Neuromonics Tinnitus Treatment: Audiometric Requirements
AUDIOMETRIC CONSIDERATIONS
Essentially, the main audiometric requirement is the ability to measure hearing thresholds in a calibrated fashion across the range from 250 Hz to 12.5 kHz. While the 250 Hz to 8 kHz is standard for audiometry, this is insufficient for tinnitus clinic purposes, as many of those patients who were thought to have normal hearing actually have a loss restricted to the very high frequencies. An important study of very high frequency audiometry in tinnitus sufferers found that there was an almost universal incidence of hearing loss in frequencies greater than 8 kHz (Domenech, et al., 1991). Recent evidence indicates that the high frequency audiometric hearing thresholds are valid and reliable, and constitute a better early indicator of noise-induced hearing loss and acoustic trauma than the lower frequencies (Ahmed, et al, 2001). That report also noted that these frequencies are rarely tested in the usual clinic assessments, despite actually constituting a very large proportion of human frequency sensitivity. It concluded that the very high frequencies were an accurate indicator of any damage to the auditory system, even if the hearing loss was very limited.
Recent animal studies of the neurophysiological response to noise trauma found that it leads to a high frequency hearing loss, particularly worse in the frequencies greater than 8 kHz (Norena & Eggermont, 2005). It was found that this loss can lead to a strongly reorganized tonotopic map, but this was also able to be prevented by providing an enriched acoustic environment in the frequency range of the hearing loss. Lower frequency sound enrichment clearly did not work as well as sounds that matched the high frequency loss. As a result of this objective evidence, those researchers have subsequently recommended that tinnitus treatments provide acoustic signals that are tailored to correct for hearing loss, so that stimulation is provided across the broadest possible range of neurons (Eggermont, 2005).
These experimental findings of the role of the higher frequencies in tinnitus are also consistent with clinical experience with patient descriptions of their tinnitus, and the greater incidence of a more conclusive pitch match when the higher frequencies are offered as an option (Vernon & Meikle, 1988). In a sample of 90 consecutive tinnitus clinic patients, it was found that the substantial hearing loss was generally only apparent in the frequencies greater than 8 kHz (Davis, Wilde and Steel, 1999). That study also found that mean pitch match was 8 kHz, and so many patients fell above that average. Thus, without those higher frequencies to offer as a comparison with their tinnitus, it is not possible to check for octave confusion.
Clinical measurement of tinnitus seems to have had less attention in recent years, perhaps because the reasons for audiometric measurement may not have been clear to early TRT researchers. Evidence for this was that the early noise generators were called white noise generators, even though they did not produce an equal frequency response across the full hearing range. One study of actual ear-level output showed that most patients with a typical hearing loss only received a narrow band of noise centered around 2 kHz (Sheldrake, Coles & Foster, 1995). Other prominent European tinnitus researchers have subsequently proposed that the efficiency of TRT might be improved by modifying acoustic stimuli to account for each individual’s hearing loss (Baguley, Beynon, and Thornton, 1997). In keeping with this, the Neuromonics treatment aims to provide the broadest possible frequency range of acoustic output to ensure maximal stimulation of all corresponding acoustic pathways.
Another of the aims of the Neuromonics Tinnitus Treatment is to provide a stimulus that is able to be effective at the lowest possible sensation level, and so enable a relaxation response to be maximized. This is in keeping with experimental comparisons of the effects of masker bandwidth by Kemp and George (1992), who found that the broadest of the bandwidths tested (0.1 to 15 kHz) tended to mask at the lowest sensation level, and that the additional high frequency response tended to most benefit those with higher pitched tinnitus. As the long term spectral average of relaxation music has a low frequency emphasis, it is the antithesis of what is required for those
with a high frequency hearing loss, and considerable spectral modification is required to counter this (Davis, Wilde & Steed, 2002). Neuromonics also aims to maximize perception of natural musicality across the whole range of frequencies that normal hearing listeners take for granted. To enable this, the hearing thresholds across the full frequency range of human hearing need to be reliably measured so that the output of the Neuromonics Processor can be individualized accordingly.
