of eSRTs across the probe conditions. Stimulus level was converted to μA. The final column provides a threshold difference in dB between the 678 Hz and reflectance-based thresholds. A negative number indicates that the reflectance-based threshold was lower than the 678 Hz threshold.

Table 4.

eSRT Comparison Across Probe Conditions

ID	Activator Electrode	Single-Hz eSRT (μA) 226 / 678 Hz	ℛ eSRT (μA)	Difference (dB)
C001R	5	CNT / 592	<344	>−4.71
C001L	2	CNT / 494	338	−3.29
C001R	12	CNT / 452	<315	>−3.14
C001L	5	CNT / 592	503	−1.41
C001L	15	592 / 592	531	−0.94
C001L	12	CNT / 592	531	−0.94
C001R	19	CNT / 452	420	−0.63
C001R	15	CNT / 413	384	−0.63
A002R	L8	>153 / > 153	144	>−0.54
C006L	5	>709 / > 709	672	>−0.47
C006L	19	>776 / 776	735	−0.47
C001L	8	CNT / 709	672	−0.47
C001R	2	CNT / 413	391	−0.47
A001R	9	306 / 286	275	−0.32
C007L	12	522 / 522	513	−0.16
C001R	8	CNT / 315	309	−0.16
C002L	19	377 / 377	377	0.00
C007L	19	494 / 452	460	0.16
C001L	19	541 / 494	513	0.31
C002L	5	219 / 219	232	0.47
C002L	12	344 / 315	332	0.47
C008R	12	CNT / 288	>344	>1.57
C009L	16 to 22	1258 / 1181	>1339	>1.09
C009L	12 to 18	1508 / 1258	>1508	>1.57

Could not test (CNT) 226 Hz due to excessive random fluctuations of the probe.

Single-frequency eSRTs •

Reflexes were observed in 8 of the 9 ears with measurable responses, and in 22 of the 46 electrodes tested when using a single-frequency (226 or 678 Hz) probe. Participant A002R did not have a measurable reflex with a single-frequency probe tone, but did in the broadband probe condition. The advantage of using a higher frequency probe tone that has been reported previously (Wolfe et al. 2017; Hernandez et al. 2018) was also observed in this data set. For one participant in particular (C001), the 226 Hz recordings fluctuated too much to be reliably interpreted, but reflexes could be reliably recorded using the 678 Hz probe tone. Even when random probe fluctuation was not an issue, the 678 Hz eSRTs were lower than the 226 Hz eSRTs in 7 instances; there were no instances where the 678 Hz eSRTs were higher than the 226 Hz eSRTs (Table 4).

Broadband eSRTs •

The pattern of reflectance changes, indicating a reflex, was similar across activator electrodes for a given ear (Fig 7), but unique to each ear (Fig 8). For the individuals in this study with measurable responses, the frequency region of maximal change in the reflectance measurements from reflexive middle ear muscle contraction was around or above 1 kHz.

Figure 7.

Similarities across difference curves obtained across multiple activator electrodes within an ear (C001L). Line style indicates electrode number (5, 12 and 19). The gray shaded region indicates the 2-octave frequency for the RMS calculation.

Figure 8.

Example reflex-induced changes in reflectance measurements across ears with responses (1 electrode, multiple activators in each panel). The gray shaded region indicates the 2-octave frequency for the RMS calculation.

Reflexes were observed in 7 of 9 ears for at least 1 electrode tested, and in 21 of 46 electrodes tested when using the reflectance-based measurements. Participants C009L and C008R had measurable single-frequency reflexes, but reflexes were not observed in the reflectance measurements. For C009L, poor probe fit (i.e. leak) affected all reflectance measurements, and is likely the reason why reflex-induced changes could not be observed in the recordings. For C008R, it is unclear why reflectance-based reflexes were not observed.

For 16 of the 24 total electrodes with measurable reflexes, the reflectance-based measurement showed an advantage (i.e. a lower threshold), though the advantage was a fraction of a dB in most instances. There were 4 instances where the advantage of reflectance methodology may be greater than indicated due to either an inability to test low or high enough activator levels to find threshold; compared to 3 instances where the 678 Hz advantage may be greater than indicated. Even so, the Wilcoxon signed-rank test was not significant (p = 0.072).

