Individual gain at the threshold level (A) and at the most comfortable listening (MCL) level (B) as a function of frequency in all 14 patients. Data points for the 7 patients (represented by symbols) with the mildest hearing loss (pure-tone average, 40- to 59-dB hearing level [HL]) are connected by broken lines, and those of the other 7 patients (represented by symbols) (pure-tone average, 60- to 80-dB HL) are connected by solid lines.
Average gain at the threshold and most comfortable listening (MCL) levels as a function of pure-tone average (PTA) (average hearing loss at 0.5, 1.0, 2.0, and 4.0 kHz). Speech gain data are also given. HL indicates hearing level.
Individual most comfortable listening (MCL) levels as a function of frequency obtained with the Vibrant Soundbridge. The MCL levels of the 7 patients (represented by symbols) with the mildest hearing loss (pure-tone average, 40- to 59-dB hearing level [HL]) are connected by broken lines, and those of the other 7 patients (represented by symbols) (pure-tone average, 60- to 80-dB HL) are connected by solid lines.
Snik AFM, Cremers CWRJ. Vibrant Semi-implantable Hearing Device With Digital Sound ProcessingEffective Gain and Speech Perception. Arch Otolaryngol Head Neck Surg. 2001;127(12):1433-1437. doi:10.1001/archotol.127.12.1433
The Vibrant Soundbridge (Symphonix Devices, San Jose, Calif) is a semi-implantable hearing device. The transducer is attached directly to the incus and is linked by telemetry to the externally worn audioprocessor. A major advantage of this semi-implantable setup, especially during its experimental phase, is that the audioprocessor can be updated. Recently, we replaced the previous 2-channel analog audioprocessor in 14 patients with a 3-channel digital device.
Prospective clinical study. Basic functions were measured, including gain as a function of input level and speech perception in quiet.
Patients (n = 14) had moderate to severe sensorineural hearing impairment (average hearing threshold at 0.5, 1.0, 2.0, and 4.0 kHz of 40- to 76-dB hearing level [HL]) and chronic external otitis, which contraindicated use of an ear mold.
Gain of the 3-channel audioprocessor for comfortable listening levels and for conversational levels varied from approximately 15- to 30-dB HL, suggesting that the device is suitable for patients with hearing loss of up to 60- to 70-dB HL. In 5 patients, identical measurements were performed using their conventional hearing aids. The other 9 patients did not use a conventional hearing device because of severe external otitis. On average, results obtained with the Vibrant Soundbridge were not as good as those obtained with the conventional device. Nevertheless, patients were satisfied with the Vibrant Soundbridge because they could use it all day without pain or itching.
The Vibrant Soundbridge is suitable for patients with hearing loss of up to 70-dB HL. Compared with conventional devices, in audiometric terms, a surplus value of the Vibrant Soundbridge was not found.
THE VIBRANT Soundbridge is a semi-implantable hearing aid for individuals with moderate to severe hearing impairment.1- 3 It has been in use since 1997, and 350 patients have been fitted worldwide. So far, sparse audiometric data on its use have been published.
In 1999, Snik and Cremers3 published sound field gain data (as a function of input level) on 7 patients who were using the Vibrant Soundbridge. At that time, the externally worn audioprocessor was an analog, dual-band, wide dynamic range compression device called type 302. Mean gain (average gain at 0.5, 1.5, and 4.0 kHz) as a function of input level was 21, 17, and 5 dB at 40-, 65-, and 90-dB sound pressure level, respectively. On an individual level, gain was, on average, below target values, especially at low frequencies and with low-level sounds. It was concluded that more gain was desirable, particularly for low-level sounds. Most patients were using their audioprocessors at high or maximum gain.3
In 1999, a new and somewhat more powerful digital audioprocessor called type 304 became available. The device contains Senso (Widex, Copenhagen, Denmark) hardware,4 a 3-channel nonlinear processor with special features for (constant) noise reduction. This audioprocessor was fitted to the 7 patients who participated in the previous study3 and to 7 more recently implanted patients.
In the previous study,3 gain was assessed using loudness growth measurements. This procedure was time consuming and could be expected to be even more so with the 304 audioprocessor. The problem with the 304 audioprocessor is that the noise reduction identifies and then reduces constant sounds (such as the measurement stimuli). Release times are relatively long. In practice, this means that the duration of the stimuli has to be short, and the time interval between sounds has to be on the order of 10 seconds. To save time, a simplified procedure was used that measured gain at the threshold level (unaided minus aided sound field thresholds) and at the patients' most comfortable listening (MCL) levels (unaided minus aided MCL levels). Gain at the threshold level estimates gain for low-level sounds, whereas gain at the MCL level assesses gain for moderately loud sounds. Usually, conversational speech is between the threshold and MCL levels, close to the MCL levels.5
With linear amplification, gain at the threshold level (often referred to as functional gain) is the same as gain at the MCL level. For nonlinear amplifiers, such as the 304 audioprocessor, this is not the case, so threshold measurements alone are not enough; in this study, they were supplemented by measurements at the MCL level. In addition, aided and unaided speech audiograms were obtained. This also enabled suprathreshold evaluation of the 304 audioprocessor.
For comparison, identical measurements were performed in 5 patients who were using conventional air-conduction hearing aids before implantation. The other 9 patients had used a conventional hearing aid once in the past but had stopped using it because of external otitis. Although the number of patients is small, the comparison is of interest. According to the general inclusion criteria,6 candidates for a Vibrant Soundbridge are typically dissatisfied users of conventional hearing aids, which might cause bias. A major advantage with our patients is that their dissatisfaction was not caused by the technical performance of the conventional hearing aid; rather, they simply could not wear it.
At the Department of Otorhinolaryngology, University Hospital Nijmegen, Nijmegen, the Netherlands, 15 patients were fitted with the 304 audioprocessor; 14 fulfilled the inclusion criteria set by the European Investigators Group,6 namely, symmetrical cochlear hearing loss (within 10 dB) with threshold levels at 500 Hz of 30- to 70-dB hearing level (HL) and at 2000 Hz of 45- to 85-dB HL. The remaining patient had high-frequency deafness (hearing threshold at 1-4 kHz that exceeded 100-dB HL) and was excluded. (Nevertheless, this patient was satisfied and used her device the entire day.)
An additional inclusion criterion used in Nijmegen was that the patients had to have severe external otitis, which made it impossible or troublesome to use (any type of) ear mold. Before implantation, the pure-tone average (average hearing loss at 0.5, 1.0, 2.0, and 4.0 kHz) of the implantation ear varied between 40- and 76-dB HL (mean, 57-dB HL). The postimplantation pure-tone average, determined at least 2 months after surgery, was 37- to 77-dB HL (mean, 60-dB HL). In one patient, deterioration of more than 10-dB HL was found.7 Age at implantation ranged from 33 to 67 years.
Five patients had been using a conventional hearing aid before implantation; the other 9 patients had tried but stopped using a conventional hearing aid because of severe external otitis. They preferred poor hearing to the inconvenience of external otitis. Some relevant data on these 5 patients are presented in Table 1. These patients were evaluated twice, once with the Vibrant Soundbridge and once with the conventional hearing aid in the implanted ear. The volume was set at the usual daily volume used by the patients.
Aided and unaided hearing threshold and MCL levels were obtained in the sound field using warble tones (generated by the Interacoustics AC 40 audiometer [Interacoustic, Vaerlose, Denmark]; frequency modulation, 5%). These tones were presented by a loudspeaker placed 1 m in front of the patient. To deal with the relatively long release times of the 304 audioprocessor, a pause of 10 seconds was applied between successive sounds. For threshold measurements, the descending method was used. For MCL measurements, the patient had to rate the loudness of the tone, as described previously.3 The MCL level was determined adaptively.
Gain at the threshold level (unaided minus aided threshold levels) was obtained at octave frequencies from 0.25 to 8.0 kHz. Gain at the MCL level (unaided minus aided MCL levels) was obtained only at the 0.5-, 1.0-, 2.0-, and 4.0-kHz frequencies.
For sound field speech audiometry, lists of 13 monosyllables were used. Phoneme scores were obtained. The presentation level per list was constant and included 65 dB (conversation level), 80 dB, and at least 3 other levels to determine the shape of the intensity-recognition curve. The sound field speech audiograms were used to determine speech gain, which was defined as the shift in decibels between aided and unaided curves for the aided score at 65 dB, expressed in multiples of 2.5 dB. Thus, speech gain is the effective amplification of conversational speech.
Gain data obtained with the 304 audioprocessor were compared with those obtained with the 302 audioprocessor.3 Although the measurements were elaborated on more in the previous study (complete loudness growth curves), this comparison was valid because the same measurement stimuli and procedures were applied. In that study, measurements were performed at only 3 frequencies, including 1.5 kHz. In the present study, this frequency was not included, but 1.0 and 2.0 kHz were included. It was decided to average the results of these 2 frequencies and compare the value to the 1.5-kHz data obtained with the 302 audioprocessor.
Sound field testing was performed in a double-walled, soundproof room. During testing, the nontested ear was blocked with a foam ear plug.
The programmer of the 304 audioprocessor (Symphonix Devices, San Jose, Calif) enables adjustment of gain, maximum output, and compression knee points in 3 frequency bands. The fitting procedure was as follows: First, gain per band was adjusted until the patient was satisfied or extreme settings had been reached. The maximum output level was not critical because the implanted part has an output limiter. Mostly, maximum output was set close to maximum. Compression knee points, variable between approximately 20 and 60 dB sound pressure level, were set such that no feedback occurred, and patients were not bothered by device- or environment-related noise. After this initial fitting, the patient tested the settings by walking around, going to the hospital cafeteria; afterward, if necessary, the settings were further adjusted. Six weeks later, patients returned to the clinic and, if necessary, the settings were adjusted once more, and a new follow-up appointment was made. Evaluation took place when the patient and the audiologist were satisfied with the result or the settings were set at maximum, thus, at least 6 weeks after fitting of the 304 audioprocessor.
Figure 1 shows individual gain at threshold level curves and at MCL level curves as a function of frequency. A large spread was seen, largely owing to the wide range of hearing loss in patients; therefore, mean gain values were calculated (average data at 0.5, 1.0, 2.0, and 4.0 kHz) and displayed as a function of patients' pure-tone averages in Figure 2. Speech gain is also presented in Figure 2. There was still considerable spread in results. Pearson correlation analyses showed a significant relation between pure-tone average and gain at the threshold level only (ρ = 0.68; P = .01). The 3 gain measures were significantly interrelated (tested at the 5% level). Averaged over the 4 frequencies and 14 patients, gain at the threshold level was 33 dB; at the MCL level it was 21 dB, illustrating the nonlinear sound processing. Mean speech gain was 22 dB, which was comparable with gain at the MCL level but not at the threshold level.
The phoneme speech recognition score at 65 dB (PS65) improved from an average of 21% (range, 0%-67%) in the unaided condition to 77% (range, 39%-100%) using the Vibrant Soundbridge. It has been suggested5 that the target value for the aided MCL levels should be approximately 65-dB HL. Individual MCL levels obtained with the Vibrant Soundbridge are presented in Figure 3. Again, a large spread was seen. The subgroup with more severe hearing loss had the highest (poorest) MCL levels. Several patients with more moderate hearing loss had remarkably low MCL levels in the middle frequencies.
Table 2 provides gain data from 5 patients who had been using conventional devices before implantation. Their results with the Vibrant Soundbridge were compared with those obtained with the conventional hearing aid in the implanted ear (the other ear was plugged). Columns 10, 11, and 12 in Table 2 show difference values: gain obtained with the conventional device minus that obtained with the Vibrant Soundbridge. A negative sign indicates a better result with the Vibrant Soundbridge, which was found only for patient 3. The last column indicates the change in PS65 score. To test the significance of the change in PS65, the method of the binomial distribution was applied, as described by Lyregaard.8 It was concluded from this analysis that the change in PS65 was statistically significant in patients 1, 3, and 5 (P = .05). For patient 2, the change in PS65 was clearly significant (P<.01).
In 6 patients, the results obtained in the previous study3 using the 302 audioprocessor were compared with those obtained in the present study with the 304 audioprocessor. Gains at the threshold and MCL levels were compared for the 0.5-, 1.5-, and 4.0-kHz frequencies. Mean ± SD gain at the threshold level was significantly higher with the 304 audioprocessor, ie, 7 ± 5 dB (t test, P<.05). Improvement was 8, 7, and 5 dB at 0.5, 1.5, and 4.0 kHz, respectively. No significant change was found in gain at the MCL level (t test, P>.05).
Using the 304 audioprocessor, mean gain for soft sounds, reflected by gain at the threshold level, was 33 dB (range, 20-45 dB) (Figure 1A and Figure 2). At more significant listening levels, reflected by gain at the MCL level, gain varied between approximately 15 and 30 dB.
Mean gain at the MCL level agreed reasonably well with speech gain, which also varied between 15 and 30 dB (Figure 2). Both gain measures were significantly related, indicating that "effective" gain of the Vibrant Soundbridge was better quantified by gain at the MCL level than by gain at the threshold level. This suggests that the application range of the Vibrant Soundbridge in terms of mean upper limit thresholds is approximately 60- to 70-dB HL at most, thus, approximately 10 dB lower than suggested previously.6 Although all the audioprocessors were adjusted according to the same protocol and were individually fine tuned, there was a large range in gain values (Figure 2).
Target values have been postulated for aided MCL levels. As indicated by Cox et al,5,9 aided MCL levels obtained with tones should be between 50- and 80-dB HL, irrespective of the frequency. Most aided MCL levels were within this target range (Figure 3). There was a clear trend toward poorer MCL levels in patients who had the most severe hearing loss, which strengthens the suggestion that the upper limit thresholds set for application of the Vibrant Soundbridge are probably too high. The low MCL levels (<50-dB HL) seen in the middle frequencies in several patients are puzzling. The resonance frequency of the transducer lies in this frequency range, so resonance phenomena might have played a role.
Table 2 gives the results of the comparison between the Vibrant Soundbridge and conventional hearing aids in 5 patients. As expected in the case of linear amplification (patients 1-4), gain at the threshold and MCL levels and speech gain were highly comparable. The conventional device used by patient 5 had nonlinear sound processing; indeed, gain was lower at higher input levels (at the MCL level and speech level) than at the threshold level. This also applied to use of the Vibrant Soundbridge. Gain at the MCL level, speech gain, and phoneme score were higher with the conventional device in 4 of 5 patients. On average, gain at the MCL level and speech gain were 8 and 9 dB higher, respectively. The phoneme score was 10% higher, which is less than expected based on the difference in gain. Ceiling scores may have played a role (compare the PS65 in Table 2 with the maximum phoneme score in Table 1). The poorer results with the Vibrant Soundbridge should not be generalized because the number of patients was small. However, the results are important because these patients did not choose implantation because they were disappointed conventional hearing aid users but because the use of any ear mold was troublesome. Therefore, our patients did not have any major complaints about their conventional hearing aids. When questioned, patients 1 to 3 and 5 said that they used the Vibrant Soundbridge all day; patient 4 used it only occasionally. Patients 1, 2, 4, and 5 still used the conventional device in the contralateral ear when communication demands were high. This enabled better speech recognition, in which binaural hearing obviously plays a role.
Patient 5 is a special case. Before implantation she had been using binaural Widex Senso C8 (Widex) devices. This device has the same sound-processing capabilities as the 304 audioprocessor. Speech perception with the Senso C8 was better than with the Vibrant Soundbridge. Nevertheless, the patient was satisfied with the Vibrant Soundbridge owing to freedom from an irritating ear mold, but she preferred the Senso C8 device for communication.
In a previous study,3 we evaluated the 302 audioprocessor. One conclusion was that on an individual level, measured gain was in fair agreement with hearing threshold–based target values, except for low-level sounds.3 A discrepancy of 20 dB was reported at 0.5 kHz and approximately 10 dB at 1.5 and 4.0 kHz. In this study, 6 patients were reevaluated after being updated to the 304 audioprocessor. Gain at MCL levels did not change, but there was improvement of approximately 5 to 8 dB for soft sounds (as assessed with threshold measurements). This means that, on average, the former discrepancy between measured and target gain for low-level sounds was reduced by a factor of 2 by fitting patients with 304 audioprocessors.
Accepted for publication July 17, 2001.
The audioprocessors were provided by Symphonix Devices Inc.
Corresponding author and reprints: Ad F. M. Snik, PhD, Department of Otorhinolaryngology, University Hospital Nijmegen, PO Box 9101, 6500 HB Nijmegen, the Netherlands (e-mail: firstname.lastname@example.org).