Calibration Considerations
The lack of calibration standards for defining dB HL in the higher frequencies had previously retarded their practical measurement, but in 1998, consensus was reached and an international standard was published (ISO/TR 389-5:1998). Prior to that, the USA (ANSI) and European Union (EU) homogenized several audiometric equipment standards. As there was no ANSI standard for equipment for the high frequencies, the USA adopted the EU standard, referred to as: IEC 645-4:1994 Audiometers. Part 4: equipment for extended high-frequency audiometry. This standard specifies the need for equipment to provide pure tones at 10 kHz, as well as at 9, 11.2 and 14 kHz. This is because the physical dimensions of the higher frequencies are acoustically quite distinct, and the tonotopic representation in the cortex can be expected to be arranged equally as distinctly.
A practical reason to determine hearing thresholds at least around 10 kHz is that many tinnitus patients with significant hearing loss across the speech range have no residual hearing left at equipment limits at 12.5 kHz (typically limited to 80 dBHL). Additionally, when the mean hearing thresholds in a tinnitus clinic sample at 8, 10 and 12.5 kHz (Davis, 1998) are graphically plotted, an extrapolation of average values at 10 kHz underestimates the actual mean thresholds by more than 5 dB. As the standard deviations were also greater for the high frequencies than for the speech frequencies, some individual’s variations would even be greater, and so directly measured (rather than extrapolated) thresholds are much more empirically accurate.
Audiometry for the >8 kHz region requires special headphones. The standard audiometric headphones have changed little in the last half century, and were not constructed with the measurement of the high frequencies in mind. The standard covers three styles. The insert type is the Etymotic Research ER-2, the closed supra-aural type earphone is the Sennheiser HDA 200. The open type supra-aural earphone is the Koss HV/1A, but it is now out of production (Frank, 2001).
The test/retest reliability of high frequency audiometry has been shown in several controlled studies to be acceptable once the purpose-built earphones are used (ISO/TR 389-5:1998). Because of the dimensions of the very high frequency sound waves, partial canal collapse has a greater effect than for the < 8 kHz frequencies, and so all high frequency earphones are either insert style or supra-aural to reduce pinna-induced collapse. Additionally, since partial wax occlusion has a greater effect in high frequencies than it does for the < 8 kHz frequencies, this may need greater care. It is also a reason why we have a preference for the supra-aural type. The insert phones also have the disadvantage of being less robust and requiring refill eartips, and they produce a sense of occlusion that reduces the validity of tinnitus pitch/loudness matching.
Some audiometers offer pure tones at 12.5 kHz while others do so at 12 kHz; practically speaking, these two are considered to be interchangeable as far as calibration is concerned, due to the purpose-built high frequency headphones having a flat frequency response in that region. Maximum output of audiometers for the higher frequencies can also be much lower than the speech range frequencies. For example, the Maico MA53 maximum output at 12.5 kHz is 80 dBHL. Thresholds obtained at higher intensities than that might be erroneous due to distortion (Fausti 2002, Personal Communication).
Other Tinnitus Clinic Measurement Considerations
There are several audiometer features required to provide the Neuromonics Tinnitus Treatment. Some of the recommended audiometric testing procedures bear much resemblance to earlier tests, but are now advocated for somewhat new reasons. For example, residual inhibition testing is now advocated, not so much as a prognostic indicator, but instead primarily to dampen down any tinnitus exacerbation that may arise from performing Loudness Discomfort Level (LDL) assessments. To perform Residual Inhibition assessments properly, the audiometer needs to be able to provide Narrow Band Noise in both ears simultaneously, and so it is included in the following list of audiometric requirements.
The ability to provide all hearing thresholds through the same headphones is required to be able to quickly provide two-alternative-forced-choice pitch match comparisons across the full frequency range. Pitch matching is performed these days more for counseling than psychophysical reasons. Having the one set of headphones also saves clinician time by not having to constantly open the booth door to change headphones.