DISCUSSION

The present study expands upon recent work (primarily Wolfe et al. 2018, but also Wolfe et al. 2017 and Hernandez et al. 2018) by characterizing wideband reflectance and reflex-induced reflectance changes in a subset of ears following cochlear implantation and by further exploring the benefit of using a broadband probe for ipsilateral eSRT recordings.

Wideband Power Reflectance

It has been suggested that cochlear implant surgery increases middle ear stiffness (e.g. Wolfe et al., 2017; 2018), and the general recommendation when performing eSRs clinically has been the use of the contralateral, non-implanted ear as the probe ear (e.g. Hodges et al., 1999). However, it is unclear whether, how, and to what extent the conductive mechanism is affected by the CI surgery. This study evaluated the use of wideband power reflectance as a tool to characterize the mechano-acoustic system for recipients of cochlear implants.

The statistically significant greater low-frequency reflectance for ears with cochlear implants compared to the normative data set is consistent with a suspected increase in middle-ear stiffness; however, individual reflectance curves highlight the variability observed across ears with CI. One aspect of the study design was the use of relatively lax inclusion criteria for the CI recipients compared to the normative data set. That is, CI recipients were not excluded for abnormal 226 Hz tympanometry (low- or hyper-compliance). This was intentional to capture the variability (and describe the potential abnormality) in the population of interest, but it is uncertain whether there is a higher incidence of abnormal 226 Hz admittance in cochlear implant recipients than the general population since the inclusion criteria was stricter for the non-implanted ears for the normative data set. If so, the results would support a difference in the conductive mechanism in ears with cochlear implants, with most ears (but not all) showing patterns consistent with increased stiffness. Wideband eSRs from ears with normal 226 Hz admittance (discussed in a subsequent section) also suggests a greater stiffness component compared to previous reports of wideband acoustic reflexes observed in non-implanted ears (e.g. Feeney & Keefe, 1999).

One limitation of this study is the small number of ears with cochlear implants and the cross-subject design for comparing reflectance measurements. A goal at the outset was to include a within-subject component for individuals implanted unilaterally (the contralateral ear used as the non-implanted control); however, there was insufficient data due to the number of bilateral CI recipients and excessive cerumen in some of the non-implanted ears. Given that intra-subject variability is smaller than inter-subject variability (Abur et al. 2014), a within-subject design would be valuable. A within-ear, longitudinal design would be most preferable to assess potential changes in the system post implantation as the body heals and reacts to the foreign body (Abur et al. 2014).

Another limitation of this study was the minimal surgical information that could be obtained on participants. The majority of participants were long-term users, and hospital records were not directly accessible by the authors. There are many variables that could affect mechano-acoustic system (e.g. electrode insertion approach, lead stabilization, device characteristics, iatrogenic injury, time after surgery, etc.) and these should be explored in a larger data set to assist in understanding individual differences.

The general issues with wideband acoustic immittance measurements that limit clinical applicability in non-implanted ears also apply to ears with CI. One issue this study attempted to address is the need for an individualized estimate of characteristic impedance (Z₀) for a more accurate reflectance calculation. In the present study, ear canal area was estimated from earmold impressions; however, this is not an ideal clinical solution as there is added time and expense in most instances (except when a retention mold may be desired). For CI recipients, a standard component of the pre-operative assessment includes imaging. Investigations are underway to evaluate how ear canal size estimates from imaging, such as CT, compare with measurements from the physical impressions (Voss, Reference Note 3). While this may be a feasible solution for CI recipients, other methods will need to be explored when imaging is not available (e.g., acoustic estimates; Rasetshwane & Neely, 2011).

Electrically Evoked Stapedial Reflexes

Across the participants and electrodes with measurable reflexes, the following characteristics were observed:

1. The frequency pattern of reflex-induced reflectance changes was unique to each ear, but similar across activators (electrodes and levels) for any given ear.

2. The maximum reflex-induced reflectance change was observed at or above 1000 Hz, even in ears with normal 226 Hz admittance.

3. For the majority of conditions, reflex thresholds obtained with the broadband probe were lower than thresholds obtained with single-frequency (226 or 678 Hz) probes; however, the threshold differences were small and not statistically significant from thresholds obtained with the 678 Hz probe.

Activator Parameters •

In ears without CIs, acoustically evoked reflex-induced changes in wideband power reflectance are similar across activator frequencies for a given ear, and vary systematically as the activator sound pressure level is adjusted (Feeney & Keefe, 1999). Electrode location and current level can be considered electrical analogs of frequency and sound pressure level, respectively. Similar response patterns were observed across electrical activators: the frequency range of the reflex-induced change in reflectance was similar across electrode sites for a given ear, and adjustments to the activator level systematically affected the magnitude of the change (Figs 7 and 8). Thus, parametric adjustments to electrical activators result in similar patterns of reflex-induced reflectance changes as seen with adjustments to acoustic activators. Overall amplitude and growth rates of the reflectance change are variable across ears and were not explored in the present investigation. Further work is warranted to determine whether these response features can provide insight into how the auditory periphery/lower brainstem is responding to electrical stimulation.

One feature of the electrical activators used in the present study that deserves additional attention is the temporal envelope. For single-frequency eSRs, a bursting pulse train (500-ms on and off; the standard measurement) was used to assist with visual detection of time-locked deflections in admittance. However, the pulse train activator during reflectance measurements was 5 seconds in duration, with a flat-amplitude envelope to provide sufficient time for signal averaging. Acoustic reflex decay diminishes when the temporal envelope is modulated (e.g., Lutman & Martin, 1978; Cook et al. 1999), and it may be that the comparison between the broadband and single-frequency probe conditions was confounded by different amounts of neural adaptation for the gated compared to flat-envelope pulse trains, and this might be different across CI recipients.

Frequency Region Most Sensitive to Reflex-Induced Changes •

Acoustically evoked reflex-induced reflectance changes in non-implanted ears are predominately observed at relatively low frequencies (i.e. below 1000 Hz; e.g., Feeney & Keefe, 1999). In contrast, the electrically evoked reflex-induced changes observed in this study were greatest at 1000 Hz or higher for all individuals. One individual with measurable reflexes (C001L) had reduced Ya, so the observing reflex-induced effects at high frequencies was expected. However, the remaining individuals with measurable reflexes had clinically normal Ya. The observance of reflex-induced changes above 1000 Hz suggests increased middle-ear stiffness, which is consistent with the reflectance patterns themselves. Wolfe and colleagues (2018) report that in their CI recipients, reflex-induced reflectance changes were typically observed across 500 to 2000 Hz, and that changes occurred at higher frequencies in some instances. However, they also observed increased reflectance below 500 Hz in some instances (see their Figures 3 and 4C)

More work is needed to further describe the variability among the population of CI users and to understand what factors underlie the reflex-induced change patterns. A theoretical benefit of using a broadband probe for reflex measurements rather than a single-frequency probe is that it is more efficient. It is not necessary to know ahead of time the frequency region of maximal change to optimize the measurement.

Reflex Thresholds •

Acoustic reflex thresholds obtained with reflectance techniques have been observed 2 to 24 dB lower than thresholds obtained with the standard 226 Hz probe tone (reviewed by Schairer et al. 2013). Although reflectance techniques appear consistently advantageous over clinical methods, the size of the advantage is potentially affected by a number of factors, including the spectral content of the activator, probe location (ipsilateral or contralateral), criterion for threshold (e.g., visual detection or statistical methods), pressurization of the ear canal or recording at ambient pressure, and age. The application of wideband immittance techniques to measuring eSRs in ears with CIs has only recently been explored (Wolfe et al. 2018; Scheperle et al. Reference Note 1), but it is expected that the factors that impact acoustically evoked reflexes would be pertinent to consider for electrically evoked reflexes.

Because advantages of using higher-frequency probe tones have been demonstrated for measuring eSRTs (Wolfe et al. 2017; Hernandez et al. 2018), the present study focused on comparing the broadband eSRTs to those obtained with a 678 Hz probe rather than the 226 Hz condition. It was desirable to determine whether the wideband admittance techniques were worth the additional complexity of measurement and analysis. Although reflex thresholds were the lowest with the broadband probe for the majority of measurements, the statistical test was not significant. In contrast, Wolfe and colleagues (2018) found that reflectance techniques resulted in significantly lower thresholds on average compared to those obtained with a 226 Hz probe tone. Even so, the average difference was 1.2, 3.6 and 3.7 CL for apical, middle and basal electrode sites (individual data are not provided; all participants had devices manufactured by Cochlear Corporation). For context, a 5-CL step size is often used in clinical measurements.

In the present study, thresholds are provided in units of μA due to the inclusion of recipients of various devices and threshold differences are in dB (Table 4). Threshold differences for individual measurements were variable. For the individual with low Ya (C001), threshold differences greater than 3 dB (equivalent of 20 to 30 CL) were observed for some electrodes tested (lower thresholds with the broadband probe). For a few electrodes tested (C001), a response was observed at all activator levels during reflectance measurements, even after adding lower levels during a return appointment. However, there were also instances where the reflectance method was not advantageous. For 2 individuals (C008R, C009L), reflexes could be obtained with the clinical method but were not observed at the limits of the activator level with the broadband probe. Although more instances of lower eSRTs when using a broadband compared to single-frequency probe were observed in the present study, the statistical analysis was not significant when comparing broadband to 678 Hz probe conditions. A larger advantage is observed when comparing to 226 Hz measurements; however, 678 Hz was similarly advantageous.

There are several parameters that require further exploration to interpret the differences observed across Wolfe et al (2018) and the present investigation, and to generally consider the broadband eSR methodology. For example, Wolfe and colleagues used a contralateral probe placement, which also happened to be a non-implanted ear in many instances. The present study used an ipsilateral placement, which was expected to result in a larger advantage of reflectance methodology (Hernandez et al. 2018), but the observed advantage was not as robust as expected. Another difference between studies is the threshold criterion. Wolfe and colleagues (2018) used a reflectance change of 0.03 as their criterion. Smaller reflectance changes were considered responses in the present study, which used the RMS amplitude across a frequency range and a noise floor estimate from interleaved no-stimulus conditions to assist with threshold determination. Even though both studies have used objective measures to influence decision making, threshold determination was ultimately subjective. Statistical methods like those used to determine acoustically evoked reflex thresholds (reviewed in Schairer et al. 2013) should be adopted for eSRT measurements as well. Additional parameters such as ear canal pressurization (compared to ambient pressure), and type of activator (single-electrode: Wolfe et al. 2018; present investigation; swept electrodes: Hernandez et al. 2018) also deserve further attention.

Clinical Implications •

The low incidence of eSRs is one reason the measurement is often not prioritized during clinical appointments and is one of the factors motivating exploration of reflectance techniques. As a reminder, the inclusion criteria were intentionally more lax in the present study than Wolfe et al (2018) to assess the potential clinical impact of improved methodology on increasing the likelihood of measuring a reflex. In this study, although the reflectance methodology was beneficial or at least no worse than conventional methods in the majority of instances, there were a few examples where standard clinical methodology was more successful. This finding might reflect limitations with the reflectance set-up specific to the experiment. For instance, the experimental set up did not allow real-time adjustments to the number of averages per condition. Replicated files were combined and analyzed offline, so it was not always apparent to the tester that additional averages would be needed to reduce the noise in the recording. Improved methodology is expected to improve the potential benefit of reflectance-based reflexes. A challenge that exists even with clinical eSR measurements is the lack of triggering capabilities across stimulating and recording computers. This does not preclude measuring the stapedial reflex, but does make the procedure more awkward, and limits the ability to investigate other features of the reflex, such as onset latency and rise time.

CONCLUSIONS

The potential clinical applications of wideband acoustic immittance measures can be extended to ears with CIs. In this study, average reflectance patterns and reflex-induced changes in reflectance were consistent with increased middle ear stiffness compared to non-implanted controls, which has been suspected. However, the variability across participants illustrates the need to focus on the individual for clinical interpretation. Electrically evoked reflex thresholds measured with a broadband probe were similar to thresholds measured with a 678 Hz probe tone. More work is needed to determine whether methodology that incorporates a broadband probe can be further optimized.