Intracochlear Electrocochleography and Speech Perception Scores in Cochlear Implant Recipients

Abstract

Objective:

Previous studies have demonstrated that electrocochleography (ECochG) measurements made at the round window prior to cochlear implant (CI) electrode insertion can account for 47% of the variability in 6-month speech perception scores. Recent advances have made it possible to use the apical CI electrode to record intracochlear responses to acoustic stimuli. Study objectives were to determine (1) the relationship between intracochlear ECochG response amplitudes and 6-month speech perception scores and (2) to determine the relationship between behavioral auditory thresholds and ECochG threshold estimates. The hypothesis was that intracochlear ECochG response amplitudes made immediately after electrode insertion would be larger than historical controls (at the extracochlear site) and explain more variability in speech perception scores.

Study Design:

Prospective case series.

Methods:

Twenty-two adult CI recipients with varying degrees of low-frequency hearing had intracochlear ECochG measurements made immediately after CI electrode insertion using 110 dB SPL tone bursts. Tone bursts were centered at five octave-spaced frequencies between 125 Hz - 2000 Hz.

Results:

There was no association between intracochlear ECochG response amplitudes and speech perception scores. But, the data suggests a mild to moderate relationship between preoperative behavioral audiometric testing and intraoperative ECochG threshold estimates.

Conclusion:

Performing intracochlear ECochG is highly feasible and results in larger response amplitudes, but performing ECochG before, rather than after, CI insertion may provide a more accurate assessment of a patient’s speech perception potential.

Introduction

It is well known that cochlear implant (CI) recipients’ performance on speech perception measures vary widely. Multiple biographic and audiologic factors influence speech perception,¹ yet these explain less than 22% of the variability in speech perception scores in quiet.² This makes it difficult to counsel patients on the potential benefits of cochlear implantation.³ Given this limitation, researchers have turned towards assessing cochlear function via electrocochleography (ECochG) to help understand how it contributes to the variability in performance. ECochG is a technique used to make electrophysiologic measurements from the inner ear, believed to originate primarily from hair cells and nerve fibers, by using a recording electrode placed at extracochlear (i.e. round window) or intracochlear sites.⁴ Fitzpatrick et al was the first to correlate ECochG response amplitudes—recorded at the round window immediately prior to CI insertion—with 6-month speech perception scores and found that ECochG response amplitudes explained 47% of the variance in CNC word scores.⁵ Subsequent studies from the group^3,6,7 have shown similar results. The round-window approach however results in smaller response amplitudes compared to intracochlear responses⁸ and is limited to being made in the operating room.

Recent advances have made it possible to use the most apical contact of a CI array as a recording electrode for responses to acoustic stimuli.^9,10 This creates an opportunity to make ECochG measurements at multiple time points during CI electrode insertion and after. In CI candidates with low-frequency hearing, using the most apical contact generates a greater likelihood of capturing the contribution of the remaining sensory cells, potentially yielding larger responses.

The majority of studies using intracochlear ECochG have focused on its utility as a surgical monitoring tool to prevent translocation of the electrode array^11,12 to preserve hearing,^10,13-17 as an estimate of hearing thresholds,^16,18-21 and as an adjunct for acoustic tuning.²² A key unknown is whether intracochlear ECochG measurements made immediately after CI insertion explain more variance in speech perception scores compared to extracochlear ECochG.

Methods

Design

A total of 37 CI candidates enrolled in this prospective study between September 6, 2016 and April 18, 2019. Eligible patients were adults (≥ 18 years old) with low-frequency hearing who chose to be implanted with an Advanced Bionics (Advanced Bionics Corporation, Valencia, CA) CI at a tertiary institution. Low-frequency hearing was defined as any acoustic hearing from 125 – 2000 Hz. Patients with pre-existing middle-ear conditions or non-native English speakers were excluded. This study was approved by the [author’s institution] Institutional Review Board and was in accordance with ethical standards.²³

The primary objective was to assess the relationship between intracochlear ECochG response amplitudes, made immediately after electrode insertion, with 6-month speech perception scores. The secondary objectives were to assess how the intracochlear ECochG response amplitudes changed over time and to determine the relationship between behavioral auditory thresholds with ECochG threshold estimates at the intraoperative and postoperative time points. The hypothesis was that the intracochlear ECochG response amplitudes would be larger than historical controls (at the extracochlear site), and thus be able to explain more variability in speech perception scores.