The loudness growth typically displayed in tinnitus patients is markedly steeper than those with a matched sensorineural loss but no tinnitus (Davis, 1998). This greatly reduced dynamic range is consistent with a central gain-type of phenomenon. Consequently 2 dB steps are required for many tinnitus measurements, and so easy access to changing dB steps is helpful, particularly for LDL testing.
The high-frequency pack’s broad band noise output is generally broader than the pink-type (speech) noise that other audiometers actually provide, and so covers more of the range of frequencies relevant to tinnitus. The true white noise Minimum Masking Level is more representative of the effect that the Neuromonics Tinnitus Treatment can provide.
Upgrading Your Audiometer, or Choosing a New One
The high frequency pack can be retro-fitted to most existing audiometers listed here. The headphones are typically the most expensive component, at around $US 1,000. The kit (including headphones), installation and calibration cost between $US 1,500 - $US 2000. Once the equipment is set up, the Tinnitus Clinic can also now offer an ototoxicity monitoring service for agencies such as Oncology departments (Frank, 2001).
Neuromonics’ own clinics in Australia currently use Maico MA 53 audiometers with HDA200 earphones. However, the Maico 53 is not available in the United States and purchasing a used unit from the internet is not recommended. The MedRx Avant Audiometer is a PC based audiometer that is quite portable and includes all necessary features. The Interacoustics Equinox is a PC-based audiometer with all of the necessary features though decibel step sizes are set at only 5dB and 1dB. The popular GSI 61 does not allow all frequencies to be accessed through one set of headphones unless properly modified. The modification to the Sennheiser headset can be made by most calibration technicians. It is important to specify that this modification be made when the appointment for calibration is made as the modification is not usually done on-site. Additionally, gold plated stereo patch cables are necessary with the GSI 61. If the headphone modification is not done the GSI 61 is quite cumbersome switching from headset to headset during pitch match testing. The Madsen Itera II is good value, but the lack of ability to perform proper residual inhibition procedures is a limitation. It can be somewhat clumsily circumvented by using a binaural CD recording of all the narrow band noises, and playing this back through an auxiliary input.
Table 1 Feature Requirements*
The following table assumes that the high frequency option has been installed and the audiometer has been calibrated to the high frequency earphones as per ISO 389-5 (1998).
MedRxAvant Stealth / GSI 61 / Interacoustic
Equinox / Madsen Itera II / Siemens
UNITY PC
Hearing threshold assessment / Essential / Ability to measure 10 kHz and 12 kHz (or 12.5 kHz) / √ / √ / √ / √ / √
Preferable / Ability to test all frequencies through the HF headphones / √ / √2 / √ / √1 / √
Preferable / Easy access to change dB steps / √ / √ / √ / √ / √
Preferable / Ability to also provide 11.2 kHz, 14 kHz and 16 kHz for pitch matching / √ / √ / √ / √ / √
Minimum Masking Level assessment / Essential / Ability to provide Broad Band Noise greater than pink noise (speech range) / √ / √ / √ / √ / √
Preferable / Ability to provide white noise that includes the 8-12 kHz range / ? / ? / ? / ? / ?
Loudness Discomfort Level assessment / Essential / Ability to provide pure tones in 2 dB steps / √ / √ / X 1dB step / √ / √
Residual Inhibition assessment / Essential / Ability to provide Narrow Band Noise (NBN) at each of the frequencies 0.25 to 12 kHz / √ / √ / √ / √ / √
Preferable / Ability to provide NBN in left and right channels simultaneously / √ / √ / √ / X / √
Graphing Requirements / Essential / Use of a paper audiogram that depicts the 10, 12 and 14 kHz thresholds. / √ / √ / √ / √ / √
Essential / Ability to graph (on paper) tinnitus measures in terms of sensation level and across time periods / √ / √ / √ / √ / √
Preferable / Use of an electronic audiogram that depicts the 10, 12 and 14 kHz thresholds / √ / √ / √ / √ / √
Preferable / The ability to electronically send test results to a database such as NOAH3, and to be able to network with other computers / √ / √ / √ / √ / √
Legend: