The electrically evoked auditory brainstem response (eABR) is used to verify the function of the auditory system from the auditory nerve to the brainstem. eABR testing is recommended as a preimplant prognostic test before cochlear implant (CI) or auditory brainstem implant surgery in patients with suspected/or confirmed hypoplastic cochlear nerves (1). eABRs can also be measured postoperatively using electrodes from the implant array and the resulting eABR thresholds could help guide CI programming (2–4).
The eABR has conventionally been recorded using either an extracochlear transtympanic test electrode placed at the promontory of the inner ear (5–9) or, more recently, using in intracochlear test electrode surgically placed inside the scala tympani via a cortical mastoidectomy and lateral tympanotomy (Supplemental Digital Content 1, http://links.lww.com/MAO/A696). The intracochlear stimulation test electrode has been developed by MED-EL and is available as a custom-made device (the Intracochlear Test Array; ITA). The ITA is requested from MED-EL in Europe under Custom-Made Device Regulations, 93/42/EEC. It is 18 mm long and has an active length of 8.4 mm. Measuring an eABR is a complex task requiring specialized equipment and expertise, sedated or very still patients, and significant clinician time. The advent of the ITA introduces the possibility of increased test resolution which could improve the test accuracy and provide a prognostic indicator for the success of a cochlear implant in these cases.
The authors are aware of one peer-reviewed eABR study concerning the clinical efficacy of the MED-EL custom-made ITA (10). In this study, researchers used the ITA and a CI to elicit eABR responses in adults with radiologically normal auditory nerves. When they compared responses, they found no significant differences in amplitude and latency of the response. There are no peer-reviewed studies with within-participant comparison of the different test electrode types (extracochlear or ITA), within participants. It is possible that eABR characteristics could vary due to different stimulation methods. Therefore, the aims of the current study were to compare methods of eABR stimulation. Differences between the two techniques could be clinically important because increased test resolution could lead to increased accuracy and better prognostic value.
The eABR has three or four positive peaks, and shows an increase in response amplitude and a decrease in response latency as stimulation levels increase (11–15). The stimulus used to evoke the eABR is a biphasic current pulse that is generated by the software used to program the CI speech processor. The presence of a response depends on the relative amplitude of a series of positive and negative peaks in the evoked response waveform. Additionally, metrics such as numbers of waveforms and quality of waves (7) and waveform growth functions (9,16) may also hold some clinical value.
Perioperative eABR recordings have been conventionally elicited transtympanically via extracochlear stimulation using a needle electrode placed at the entrance to the round window of the cochlea or on the cochlea promontory (5,7,17–19). The shape of the electrode varies from a needle, to a needle topped with a ball (golf-club) electrode. As mentioned above, this perioperative test is recommended for patients with suspected auditory nerve aplasia or hypoplasia, in order to determine whether an auditory brainstem implant or a CI is the better auditory prosthesis to use (1). In this clinical application, it is mainly the presence or absence of a wave V response which is sought after.
There are no published peer-reviewed normative eABR data measured using the ITA, and limited data with the golf-club extracochlear electrode. Additionally, there are no peer-reviewed studies that compare electrode type (CI, extracochlear, or ITA), within participants, to determine if there is a difference in the eABR responses or the amount of electrical stimulation required to elicit responses when different electrodes are used.
The aims of the study were to: 1) record eABRs using three techniques (extracochlear from the ‘golf-club’ electrode and intracochlear using the ITA, and the CI) from each participant, and 2) compare the following characteristics of the eABR: (i) the amount of charge required to elicit a threshold response, (ii) latency of identifiable peaks in the waveform, and (iii) general morphology and growth of responses.
HYPOTHESIS 1: CHARGE REQUIRED TO ELICIT A THRESHOLD RESPONSE
The distance between the extracochlear electrode and the spiral ganglion is larger than for the intracochlear electrodes; therefore, the total charge in micro-Coulombs (μC) required to elicit a threshold eABR using the ITA and CI will be smaller than that required using the extracochlear Golf-club electrode. This can be of clinical significance as higher charges are likely to introduce greater stimulus artifact into a recording; having deleterious effects on the quality of the response.
HYPOTHESIS 2: LATENCY OF THE RESPONSE
The latency (ms) of wave V of threshold eABRs will not significantly vary between the three different electrode types because the stimulation rate is not being varied, and is the only known factor to may marginally increase the latency of an eABR as it increases.
HYPOTHESIS 3: MORPHOLOGY OF THE RESPONSE
eABR traces elicited via the ITA and CI will share more morphological characteristics (numbers of identifiable positive peaks and response growth patterns) than the extracochlear Golf-club-elicited responses. This is because the ITA and CI will be simulating similar parts of the cochlea. Furthermore, if higher amounts of charge are needed to elicit an extracochlear eABR, there may be differences in stimulus artifact affecting the eABRs.
MATERIALS AND METHODS
The procedures followed were in accordance with the ethical standards of the National Health Service (NHS) in the United Kingdom and with the Helsinki Declaration, and were approved by the NHS Research Ethics Service (reference: 17/NW/0151). All participants were of the age and capacity of consent, and gave their written informed consent before participating in the study.
Participants were new CI patients from the Institution, being fitted with a MED-EL Synchrony implant or a Cochlear Nucleus Profile implant (Cochlear Ltd., Sydney, Australia), the two most commonly used implants within the service. The decision regarding the type of device was not influenced by the study and involved the patient in shared decision-making with their clinical team. Inclusion criteria included an anatomically normal auditory nerve on the side of implantation, assessed via CT/MRI scan. Participants with significant residual hearing: hearing thresholds in more than one out of five main octave points on their audiogram better than 80 dB HL at 125 Hz and 90 dB HL at 250 to 8000 Hz were excluded from the study. This was to ensure that the additional insertion and removal of an intracochlear test electrode (for the purposes of the study) would not risk impacting on the preservation of residual hearing.
Of the 117 adults who underwent CI surgery at the Institution between April 2017 and April 2018, 47 were identified as potential candidates and approached for the study. From 47 patients, 30 potential candidates responded to the invitation to participate and consented to the study; however, due to other health limitations, surgery timings, custom-made ITA availability, and one instance of technical difficulties in theater; 16 participants (8M, 8F, mean age 50 years [SD = 15]) were included in the study. The patients received Cochlear CI522 arrays (n = 5) or MED-EL Flex 28 arrays (n = 10) as decided by the multidisciplinary team irrespective of the patient's participation in the study. Table 1 gives participant demographic details of age, gender, duration of deafness, etiology, implant, and side of implant.
eABR recordings were measured during cochlear implant surgery while participants were under general anesthesia. Single-use subdermal needle electrodes were placed in the following montage for recording: vertex (positive), forehead (ground), and contralateral mastoid (negative). Additionally, two electrodes were placed superior and inferior to the pinna; to act as references (negatives) for the extracochlear golf club electrode. A cortical mastoidectomy was performed as part of the CI surgery and when the round window niche was exposed, paralysis was induced using 25 mg of atracurium or rocuronium. This was to prevent myogenic interference from incidental facial nerve stimulation during electrical stimulation at the promontory. The recording paradigm was: 1) extracochlear eABR via the golf club electrode, 2) intracochlear eABR via the ITA, and 3) intracochlear eABR via the CI. All ITAs and CI arrays were inserted via the round window of the cochlea. The order of testing could not be balanced because the CI electrode was always the last array to be used to minimize insertions and removals of arrays into the cochlea. This was unlikely to have an effect as no active patient participation is required for an eABR and as the response is not known to fatigue. The study increased each surgery time by approximately 30 min.
Identification of the eABR was based on visual inspection of the waveform by two audiologists. Blind reassessment of the waveforms established an interrater reliability of 94% and intrarater reliability of 96%. Threshold was defined as the lowest current level at which a repeatable response could be seen. A repeatable response was defined as two wave V's with similar amplitude greater than 0.4 μV, and similar latency at the same stimulation level, with background noise not exceeding one-third of the size of the waveform; as per UK Newborn Hearing Screening Protocol (NHSP) ABR criteria (20). Response growth at higher current levels was also necessary to ensure the measured response was neural.
All eABR traces and raw data are included in Supplemental Digital Content 2 (http://links.lww.com/MAO/A697). GraphPad Prism (San Diego, CA), version 7.04, was used for statistical analyses and to generate Fig. 1. Microsoft Excel Professional Plus (2013), version 15.0, was used to generate Figs. 2 and 3. Before statistical group analysis, group data were analyzed for each variable collected (μC, ms). D’Agostino and Pearson normality tests were conducted for each group (Supplemental Digital Content 2, http://links.lww.com/MAO/A697). Charge data (μC) for the medial electrode CI group and for the basal electrode ITA group were not normally distributed (p < 0.05). Therefore, a Kruskal–Wallis H test was used to test the effect of stimulating electrode type on the charge (μC) required to elicit a threshold wave V response. A one-way repeated measures ANOVA with a Geisser-Greenhouse correction was used to test for an effect of stimulating electrode type on the latency (ms) of the peak of a threshold wave V response.
A post hoc power calculation for the repeated measures ANOVA, with within-between interactions was calculated using GPower 3.0.10 (Universität Düsseldorf, Germany). The test was found to have a power (1−β) of 83%; indicating adequate study power.
From the 16 participants recruited, a full dataset of all test conditions was collected in 11 patients. In one subject; eABR008, a technical error with the grounding electrode resulted in aborting data collection. For subject eABR002, the basal ITA electrode and medial CI electrode were missed due to procedural error. For subject eABR001, no medial CI eABR could be recorded, due to open circuits (Supplemental Digital Content 2, http://links.lww.com/MAO/A697). Open circuits are relatively common occurrences immediately after CI array insertion due to air bubbles around the array (21). The basal CI eABR and both basal and medial ITA eABRs for subject eABR012 could also not be measured due to open circuits. A basal eABR recording could not be measured for the same reason from subject eABR009. eABR011 and eABR015 had open circuits on certain CI electrodes but low impedances at neighboring electrodes, and therefore those were used to prevent the loss of further data points. Considering the variability in array insertion depth between patients, and the proximity of neighboring electrodes, this is unlikely to constitute a significant difference. A total of 15 participants contributed data to the study.
Primary Outcome Measures
A common unit of electrical charge per second (Coulombs, C) was used to harmonize data from Cochlear and MED-EL devices. Charge is the maximum amplitude of the current (Amps) in a pulse multiplied by its width (in seconds). The charge required to elicit a threshold wave V response for each of the stimulating electrode is shown in Fig. 3A. For intracochlear electrodes, charge was tightly clustered, and less charge was required to stimulate a response than was required when using the extracochlear electrode. The Kruskal–Wallis H test showed a significant effect of stimulating electrode type on the Charge (μC) required to elicit a threshold wave V response (H = 30.88, p < 0.0001). Posthoc Dunn's multiple comparisons were used to test for between-group differences. The Charge required to elicit a response using the extracochlear electrode (median = 0.08 μC) was significantly larger than all other electrodes: basal ITA electrodes (median = 0.01 μC, p < 0.0001), medial ITA electrodes (median = 0.02 μC, p < 0.005), basal CI electrodes (median = 0.02 μC, p < 0.02), and medial CI electrodes (median = 0.02 μC, p < 0.001). Charge required to elicit a threshold wave V response did not vary significantly between any of the intracochlear electrodes.
The latency of wave V responses elicited by each stimulating electrode is shown in Fig. 3B. Latency data for all electrodes was more variable than charge; however, an indication of responses from extracochlear electrodes having a longer wave V latency is still visible. The ANOVA showed a significant effect of stimulating electrode type on the latency (ms) of the peak of a threshold wave V response F (2.533, 25.33) = 19.42, p < 0.0001). Post hoc Sidak's multiple comparisons were used to identify differences between electrode groups. The latency (ms) of the peak of a threshold wave V response when using extracochlear electrodes (mean = 5.1 ms) was significantly larger than all other electrodes: basal ITA electrodes (mean = 4.5 ms, p < 0.005), medial ITA electrodes (mean = 4.7 ms, p < 0.01), basal CI electrodes (mean = 4.6 ms, p < 0.01), and medial CI electrodes (mean = 4.4 ms, p < 0.0001). The latency of threshold wave V responses did not vary significantly between any of the intracochlear electrodes.
Secondary Outcome Measures
eABR response growth for each of the stimulating electrodes was analyzed. Response growth is calculated by plotting the size of wave V (μV), from threshold stimulation, as it grows with increased electrical stimulation (μC). When averaged across all study participants, an average eABR input/output (I/O) function, is produced. An eABR I/O function was created for each stimulation method; extracochlear, ITA, and CI (cf.Fig. 1). This was done by plotting the growth curve for each patient and calculating the regression line. Regressions in all patients were then averaged, and standard deviations were used in these discrete models.
The extracochlear eABR average I/O function has a threshold wave V amplitude of 0.22 μV (SD = 0.14) at threshold; which increases by 0.03 μV for every 0.01 μC of additional stimulation. The ITA average eABR I/O function has a threshold wave V amplitude of 0.33 μV (SD = 0.12) at threshold, which increases by 0.05 μV for every 0.0015 μC of additional stimulation. The CI average eABR I/O function has a threshold wave V amplitude of 0.26 μV (SD = 0.13) at threshold; which increases by 0.04 μV for every 0.0015 μC of additional stimulation. As indicated by the values, and can be seen from the figure (cf.Fig. 1), the three I/O functions are very similar.
The extracochlear electrode eABR I/O function is slightly wider than that of the ITA and CI, particularly at higher stimulation levels. This indicates marginally higher variability in extracochlear electrode measurements. ITA and CI I/O functions appear almost identical. The main difference between I/O functions is the charge required to elicit responses by the extracochlear electrodes; which has similarly been shown by the Kruskal–Wallis H test. When looking at the mean growth curve across the three stimulation methods, the gradients are fairly similar.
Qualitative Analysis of Suprathreshold Wave Characteristics
eABRs were collected at two stimulation steps (40 CL using Cochlear Custom Sound, Cochlear Ltd, Sydney, Australia) above threshold; in order to demonstrate waveform growth. These suprathreshold responses were qualitatively rated by two audiologists independently. Blind reassessment of the traces established an interrater reliability of 86% and intrarater reliability of 92%. Waveforms in each response were counted, and responses were grouped into one of three groups: “Clear”: No artifact pervading beyond 1 ms, less than 10 nV noise, flat base, immediately identifiable wave V. “Adequate”: No artifact pervading beyond 2 ms, between 10 and 25 nV noise. “Poor”: Artifact pervading beyond 2 ms, more than 25 nV of noise (cf.Fig. 2 A).
The analysis established that while CI most often produced the best-quality eABRs, the extracochlear electrode was the only electrode not to produce any poor responses; with all responses being classed as clear or adequate. Intracochlear electrodes most commonly produced a response with three peaks in the waveforms, whereas the extracochlear electrode most commonly produced an eABR with two peaks in the waveforms (cf.Table 2).
This study compared eABR data collected using different stimulation methods from patients with normal auditory nerves. The extracochlear electrode technique requires a tympanotomy for the electrode to be placed on the promontory of round window niche, whereas the ITA requires a cortical mastoidectomy and posterior tympanotomy for the array to be inserted into the cochlea. The radical differences between the ITA and extracochlear stimulation gave rise to the need to clinically investigate if one approach held any advantages. There are no published peer-reviewed normative eABR data measured using the ITA, and limited data with the golf-club extracochlear electrode.
Primary Outcome Measures
Hypothesis 1, which predicted a significant difference between extracochlear and intracochlear electrodes in the charge required to elicit a threshold wave V response, is supported by the results of this study. The charge required by the extracochlear electrode was significantly larger than with ITA and CI electrodes, and no differences were found between the ITA and CI groups. The difference can be explained by the larger amount of charge required to breach the round window and reach the spiral ganglion cells within the cochlea modiolus when using an extracochlear stimulating electrode. The charge required when using the extracochlear electrode was approximately six times larger than intracochlear methods. The effect the difference in charge had on response quality is discussed in secondary outcome measures.
Hypothesis 2, which predicted no significant difference in latency between electrodes, is not supported by the results of this study. The latency of wave V responses elicited by the extracochlear electrode was significantly larger than with the ITA and CI, and no differences were found between ITA and CI. The mean latency of the golf-club eABR group was 5.1 ms, which is notably longer than the 4.4 ms reported by Kileny and Zwolan (7) who also used a transtympanic golf-club electrode. Arguably, Kileny and Zwolan (7) measured latency at suprathreshold stimulation. Some studies have described a correlation between latency and amplitude of stimulation (5); however, the majority of studies do not report latency change with increased stimulation (11,22,23). Kileny and Zwolan (7) used a pediatric population with atypical auditory nerves and it remains unclear why normal adults would have a longer extracochlear eABR latency. The mean duration of deafness in the adult patients was approximately 13 years (SD = 12), which is different from pediatric populations. With Kileny and Zwolan (7), different etiology groups had different latencies. Participants who lost their hearing following meningitis (n = 8) demonstrated the longest mean latency of 4.9 ms (SD = 0.57), though they did not find a statistically significant difference between etiology groups. Therefore, it is difficult for us to conclude that this may be due to the mixed etiologies in the current study. The extracochlear wave V latency measured in this study matches the mean latency measured by Mason et al., of 5.2 ms5, which was also measured transtympanically at suprathreshold stimulation; however, this was measured at a much higher stimulation rate of 86 pps.
