INTRODUCTION

Today, a technological alternative for patients suffering from profound bilateral sensorineural hearing loss is the cochlear implant (CI) which, based on banded spectrum analysis for extraction of significant voice attributes, obtains an electrical stimulation code which, applied to the patient’s hearing system, generates an auditory sensation. The FDA (Food and Drug Administration) recommends using CI in children from 12 months of age, due to the difficulty of performing behavioral hearing tests in younger patients. This is a limitation when seeking successful rehabilitation in the implanted patients, considering that the critical period for language acquisition comprises the first years of life ^{[1] [2] [3]}.

Four to six weeks following implant surgery, CI is programmed for the first time, starting the process of customizing the device to the patient’s needs, in order to achieve useful, safe, and comfortable hearing. The different parameters involved in CI programming include establishing the dynamic range of electrical stimulation current in each intracochlear electrode, delimited by the minimum (T) and maximum (M) levels corresponding to the threshold and maximum tolerable auditory sensation, respectively ^[4]. In clinical practice these current levels are determined by observing the patient’s auditory behavior in response to the electrical stimulation provided for a limited number of electrodes. This behavioral CI fitting can be more difficult in pediatric patients who lack the communication skills needed to obtain reliable feedback information.

This scenario has led to the use of electrophysiological methodologies such as Brainstem Auditory Evoked Potentials (BAEPs) ^{[5] [6]}, the Electric Stapedial Reflex Threshold (ESRT) ^{[7] [8] [9] [10]}, and the Electrically-Evoked Compound Action Potential (ECAP) ^{[5] [9] [10] [11]}, to estimate T and M current levels for the electrode array. Of these methodologies, ECAP is the most used because it can be recorded for selected electrodes during or after implantation surgery ^[12].

Unfortunately, neither ECAP nor other electrophysiological methodologies mentioned previously, allow the estimation of hearing thresholds. The audiometry is the only clinical test available to determine hearing thresholds for a number of standardized frequencies ^{[13] [14] [15] [16]}, which requires the conscious participation of the patient.

Audiometry results are reliable in patients older than 4 years old, while in younger patients it is difficult to perform because they frequently do not present perceptible behavioral changes. It is even more difficult to observe the changes in implanted patients, especially in their first CI programming session because they are not familiar with the use of the device ^[17].

Furthermore, standard audiometry was not originally designed for implanted users. Test standardized frequencies set does not consider the band pass filter number for the spectral analysis that CI sound processor performs on the external input sound. The number of the band pass filter depends on the number of active electrodes and the IC manufacturer. Therefore, if any audiometry test frequency fall outside the band-pass filter bandwidth, corresponding to a certain intracochlear electrode, the hearing threshold may not be evaluated correctly.

Despite audiometry application difficulties and limitations in implanted pediatric patients, this test is used as a guide to readjust stimulation current level in the intracochlear electrode array, in order to lead the patient from an auditory status equivalent to profound deafness to normal hearing. The methodological limitations to determine the optimal M and T current levels across the electrode array, by using electrical stimulation tests like ECAP, as well as the difficulties in estimating the hearing thresholds due to the adjustment of the current levels, especially in pediatric patients, has generated the need to develop new alternative methods to evaluate hearing.

Our research group at the Metropolitan Autonomous University (UAM) Audiology Laboratory has developed a test called Electrical Cochlear Response (ECR) ^[18], as an alternative method for hearing threshold estimation in implanted pediatric patients. ECR measures the electrical response of the auditory nerve due to electric stimulation every time CI processes an incoming external sound. The ECR considers the auditory nerve portion responsiveness to the electrical current involved with each intracochlear electrode. For a given dynamic range of electrical current in an intracochlear electrode, ECR allows establishing the minimum sound intensity level for which auditory nerve portion near to intracochlear electrode generates a threshold electrical response.

Through individual electrode current adjusting, ECR test can help to estimate the auditory threshold at each of the intracochlear electrodes. This is done by quantifying the change in amplitude of the ECR due to variations in the sound intensity level of the input sound.

Previous ECR results have shown that the activation profile of the intracochlear electrodes follows the audiometric threshold profile ^[19]. These results have shown the feasibility of estimating the patient’s hearing thresholds based on ECR, which is independent of the age, general health condition, previous time of use, or brand of the CI.

On the other hand, considerable efforts have been made in the clinical field to establish an auditory-verbal rehabilitation program for pediatric CI users, by using a standardized set of tests that include questionnaires about child’ hearing abilities development, and evaluation of speech and sounds discrimination ^{[20] [21]}. However, up to now, there are no standardized methodologies to quantify the patient’s hearing performance during the period of adaptation to the device.

The purpose of this work is to show the use of ECR to estimate hearing thresholds in implanted pediatric patients and to evaluate hearing performance through the application of a set of tests designed to assess the patient's hearing abilities (see Table 1).

MATERIALS AND METHODS

Population

We included a group of 25 pre-lingual CI users between 1.4-6.6 years of age, who underwent CI surgery at the National Institute of Rehabilitation (INR), in Mexico City. Patients were diagnosed with bilateral profound sensorineural deafness and used hearing aids for at least six months previous to implantation surgery. All patients had a full insertion of the electrode array. Patients with neurological disease or ossified cochlea, and those without audiometry were excluded. Patients who did not complete the follow-up period were eliminated. The patient’s parents accepted to wake up their children around 10 hours before ECR test. Ethical Research Committee of INR approved and supervised the study complied with all applicable research and ethical standards and laws followed by the Declaration of Helsinki principles. All patients’ parents signed informed consent.

Two evaluation sessions were considered, T1: 1.5 to 9 months after CI implantation, and T2: 7 months after T1. The Figure 1 shows a flow diagram of the procedure used for patient´s hearing evaluation in each session.

Figure 1 Flow diagram procedure used for patient´s hearing evaluation in T1 and T2 sessions. 

Tests

Tests were performed at the INR audiology service in a 2.5x2.5x3.0 m audiometric test booth. The free field was calibrated according to ISO 389-7 Standard using a B&K 2235 sound meter, B&K 1625 filter, and a B&K 4230 microphone calibrator. The free field audiometry and ECR test were obtained in two different and successive days. Audiometry was performed using an Interacoustics clinical audiometer AC-40, while the patient was wearing the CI, seated a meter away from the loudspeaker. Pure Tone Average (PTA) was calculated in the frequency range of 500 to 2000 Hz to determine the patient's hearing level ^[15].

A prototype of ECR Monitor developed by the UAM Audiology Laboratory jointly with the company Innovamedica was used to obtain ECR. The prototype consists of one module to generate and reproduce sound stimuli, and other to acquire EEG, and a PC platform with software which synchronizes and organizes the acquisition of EEG with the reproduction of the stimuli (Figure 2).

Figure 2 Block diagram of the system for ECR acquisition. The tone pip is generated in the Stimulation Module and sent to the patient while the patient is asleep inside the audiometric test booth. Two channels of EEG recording are used in the Acquisition Module. ECR waveforms obtained from an average of 100 EEG epochs due to a tone pip of 1278 Hz, for an increasing intensity level. 

Two recording EEG channels with A1(-), A2(-), Cz(+) and FPz(GND) according to 10-20 system ^[22], observing an electrode-skin impedance < 5 k_Ω at 100 Hz were used. One hundred 50 ms EEG epochs were averaged, previously filtered with a second-order Butterworth low-pass filter of 0.1 to 300 Hz with cutoff slope of 12 dB/Octave, zero phase, _±10 _μV artifact rejection window, and a sampling frequency of 20 kHz.

The waveform of ECR response (*) is shown in Figure 2 for a stimulation tone pip of 1278 Hz (central frequency assigned to the intracochlear electrode number eight). The increase in the ECR amplitude and the decrease in latency are due to the increase in the intensity of the tone pip. The ECR threshold is defined as the minimum intensity for robust ECR detection.

During the ECR test the patient remained asleep on a cot with the head facing a loudspeaker positioned at a distance of one meter, with CI operating under everyday usage conditions, according to the CI programming parameters previously defined by the audiologist. The test was performed in the presence of the patient’s parents and supervised by the audiologist.

A PTA_ECR value was calculated considering intracochlear electrode central frequencies in the frequency range of 500 to 2000 Hz (540, 642, 762, 906, 1076, 1278, 1518 and 1803 Hz). Patients' hearing level was classified as normal (0-20 dB_HL), mild deafness (20-40 dB_HL), moderate deafness (40-60 dB_HL), severe deafness (60-80 dB_HL), and profound deafness (>80 dB_HL) ^[9]. Additional intermediate audiometric threshold values were interpolated to compare audiometry with ECR test result frequency by frequency.

Once ECR test finished, patients' parents answered the ASQ to provide information about the child’s ability to use the CI in daily situations and the ability to detect, identify, and repeat 18 ambient sounds, as indicated in Table 1. Items evaluated were taken from the IT-MAIS/MAIS scale and the LiP Profile ^{[20] [23]}. The CST was designed to evaluate the patient’s ability to detect, identify, and repeat 6 seconds duration real sounds and two-syllable Spanish words (Table 1).

All the CST items were normalized in amplitude and grouped in the following five groups: 1. An extended version of Ling sounds, which include eight representative phonemes of the speech spectrum ^{[1] [24]}; 2. Thirteen animal sounds; 3. Seven transportation sounds; 4. Ten ambient sounds ^{[1] [25]}; 5. Fourteen two-syllable Spanish words ^[1]. Ling sounds -extended version- and two-syllable Spanish words were recorded in a soundproof chamber at INR Audiology Service.

During CST, the patient remained seated at a distance of one meter from the loudspeaker. Three different sessions of 40 minutes were necessary to complete CST. At each session, cards with allusive images to sounds were given to the patient to keep their attention during the test. Each item was randomly played up to three times, starting at 60 dB_HL, while the therapist registered the patient's responses. The test was performed in the presence of the patients’ parents, under the everyday CI program mode.

ASQ and CST scores achieved by patients were used to quantify the hearing performance at the end of the follow-up period. Table 1 details all test items and the evaluation criteria considered to achieve a hearing performance of 100%.

RESULTS AND DISCUSSION

Of the 25 patients initially included, seven were eliminated because they did not complete the follow-up period. Two of them experienced problems with the implant and five did not complete the battery of tests. The average age of the 18 patients included in the analysis (9 girls and 9 boys) was 3.2±0.9 (1.8-5.7) years. The average auditory age (elapsed time between CI activation and ECR recording) was 7.2 months. Information about the patients is presented in Table 2.

Figure 3 shows the mean thresholds values for audiometry and ECR for all patients in sessions T1 and T2. A significant decrease in these thresholds is observed in T2 in relation to T1 (p<0.05), 11±2 dB_HL and 7±3 dB_HL, respectively, indicating an improvement in hearing thresholds over time. ECR thresholds were obtained at lower intensity than audiometry, 7 dB_HL at T1 y 3 dB_HL at T2, on average. Significant differences (p<0.05) were observed for 1076, 1278 and 6665 Hz in T1. In both testing sessions, the ECR threshold values showed less dispersion (≈10 dB_HL) than the audiometry (≈20 dB_HL).

Figure 3 Mean and standard deviation values of the audiometric and ECR thresholds by frequency obtained from 18 patients, ages 3.2_±0.9 years, at T1 and T2 sessions. 

Audiometric and ECR thresholds, PTA, PTA_ECR, ASQ and CST scores did not show significant differences when analyzed by gender. For the analysis by chronological age, patients were divided into 2 groups: 3 years and younger (G1; n=8) and older than 3 years (G2; n=10).

Table 3 shows significant PTA differences between G1 and G2 groups in the two sessions, with lower values in G2; while the PTA_ECR do not show significant changes between G1 and G2. In both sessions, PTA_ECR values were lower than PTA values for G1 group (p<0.05); 22 dB_HL at T1 session, and 13 dB_HL at T2 session. On the other hand, G2 group did not show significant differences between PTA and PTA_ECR values, 4 dB_HL at T1 (p=0.37) and 5 dB_HL at T2 (p=0.29).

Regarding ASQ, G1 group improved significantly their scores (p<0.05) from T1 to T2 session for IT-MAIS/ MAIS and Ambient Sounds Response -Detection-. The scores improvement for Identification (p=0.13) and Repetition (p=0.23) ambient sounds were not significant. For G2 group there was not a score significantly improvement for any ASQ items. Better scores were obtained by G2 group at T1 and T2 sessions and only for Identification, significantly differences between groups were found at both sessions.

Figure 4 shows mean audiometric and ECR thresholds for both patient groups at T2 session when patients’ cooperation got better allowing reliable audiometry. For G1 group ECR had lower threshold and less dispersion than audiometry. Significant differences (p<0.05), 17±2 dB_HL on average, between audiometric and ECR thresholds for frequencies above 1278 Hz were observed. For G2 group, audiometry and ECR had lower threshold than G1 group without significant differences between tests. Audiometry showed lower threshold, 5±3 dB_HL on average, and greater variability than ECR. In T2, after more than a year wearing the CI in most patients, the CST was evaluated to quantify the hearing performance, once the children became familiar with the use of the CI. G2 had better CST detection scores than G1.

Figure 4 Audiometric and ECR thresholds of patients 3 years of age or younger (group G1), and patients older than 3 years (group G2), in session T2. 

The largest changes between the two groups (G1 and G2) were found in animal sounds, environmental sounds, and two-syllable Spanish words at 40 dB_HL (Table 4).For group G1, the PTA_ECR of 38±6 dB_HL seems to show better relationship with CST detection scores at 40 dB_HL than PTA of 51 ±17 dB_HL showed in Table 3. The average hearing performance obtained for G1 and G2 groups was 32% and 58% respectively, (p<0.05).

Below is an example of a patient (S2) older than 3 years, who is already familiar with the use of his CI, which has the audiometric and ECR thresholds of Figure 5 and the CST detection scores of Table 5.

Figure 5 Audiometric and ECR thresholds obtained for patient S2 in session T2. 

According to audiometry patient S2 should detect all sounds at 40 dB_HL, however he is not. His CST detection scores are more related to ECR, where thresholds are below 40 dB_HL only at some frequencies.

A common practice for CI fitting purposes is to use ECAP threshold to establish M and T levels of the dynamic range of electrical current for a limited number of intracochlear electrodes, without considering the CI sound processor operation. Unfortunately, ECAP is not a predictor of post-operative performance ^[27]. Some studies had shown that ECAP threshold tends to fall over time ^[11], and is absent in approximately 5% of patients ^{[12] [28]}. ECAP thresholds profile does not fully coincide with M levels profile ^[12], hence final M and T levels values should be confirmed by expert observation of the patient’s auditory behavior and successive application of audiometry, which is not an easy task in pediatric patients.

Bear in mind that audiometry was not originally designed for implanted patients. Thus, the standard test frequencies have no relationship with the frequency bands assigned to the intracochlear electrodes. This may cause to underestimate the hearing threshold by up to 3 dB when test frequency falls within the bandwidth of any band-pass filter or an inadequate estimation of hearing threshold if it falls outside bandwidth.

Due to its objective nature, ECR can be use since the initial CI programming. Furthermore, the utilization of sound stimulus of frequency equal to the central frequency assigned to each stimulation electrode allows individual addressing of intracochlear electrodes.

Additionally, ECR thresholds showed less dispersion than audiometry, probably due to its objective nature compared to the subjectivity of the audiometry which generated greater deviations.

The high variability found in the audiometry of the youngest patients, G1, may be due to their short age, limited experience using the CI, and the difficulty they presented in performing the test.

In the other hand, it was difficult to obtain thresholds at high frequencies in these patients. For older children, G2, according to the audiometry, all of them would be in mild deafness, with hearing thresholds below 40 dB_HL, Figure 4, however the CST shows detection scores less than 80% for 40 dB_HL.

This is probably due to the fact that most patients have not developed the skills to detect the different sounds evaluated. G1 achieved a total CST detection score of 55% and G2 of 82% after approximately one year of CI use, Table 4.

The patient S2 hearing thresholds illustrate how the inherent subjectivity of audiometry affected the determination of hearing thresholds for test frequencies below 906 Hz, where an average difference of 20 dB_HL was observed with respect to the ECR thresholds after 8.4 months of CI activation (Figure 5).

This study corroborated the feasibility of evaluating the patient-CI pair making the use of known frequency and variable intensity tone pips with the CI operating under everyday usage conditions, which take account of the effect of microphone sensitivity and sound processor gain.

A limitation of the ECR test is that requires patients to remain asleep during the entire test. Nevertheless, the results showed the possibility of knowing the minimum sound intensity level for the detection of ECR corresponding to the probable hearing threshold. The relationship between patients’ hearing performance improvement and ECR was also shown after modifying T and M levels, and/or extent of the dynamic range of electrical current derived from the amplitude and waveform analysis of the ECR.

The promising results provided in this study lead us to propose ECR as a tool for the evaluation and follow-up of CI users, especially for ages ≤ 3 years. Although more studies are needed to show the possible benefits of this objective approach from the activation of the CI.

CONCLUSIONS

The most significant differences between audiometric and ECR thresholds obtained by PTA and PTA_ECR were observed for patients aged 3 years or younger. Additionally, no statistically significant differences were obtained for older patients. However, the similarity in the threshold profiles obtained in both tests allows us to think about the possibility of obtaining hearing thresholds using ECR.

Furthermore, our study highlighted the limitations of audiometry to provide information on the hearing threshold due to subjectivity and the absence of a relationship between test frequencies and the central frequencies of the band pass filter assigned to each CI intracochlear electrode.

The application of ASQ and CST tests was very difficult due to the age of the patients and the lack of experience in the use of the CI; however, these made possible to relate auditory behavior with audiometry and ECR, where a better relationship was shown between ASQ and CST scores and ECR thresholds.

This research project showed the possible benefits of employing ECR to estimate hearing thresholds per frequency in pediatric implanted patients, especially younger than 3 years, as well as the implementation of ASQ and CST to quantify hearing performance. During the patients' follow-up period, the methodology developed helped audiologists and therapists to identify some difficult situations related to CI programming and rehabilitation therapy strategy, helped them to quantify and improve the performance of their patients, and to have additional information about their hearing.