Variables

Baseline demographics and behavioral auditory metrics of the recipient ear were obtained. Preoperative (aided) and postoperative, electric-only speech perception scores (3, 6, 12, and beyond 12 months) in the implant ear consisted of CNC word scores and AZBio sentence scores in quiet, both presented at 60 dB SPL. The contralateral ear was plugged and muffed during testing. Patients with residual hearing were tested in the electric-only condition. Unaided audiograms of the implanted ear were obtained at the 1-month postoperative appointment and at various time points thereafter. Participants underwent a pre-and post-operative computed tomography (CT) temporal bone scan to determine scalar location of the CI array as previously described^24,25 since translocation is associated with worse performance.¹ Scalar locations included all in scala tympani (ST), intermediate (ST and scala media), or in ST and scala vestibuli (SV).

ECochG measurements

Intracochlear ECochG measurements were made using the apical electrode of the CI. The intraoperative setup consisted of placing an in-ear insert connected to an Etymotic ER-3 speaker. The speaker was connected to custom equipment that provided the acoustic tone bursts. A standard transmastoid facial recess approach was performed. Once the internal receiver was placed, an external transmitter was placed in a sterile bag and positioned over it. During CI insertion, single-frequency ECochG monitoring was evoked using 110 dB SPL 50-ms tone bursts (alternating in polarity), centered at the 500 Hz, to understand insertion trauma. These yielded the “peak response.” Immediately after CI insertion, ECochG measurements were similarly evoked but centered on five octave-spaced frequencies: 125, 250, 500, 1000 and 2000 Hz with 40-80 averages to yield the “total response.” ECochG measurements were evoked postoperatively using up to 1000 averages in the same manner.

The sum and difference of responses from alternating stimuli were calculated. The cellular and spatial origins of these are not fully understood, though the sum may represent the predominant neural contribution to the response and the difference may represent the cochlear microphonic (CM). Fast Fourier transformation (FFT) using custom MATLAB software (The MathWorks, Inc., Natick, MA) was used to obtain the amplitude of the first and second harmonic of the response for each frequency. Significant responses had to exceed the mean noise floor (calculated from three spectral bins on each side of the target frequency starting two bins away from the peak) for both the summation and difference of the first and second harmonics at each frequency (Figure 1). Only significant amplitudes of the target frequency and their harmonics were summed and then converted to dB relative to 1 uV (dB re: 1uV) to generate an ECochG “total response” (TR).⁵ The ECochG recording software used difference response as previously described^20,21 to yield ECochG estimated audiometric thresholds.

Figure 1.

An example of an intracochlear ECochG recording immediately following electrode insertion. Left panels (A,C,E,G and I) show Difference and Summation responses in the time domain; right panels (B,D,F,H and J) show Difference and Summation responses in the frequency domain for frequencies 125 Hz (A and B), 250 Hz (C and D), 500 Hz (E and F), 1000 Hz (G and H) and 2000 Hz (I and J). Significant peaks above the noise floor are marked with a closed Circle symbol. The sum of all significant peaks contributed to the Total Response. In this example, seven significant points contributed to the Total Response.

Statistical Analysis

Demographic and clinical characteristics were run in SPSS Statistics for Windows version 26 (IBM Corp., Armonk, NY) and summarized using descriptive statistics. Distribution of the baseline continuous level variables were not normally distributed, so data was reported using the median [minimum – maximum]. A clustered robust regression approach was used to determine the association between intraoperative ECochG TR and change in speech perception scores, and the association between intraoperative ECochG TR and 6-month speech perception scores. Spearman’s correlation assessed the relationship between the behavioral audiometric thresholds and ECochG threshold estimates. Post-hoc analyses used Spearman’s correlation to evaluate the relationship between intraoperative ECochG TR and 6-month speech perception scores in patients ≤ 80 years of age to compare results with a recent round window ECochG study.³ Clustered linear regressions were used to evaluate the relationship between 6-month speech perception scores with the “peak responses” or 6-month postoperative ECochG TR. This was a pilot study so the sample size was determined by feasibility as no preliminary data exists on intraoperative ECochG responses using the apical electrode.

Results

A total of 37 patients were enrolled. Fifteen participants were excluded for the following reasons: withdrew from study due to time constraints (n=1); ECochG equipment malfunction or user error (n=6); and missing 6-month speech score (n=8). No adverse events were reported. A total of 22 participants had data available for analysis.

Baseline demographics are summarized in Table 1 and outcome measures are summarized in Table 2. Electrode type was not an effect modifier based on distribution of TR or performance over time, so results were combined. Translocation into scala vestibuli was minimal (Table 2) and was not an effect modifier. At the end of the study period, five participants were identified as having potential device failure based on a manufacturer recall, but they were not effect modifiers in any of the analyses and were kept in the analysis. Two separate participants, however, were clinical outliers (n=2) and effect modifiers due to clinically meaningful low performance (6-month CNC score < 5%). Thus, the statistical analysis was reported without those two outliers.

Table 1.

Baseline Patient Demographics (n=22)

Demographic	N (%)
Sex:
Male	13 (59.1%)
Female	9 (40.9%)
Recipient site
Left	11 (50%)
Right	11 (50%)
Cochlear implant electrode:
Mid-Scala	16 (72.7%)
Slim J	6 (27.3%)
Etiology:
Unknown	14 (63.6%)
Meniere’s disease	4 (18.2%)
Meningitis	1 (4.5%)
Noise induced Genetic	1 (4.5%) 2 (9.1%)
Progression of hearing loss:
Progressive/gradual	18 (81.8%)
Sudden	4 (18.2%)
Demographic	Median (range)
Age at onset of hearing loss (years)	47 (5 – 84)
Duration of hearing aid use (years)	13 (0 – 28)
Duration of severe/profound hearing loss (years, n=10)	3 (0.25 – 17)
Age at implantation (years)	65 (20 – 87)
Pure-tone average (125 Hz – 2000 Hz)	81 dB HL (50 – 110)
Duration between date of preoperative audiogram and intraoperative ECochG measurements (months)	5 (0.1 – 12)

Table 2.

Outcome Measures

Variable	Median (range)
Consonant nucleus consonant word score (CNC):
Baseline (n=20)	5% (0 – 22%)
3 months (n=21)	52% (3 – 84%)
6 months (n=22)	53% (2 – 90%)
12 months (n=16)	55% (20 – 85%)
Greater than 12 months (n=9)	74% (22 – 88%)
AZBio sentence score in quiet:
Baseline (n=16)	2% (0 – 50%)
3 months (n=20)	62% (0 – 99%)
6 months (n=20)	71% (3 – 99%)
12 months (n=16)	64% (12 – 91%)
Greater than 12 months (n=9)	77% (29 – 92%)
Scalar location of electrode (n=20):
All scala tympani	11 (50%)
Scala tympani/scala media	7 (31.8%)
Scala tympani and scala vestibuli	2 (9.1%)
ECochG Total Response:
Intraoperative (n=22)	14.97 dB re: 1uV (−13.98 – 51.16)
1 month (n=13)	2.88 dB re: 1uV (−13.98 – 23.56)
3 months (n=12)	7.41 dB re: 1uV (−13.98 – 26.05)
6 months (n=17)	6.39 dB re: 1uV (−13.98 – 44.57)
12 months (n=12)	1.58 dB re: 1uV (−13.98 – 23.58)
Greater than 12 months (n=6)	1.49 dB re: 1uV (−13.98 – 23.86)
Residual hearing at 1 month (n=20) †	7 (31.8%)
Residual hearing pure-tone average (125 Hz – 2000 Hz)	87 dB HL (63 – 106)
Clinical outlier (n=22) †	2 (9.1%)

All pre-operative speech testing scores reported above reflect the pre-implant ear in the best aided condition. Speech scores were obtained in aided cochlear implant only condition with contralateral ear plugged and muffed.

^†Numbers (percentage) are used for reporting distribution of categorical-level characteristics.

Relationship of Intraoperative and Postoperative ECochG TR with Speech Perception Scores

A clustered linear regression was performed to determine the relationship between intraoperative ECochG TR with the change in speech perception scores from baseline to beyond 12-months (Figure 2). Time explained 60% of the variability in the change of CNC scores (r²= 0.60, 95% CI 3.82 – 10.29) and 57% of the variability in the change of AZBio scores (r²= 0.57, 95% CI 0.99 – 17.01). Adding intraoperative ECochG TR improved the model by 3% for both CNC scores (r²= 0.63, 95% CI −4.37 – 10.78) and AZBio scores (r²= 0.60, 95% −3.13 – 13.65).

Figure 2.

Line graph demonstrating the change in CNC word scores (left) and AZBio sentences in quiet (right) over time, stratified by the magnitude of intraoperative ECochG TR (upper, middle, and lower third values). Clinical outliers are represented by an “X” and were excluded from the statistical analysis. Time explained most of the variance in speech perception scores (r² = 0.60, 95% CI 3.82 – 10.29 for CNC scores and r² = 0.57, 95% CI 0.99 – 17.01 for AZBio scores). The addition of ECochG TR accounted for an additional 3% of the variance.

A clustered linear regression was run to determine whether the intraoperative ECochG TR was associated with 6-month speech scores (Figure 3). There was no relationship between the intraoperative ECochG TR and 6-month speech perception scores, with ECochG TR explaining 5.2% of the variability in CNC scores (r²=0.052, 95% CI −0.27 – 0.69) and 4.0% of the variability in AZBio scores (r²=0.040, 95% CI −0.33 – 0.75). A post-hoc analysis evaluating the relationship between the intraoperative ECochG TR in adults ≤ 80 years with 6-month speech scores suggested a weak, positive trend between the two variables (Figure 3, CNC: Spearman’s rho = 0.33, p-value=0.19, n=17; AZBio in quiet: Spearman’s rho=0.31, p-value =0.25, n=16). A post-hoc analysis evaluating the relationship between the “peak response” with 6-month speech perception scores demonstrated that the “peak response” explained 5.7% of the variability in 6-month CNC scores (r²= 0.057, y= 49.64 +0.31*x, 95% CI −0.41 – 1.02) and 4.3% of the variability in AZBio scores (r²= 0.043, y= 63.20 + 0.31*x, 95% CI −0.36 – 0.98).

Figure 3.

Scatterplot demonstrating the relationship between intraoperative ECochG TR and 6-month CNC word scores (left) and AZBio sentences in quiet (right). Patients > 80 years of age are represented by open black circles. Patients at risk for device failure are represented by gray circles. Clinical outliers are represented by an “X” and were excluded from the statistical analysis. Intraoperative ECochG TR explains 5.2% of the variability in 6-month CNC word scores (r²=0.052, 95% CI −0.27 – 0.69) and 4.0% of the variability in 6-month AZBio in quiet scores (r²=0.040, 95% CI −0.33 – 0.75).

The median ECochG TR slightly decreased over time when all cases were combined (Figure 4). The range of the ECochG TR was wider at the intraoperative time point (median 14.97 dB re: 1uV [−13.98 – 51.16]) than beyond 12-months (median 1.49 dB re: 1uV [−13.98 – 23.86]). A clustered linear regression was run to determine how time was associated with change in ECochG TR. Time explained 11% of the variability in the ECochG TR (r²=0.11, 95% CI 5.48 – 23.33). Figure 5 illustrates the change in the ECochG TR over time stratified by whether patients had residual hearing at 1-month. Patients with residual hearing (n=7) had a trend of higher median ECochG TR’s that appeared stable. A post-hoc analysis evaluating the relationship between postoperative ECochG TR at 6 months with concurrent speech perception demonstrated no meaningful relationship, as postoperative TR explained 5% of the CNC scores and 0% of AZBio scores (CNC scores: r²=0.049, 95% CI −0.95 – 0.40; AZBio: r²=0.00, 95% CI −0.73 – 0.67).

Figure 4.

Box-whisker plots demonstrating the change in ECochG TR over time in all patients from immediately after CI insertion to beyond 12 months. The star symbols represent outliers, whose value is greater or lower than 1.5 times the interquartile range.

Figure 5.

Box-whisker plots demonstrating the change in ECochG TR over time in patients without residual hearing at 1-month (blue plots, n=13) and with residual hearing at 1-month (red plots, n=7). The star symbols represent extreme cases whose value is lower than 3 times the interquartile range. For those without residual hearing at 1 month, the number of cases with postoperative TR per time point were as follows: 1 month (n= 8), 3 months (n= 6), 6 months (n=10), 12 months (n=9), and greater than 12 months (n=3). For those with residual hearing at 1 month, the number of cases with postoperative TR per time point were as follows: 1 month (n= 5), 3 months (n= 5), 6 months (n=6), 12 months (n=3), and greater than 12 months (n=2).

Relationship between behavioral audiogram and ECochG threshold estimates

The relationship between the intraoperative ECochG threshold estimates and the preoperative behavioral audiogram is shown in Figure 6. Most patients had higher ECochG threshold estimates compared to their preoperative audiogram from 250 – 1000 Hz. There were no obvious outliers. There was a positive trend in the data with Spearman’s rho ranging between 0.18-0.46 for 250-1000 Hz. At 500 Hz, though not significant, there was a moderate, positive trend between ECochG threshold estimates and the behavioral audiogram (Spearman’s rho = 0.46, p-value =0.13). The relationship between the postoperative ECochG threshold estimates and the behavioral audiogram could not be analyzed because there were less than 5 cases per time point and frequency.

Figure 6.

Scatterplots demonstrating the relationship between intraoperative ECochG predicted hearing thresholds and preoperative behavioral hearing thresholds (n=22) at 250 Hz to 2000 Hz. Intraoperative ECochG thresholds were estimated by using a 512-point FFT with a frequency resolution of 18 Hz. A Hanning window was used to analyze the CM responses and estimate the CM amplitude and noise floor. The CM amplitude is based on FFT amplitude at the stimulus frequency bin and the noise floor is estimated as the average of bins (−6,−5,−4,+4,+5, +6 where “0” bin corresponds to the stimulus frequency bin). An adaptive averaging between 8 to 40 repetitions was used to measure ECochG responses per stimulus frequency. An ECochG waveform with a signal-to-noise ratio of at least 24 dB was used for switching to the next stimulus frequency. The signal was treated as “No Response” if an ECochG signal could not be measured 6 dB above the noise floor after 40 repetitions. The CM amplitude measured at single stimulus presentation level was used to estimate CM thresholds using the following formula: Estimated CM (ECochG) threshold (dB HL) = Acoustic stimulus level (dB HL) – 20 * log10 (CM amplitude in μV/0.25 μV).” The scatterplot demonstrating hearing thresholds at 125 Hz was unable to be created as less than 5 participants had measurable thresholds at this frequency. At 250 Hz, Spearman’s rho = 0.175 (p=0.57), at 500 Hz, Spearman’s rho = 0.462 (p=0.13), at 1000 Hz, Spearman’s rho= 0.2 (p=0.51), and at 2000 Hz, Spearman’s rho = 0.06 (p=0.91).

Discussion

This is the first study to demonstrate the feasibility of reporting the relationship between intracochlear ECochG responses recorded immediately after CI insertion and speech perception scores in adult CI recipients using the most apical electrode of the CI. This is also the first study to report the relationship between preoperative behavioral audiometric thresholds and intraoperative ECochG threshold estimates. We found that intracochlear ECochG response amplitudes were not associated with 6-month speech perception scores or with the change in speech perception scores in the postoperative period. Our data suggests a mild to moderate trend between preoperative behavioral audiometric thresholds and intraoperative ECochG thresholds.

Does location and timing of the recording electrode matter for intraoperative ECochG?

Multiple studies recording round window ECochG responses prior to CI insertion have consistently demonstrated that it can explain around 47% of the variability in 6-month speech perception scores.^3,5-7 But our study did not find this relationship between these two variables. Key differences were the location of the recording electrode (intracochlear versus extracochlear) and the timing of the measurements (immediately after CI electrode insertion versus immediately prior to CI insertion). We hypothesized that the intracochlear ECochG response amplitudes would be larger than previously published round window measurements.^3,5-8 Our median ECochG TR amplitudes were ≥ 5 dB re: 1uV larger than those recorded at the round window prior to³ or after^8,26 CI insertion. However, our larger response amplitudes may have negatively affected the relationship with CNC scores; the lower round window ECochG responses appear to have had a strong statistical impact on the correlation. Furthermore, we were also limited by the noise floor in the CI device recording system and a more limited recording window. The degree to which larger intracochlear ECochG response amplitudes are balanced against these shortcomings is unknown.

While we did not find a significant relationship with speech perception scores as we had hypothesized, we did find, on post-hoc analysis, a similar relationship between ECochG response amplitudes and speech perception scores when excluding older patients. Consistent with Fontenot et al, addition of patients over 80 years old weakened the relationship,³ which could be from non-cochlear-neural factors.^27-29

Though studies using round window ECochG have demonstrated that the presence of a CI affects the ECochG response amplitude,²⁶ it remains unknown how much a CI impacts the intracochlear ECochG responses prior to full CI insertion using the same recording electrode. We attempted to determine this by examining the “peak response” during insertion with 6-month speech scores. The peak response originates from different cochlear locations at different time points during insertion for each participant. But, it can provide information on the cochlear environment prior to full electrode insertion. The “peak response” had a similar relationship with speech perception scores as the ECochG TR. Future studies using multi-frequency intracochlear ECochG during insertion as done by Saoji et al³⁰ might provide a better prediction of CI performance similar to that seen by Fitzpatrick et al.⁵ It is also possible that the lack of relationship seen with speech perception scores may be related to acute disruptions in the cochlear biomechanics that would affect the ECochG TR.^31,32

Lack of relationship between postoperative ECochG TR with speech perception scores

Our postoperative ECochG TRs, when combining all cases, slightly decreased over time, which was consistent with a prior study.¹⁰ One patient demonstrated an increase in postoperative ECochG TR at 1-year, but we suspect this was due to variation in the in-ear insert placement during the intraoperative recording. Though limited by small sample size and variation in number of cases per time point, those with residual hearing appeared to have stable postoperative ECochG TRs. This trend is similar to the work by Tejani, Abbas, and Brown et al^{33, 34} who found that postoperative ECochG responses remained stable in patients with residual hearing. Kim et al looked at the relationship between postoperative ECochG response amplitudes and speech perception scores in hybrid CI recipients³⁵ and did not find a relationship between the two in the acoustic and electric hearing condition. A post-hoc analysis in our dataset confirmed a lack of relationship between the 6-month postoperative ECochG TR and 6-month speech perception scores. While device failure could potentially impact these results, those patients flagged for possible device failures were not effect modifiers on exploratory analysis. The lack of relationship between postoperative ECochG TR and speech perception scores suggests that factors outside of acoustically-evoked cochlear physiology drive performance following electrode insertion.

Comparison of Behavioral Audiometric Thresholds with ECochG Threshold Estimates

Intracochlear ECochG-based measurements can estimate behavioral audiometric thresholds in CI recipients.^16,19-21 While not significant, our data suggests a mild to moderate trend between preoperative behavioral thresholds and intraoperative ECochG threshold estimates. There are several possible explanations as to why we did not find a significant correlation. First, the measurements in the present study are comparing thresholds taken at two separate time instances (preoperative versus intraoperative) as opposed to thresholds taken at the same time point.^16,19-21 The status of the middle ear, mastoid, and inner ear are all very different in the intraoperative condition. Factors such as (1) blood and fluid in the middle ear, inner ear, and mastoid, (2) lack of a well-formed seal at the round window, and (3) a lack of skin covering the mastoid cavity create acoustic conditions different to those of the preoperative condition. Nonetheless, these findings may support those of Koka et al suggesting ECochG can be used to estimate frequency-specific auditory thresholds.¹⁹ ECochG might be useful clinically both intraoperatively and postoperatively as a way to estimate auditory thresholds outside of the clinic as well as in pediatric populations that are challenging to capture frequency-specific behavioral thresholds.

The strengths of our study demonstrate (1) the feasibility in making intracochlear ECochG measurements intraoperatively and postoperatively to monitor the change in responses and (2) the feasibility of estimating audiometric thresholds at the time of cochlear implantation and beyond. The limitations of our study include a small sample size, which prevents us from making definitive conclusions due to imprecision in the estimates, and inclusion of participants (n=5) that were identified as having a potential device failure though it was not a significant modifier. Future studies should compare the change in ECochG response amplitude at the round window (prior to CI insertion) and after CI insertion using the apical electrode of the CI to determine the degree of change per subject. Such studies could provide insight as to whether full electrode insertion acutely changes the cochlear environment and prevents acquisition of meaningful ECochG responses.

Conclusion

There was no association between intracochlear ECochG response amplitudes and speech perception scores. But the data suggests a mild to moderate relationship between preoperative behavioral audiometric testing and intraoperative ECochG threshold estimates. Performing intracochlear ECochG before, rather than after, CI insertion may provide a more accurate assessment of a patient’s speech perception potential.