All ITA and CI wave V latencies fell into the range of 4.1 to 5.1 ms, which is similar to the 4.1 to 4.7-ms range cited by Hughes (15) for stimulation levels closer to threshold. The significant difference in latency of the wave V waveform at threshold is hypothesized to be due to the time it takes for enough charge to build and spread from the round window membrane and reach the spiral ganglion. The latency of responses elicited with an extracochlear electrode was approximately 0.5 ms longer than other methods.
Secondary Outcome Measures
Hypothesis 3 concerned morphology of the eABR traces elicited. High intra and interrater reliabilities confirm that a simple subjective grading system of “clear, adequate, or poor”. Both Figures 2A and B support the hypothesis that the ITA and CI share more morphological characteristics than the extracochlear Golf-club elicited responses. However, the difference between extracochlear and intracochlear electrodes is not great. The results support the hypothesis that intracochlear stimulation gives slightly more detail due to cleaner waveforms with less electrical stimulation artifact. Notably, the stimulation (charge) level that has been used for suprathreshold responses is approximately 40 CL greater than threshold stimulation. Stimulating at this level produced a response with an amplitude greater than that at threshold, but even more charge could have possible evoked a larger amplitude response. The difference between intracochlear and extracochlear electrodes may become more pronounced at higher levels; however, such an assertion would require further testing.
The eABR response displays an amplitude growth response with increasing stimulus. The wave V I/O function of the eABR is thought to be linked to spiral ganglion survival. In an animal study where ganglion cells were surgically ablated, researchers found a much flatter response curve in cats with only 5 to10% of spiral ganglion cell survival (24). This suggests that high loss of spiral ganglion cells could produce a different I/O function than healthier inner ears; which could be an indicator of postoperative CI performance.
Wave V amplitude I/O appears to be influenced by the stimulating device used (13). Abbas and Brown (13) found that I/O functions varied between Nucleus and Symbion devices and attributed it to the difference in artifact which may be generated with different devices. The results in the present study show similar I/O functions between the three stimulation techniques with somewhat more variability as stimulation grows in the extracochlear electrode I/O function. As with the morphological analysis, the top end of the eABR I/O functions are not at very high stimulation levels and so there is a possibility that I/O functions may be less similar at higher stimulation currents. Also, stimulation across multiple electrodes on an array may provide further variations which a single extracochlear electrode would not be capable of.
Intracochlear stimulation, particularly with the ITA, was prone to poor or disrupted recordings owing to electrodes with high impedances or open circuits. This occurred in 33% of the recordings from the ITA and in two cases, data collection was not possible. The extracochlear electrode, in contrast, did not produce any poor recordings and enabled data collection in all 15 patients. When impedances were low, intracochlear stimulation with both the ITA and CI was more likely to produce clear recordings with more waveforms than the extracochlear electrode.
This study has not tested if an additional waveform in an eABR, as a result of using the ITA, will improve predictive value of the test. That would require a larger participant cohort to produce the substantial number of responses with the three to five waves required for detailed analysis against outcomes. Such an analysis may also benefit from including participants with atypical auditory nerves as well. What this study has determined is that ITA eABR results are very similar to CI eABR results, so it is reasonable to assume that CI stimulation would give eABRs representative of intracochlear stimulation with the ITA, as long as stimulation depth in the cochlea approximately respected. The predictive power of ITA eABR is probably similar of that of CI eABR; for which there is somewhat of a body of knowledge. Such studies tend to produce correlations with relatively small effect sizes with parameters associated with listening outcomes, such as T-levels, but rarely produce correlations with large effect sizes directly with listening outcomes. Perioperative CI eABR, and therefore probably also ITA eABR, is not the preferred preimplant test if transtympanic eABR is available.
Extracochlear eABR is a less invasive and less expensive option that always produced responses in patients with radiologically normal nerves. The ITA did not always produce responses, but produced marginally finer detail in responses for some participants. The clinical significance of such detail is currently unknown. When attempting to clinically determine the presence of an auditory brainstem response via electrical stimulation of the inner ear, extracochlear transtympanic stimulation may well be sufficient.
This study was supported by the NIHR Manchester Biomedical Research Centre.
1. Sennaroğlu L, Colletti V, Lenarz T, et al. Consensus statement: long-term results of ABI in children with complex inner ear malformations and decision making between CI and ABI. Cochlear Implants Int
2. Brown C, Abbas P, Fryauf-Bertschy H, Kelsay D, Gantz B. Intraoperative and postoperative electrically evoked auditory brain stem responses in nucleus cochlear implant users: implications for the fitting process. Ear Hear
3. Lo T, Chen Y, Horng M, Hsu C. Efficacy of EABR
and ECAP in programming children with Nucleus-24 cochlear implants. Cochlear Implants Int
2004; 5 (Supplement 1.47):47–49.
4. Shallop J, VanDyke L, Goin D, Mischke R. Prediction of behavioral threshold and comfort values for nucleus 22-channel implant patients from electrical auditory brain stem response test results. Ann Otol Rhinol Laryngol
5. Mason S, O’Donoghue G, Gibbin K, Garnham C, Jowett C. Perioperative electrical auditory brain stem response in candidates for pediatric cochlear implantation. Am J Otol
6. O’Driscoll M, El-Deredy W, Ramsden R. Brain stem responses evoked by stimulation of the mature cochlear nucleus with an auditory brain stem implant. Ear Hear
7. Kileny P, Zwolan T. Perioperative, transtympanic electric ABR in paediatric cochlear implant candidates. Cochlear Implants Int
2004; 5 (Suppl. 1):23–25.
8. Sauvaget E, Péréon Y, Nguyen The Tich S, Bordure T. Electrically evoked auditory potentials: comparison between transtympanic promontory and round-window stimulations. Neurophysiol Clin
9. Walton J, Gibson W, Sanli H, Prelog K. Predicting cochlear implant outcomes in children with auditory neuropathy. Otol Neurotol
10. Lassaletta L, Polak M, Huesers J, et al. Usefulness of electrical auditory brainstem responses to assess the functionality of the cochlear nerve using an intracochlear test electrode. Otol Neurotol
11. Chouard C, Meyer B, Donadieu F. Auditory brainstem potentials in man evoked by electrical stimulation of the round window. Acta Otolaryngol
12. Starr A, Brackmann D. Brain stem potentials evoked by electrical stimulation of the cochlea in human subjects. Ann Otol Rhinol Laryngol
13. Abbas P, Brown C. Electrically evoked auditory brainstem response
: growth of response with current level. Hear Res
14. O’Driscoll M, El-Deredy W, Atas A, Sennaroglu G, Sennaroglu L, Ramsden R. Brain stem responses evoked by stimulation with an auditory brain stem implant in children with cochlear nerve aplasia or hypoplasia
. Ear Hear
15. Hughes M. Zwolan T, Wolfe J. Part III physiological objective measures. Objective measures in cochlear implants
. San Diego, CA: Plural Publishing Inc; 2013. 93–149.
16. Alfelasi M, Piron J, Mathiolon C, et al. The transtympanic promontory stimulation test in patients with auditory deprivation: correlations with electrical dynamics of cochlear implant and speech perception. Eur Arch Oto-Rhino-Laryngology
17. Pau H, Gibson W, Sanli H. Trans-tympanic electric auditory brainstem response (TT-EABR
): the importance of the positioning of the stimulating electrode. Cochlear Implants Int
18. Chouard C, Koca E, Meyer B, Jacquier I. Test of electrical stimulation of the round window. Diagnostic and prognostic value of the rehabilitation of total deafness by cochlear implant. Ann d’oto-laryngologie Chir cervico faciale
19. Meyer B, Drira M, Gegu D, Chouard C. Results of the round window electrical stimulation in 460 cases of total deafness. Acta Otolaryngol Suppl
21. Carlson ML, Archibald DJ, Dabade TS, et al. Prevalence and timing of individual cochlear implant electrode failures. Otol Neurotol
22. van den Honert C, Stypulkowski P. Characterization of the electrically evoked auditory brainstem response
(ABR) in cats and humans. Hear Res
23. Abbas P, Brown C. Electrically evoked brainstem potentials in cochlear implant patients with multi-electrode stimulation. Hear Res
24. Smith L, Simmons F. Estimating eighth nerve survival by electrical stimulation. Ann Otol Rhinol Laryngol