Non-linear amplification that incorporates compression with a low compression ratio has advantages over linear amplification for children.1 Because children (especially infants) cannot adjust the volume control of a hearing aid, a non-linear hearing aid that automatically provides higher gain for soft sounds and lower gain for loud sounds ensures adequate audibility for low input levels and listening comfort for high input levels.
The National Acoustic Laboratories (NAL) have developed a procedure for prescribing non-linear hearing aids, which is known as the NAL-NL1 procedure.2 It aims to maximize speech intelligibility while maintaining overall loudness at a level similar to that perceived by a normal-hearing person listening to the same sound. For each input level, a different gain-frequency response is prescribed to achieve these aims. NAL-NL1 is implemented as a stand-alone software program, and is also incorporated in many manufacturers software for hearing aid fitting and measurement. The way in which NAL-NL1 differs from other prescriptions3 and the method by which the effectiveness of audibility for speech intelligibility is calculated in the derivation of the NAL-NL14 are well documented. Additionally, the rationale of maximizing speech intelligibility relative to normalizing loudness at each frequency has been validated for adults.5
This article focuses on the implementation of NAL-NL1 in fitting hearing aids to children using a real-ear aided gain (REAG) approach that incorporates the measurement of individual real-ear-to-coupler difference (RECD). It is well accepted that the hearing aid prescription procedure for infants and young children should be based on REAG rather than real ear insertion gain (REIG).6,7 A procedure based on REAG prescribes the same gain from the free field to the eardrum for the same degree of hearing loss, irrespective of the gain provided by the unaided ear canal. A prescription based on REIG, by contrast, prescribes gain in SPL at the eardrum when aided, relative to the SPL at the eardrum when unaided.
Using an REAG procedure for young children seems reasonable, given that the amplified levels at a childs eardrum must provide appropriate stimulation to the childs auditory system despite changes in the resonant frequencies of the childs unaided ear canal during the first few years of life. It is not until the acoustic characteristics of the childs ear canal approximate those of an average adult ear canal when insertion gain targets can provide an equivalent fitting. Before then, as one may surmise, different insertion gains would be required for a 6-month-old baby than for a 12-month-old infant to achieve the same excitation levels within the cochlea.
The National Acoustic Laboratories support this REAG approach, developed as part of the DSL method6, for prescribing hearing aids for infants and young children. Fitting to REAG targets is readily applicable clinically through the use of contemporary hearing aid fitting systems with the NAL-NL1 stand-alone software.
The REAG Approach
The real-ear-to-coupler difference (RECD) is a vital component in implementing an REAG approach when verification of gain targets is carried out in a coupler. The RECD is defined as the difference between the sound pressure level near the eardrum and the sound pressure level measured in a coupler. When the RECD for an individuals ear is known, a hearing aid can be adjusted in a coupler to accurately achieve a specific REAG target in the real ear.8.9 Because the coupler response required for achieving the REAG target depends on the acoustics of the individuals ear and the earmold used, it is recommended that the measurements be carried out using insert earphones and the childs own earmold.
The accuracy of RECD measurement is directly related to the insertion depth of the probe tube in the ear canal. Two factors need to be considered in the positioning of the probe:
1) The tip of the probe tube needs to be at least 5 mm past the end of the earmold so that the point of measurement is not within the region where the sound wave is making a transition from the narrow sound-bore in the earmold to the wider ear canal.
2) The probe tip needs to be close enough to the eardrum to control for the measurement error caused by standing waves in the ear canal, but not so close as to lead to discomfort.
A method for positioning the probe tube at a desired distance from the eardrum is available.7 This method involves finding the location of a minimum (the result of standing waves in the ear canal) by presenting a 6 kHz tone and adjusting the insertion depth of the probe tube. The location of the 6 kHz minimum corresponds to a region 15 mm from the eardrum. If an error of 2 dB for frequencies up to 4 kHz is acceptable, the probe needs to be within 9 mm of the eardrum. The probe tube can be accurately positioned at this distance by extending it for a further 6 mm from the location of the 6 kHz minimum. A less desirable, but sometimes more practical, method of probe tube positioning with less compliant children is to use age-appropriate fixed insertion depths. The recommended probe tube insertion depths are 15 mm past the inter-tragal notch for babies under 12 months, 20 mm for children 1-5 years old, and 25 mm for older children.9 These guidelines must be used in conjunction with otoscopic observation during insertion, so that the location of the probe tip relative to the eardrum could be carefully monitored and varied if required.10
The NAL-NL1 software allows individually measured RECDs to be used for calculating the coupler response required for the prescribed REAG target. If measured values are not available, the software calculates the required coupler response based on age-appropriate RECD values according to the childs date of birth entered into the program.
Detailed below is a step-by-step procedure for fitting hearing aids to NAL-NL1 targets for hearing-impaired children using an REAG approach. Two case studies are also presented. Finally, data is presented on the goodness-of-fit achieved for a group of children using a two-channel non-linear hearing aid.
Using NAL-NL1 Software with Children: Step by Step
Step 1. Measure hearing threshold levels: Insert earphones can be connected to a childs individual earmold for measuring hearing thresholds in terms of real-ear SPL (the rationale for doing this is well described in Seewald & Scollie11). The hearing thresholds can be entered into the NAL-NL1 software directly by selecting the appropriate transducer in the Audiological Input screen. The program also allows you to enter hearing thresholds in several forms, depending on how the thresholds were measured, and calculates the targets using appropriate correction factors.
Step 2. Measure real-ear-to-coupler difference: Detailed protocols for measuring RECD in children are available.7,9 Briefly, a real-ear analyzer is used with an insert earphone and a childs own earmold to measure RECD. The following steps assume that a BTE hearing aid is selected:
- The coupler part of the measurement is carried out by connecting the earphone to an HA2 coupler via 25 mm of tubing.
- The real-ear part of the measurement is carried out by inserting a probe tube and the individuals earmold into the ear. The difference between these two measurements equals the RECD.
The measured RECD values for each ear can be entered directly in the Audiological Input screen. If you did not measure RECD, ensure that you have entered the childs date of birth so that the software will use age-appropriate average RECD values.
Step 3. Derive NAL-NL1 coupler gain targets:
- In the Selection screen, select the type and number of hearing aids, the number of compression channels, and the earmold acoustics.
- In Target Type under Options, select Coupler response to match REAG target.
- In the Target screen, set one of the graphs to display REAG or 2cc Coupler Gain, and select the input levels you wish to display. For example, 55, 70, and 80 dB can be chosen to represent low, medium, and high input levels. Select Speech Level and Coupler Gain Targets if you are going to use a broadband signal for electro-acoustic verification in the coupler.
The parameter tables provide the compression threshold, compression ratio, and speech level gain targets at low, medium, and high input levels for each channel. In the second window, select the 2cc OSPL 90 curve to display the maximum output (OSPL 90) targets for NAL-NL1.
Step 4. Adjust the hearing aid in a coupler to achieve the targets:
- Connect a hearing aid to a programming device and a computer that contains the appropriate manufacturers software.
- Adjust the compression thresholds and compression ratios to approximate the prescribed values.
- Adjust the gain and the filter settings to match the prescribed frequency response for an average input level (70 dB SPL).
- Connect the hearing aid to a 2cc coupler and measure gain using a broadband stimulus at 70 dB to verify the gain targets for average level inputs. Adjust gain at different frequencies, if required.
- Measure the hearing aid gain for a low and high input level (for example, at 55 dB and at 80 dB) using a broadband stimulus in the test box, if desired. Adjust compression ratio if necessary.
- Measure the maximum power output of the hearing aid using a pure tone frequency sweep at 90 dB SPL input level. Adjust the maximum power output of the hearing aid to match the OSPL 90 targets.
Two cases are presented here to illustrate the application of the NAL-NL1 to fitting children with non-linear hearing aids. The first case is an older child being fitted with Bernafon Smile 110 BTE hearing aids in both ears, and the second case is a young child with bilateral severe hearing losses fitted with Siemens Prisma 2 SP+ BTE hearing aids. In each case, fitting of the left ear is shown relative to the childrens audiograms, derivations of the NAL-NL1 targets, the hearing aid adjustment process, and the verification of the fittings in a coupler.
Although many hearing aid fitting software systems include NAL-NL1 as a fitting rule, the current versions of the manufacturers software do not take account of the individuals RECD in calculating coupler gain targets. Therefore, the NAL-NL1 stand-alone software needs to be used for deriving individualized coupler gain targets. The fitting rule in the manufacturers software can be used as a basis, and further adjustments can be made to achieve the gains prescribed by the NAL-NL1 stand-alone software.
Case 1AM: AMs hearing threshold levels and RECDs were measured using insert earphones and her own earmolds. The thresholds were entered into the NAL-NL1 stand-alone software in the Audiological Input screen as real ear SPL levels by selecting Insert earphone+ own earmold, and dB SPL in ear canal. The measured RECDs were entered by selecting Measured own mold/HA2. Hearing threshold levels and RECDs measured for both ears were entered into the appropriate fields (Figure 1).
|Figure 1. Audiological input screen from the NAL-NL1 stand-alone program.|
In the Selection screen, BTE and Bilateral fitting were selected. The hearing aid selected for the fitting was a two-channel aid, so 2 was selected. An earmold with a 2 mm vent and standard #13 tubing was selected for this ear.
In the Target type field, found under Options, Coupler response matches REAG targets was selected. In the Target screen, Real-ear aided gain curves for input levels of 55, 70, and 80 dB were selected for display in the left graph, and the 2cc OSPL 90 in the right graph. We selected 2cc coupler and Speech levels to be displayed in the Parameter table below the graphs, as shown in Figure 2.
|Figure 2. Target screen showing real ear aided gain curves and 2cc OSPL 90 curves together with the targets for speech input expressed in 2cc coupler terms listed in the parameter table.|
The Bernafon BTE hearing aid was fitted to this ear. Using the manufacturers fitting software (Oasis on a NOAH platform), the audiogram was first entered, and the Smile 110 was selected. The hearing aid was connected via a HIPRO box to the computer.
In the Tuning screen in Oasis, NAL-NL1 was selected as the fitting algorithm, so that the hearing aid was preset to a first approximation of the NAL targets in the fitting software. The compression thresholds for the low and high frequency bands were set as prescribed by the NAL-NL1 stand-alone software, and the hearing aid was measured in a coupler using a speech-shaped noise presented at 70 dB. The measured gains were compared to the coupler gains prescribed by the NAL-NL1 stand-alone software that included the individuals RECD.
The gain-frequency response of the hearing aid was modified by adjusting the gains for different frequencies using the equalizer slides, and the overall gain controls for each band. The hearing aid was re-measured and further adjusted until the achieved gains closely matched the prescribed gains. This process was repeated for input levels of 55 and 80 dB SPL. Figure 3 shows the coupler gains prescribed by NAL-NL1 stand-alone software (represented by broken lines) and the achieved gains (represented by solid lines) for input levels of 55, 70, and 80 dB SPL, together with the hearing aid parameters used.
|Figure 3. Top: Targets prescribed for Client AM by the NAL-NL1 stand-alone program (broken lines) compared to achieved coupler gains (solid lines). Bottom: Hearing aid parameter settings used.|
At audiometric frequencies between 250 Hz and 4000 Hz, the achieved gains were within ±5 dB of the prescribed gain targets. Averaged across the audiometric frequencies, the rms error was 2.8 dB.
The teachers and parents of the child were interested in visualizing how much of the longterm average speech spectrum would be audible to the child via acoustic amplification when the hearing aid is worn. By putting the coupler gain measurements of the fitting into the Speech screen of the NAL-NL1 software, the amount of audible signal for different input levels can be displayed. Figure 4 shows the Speech-o-gram for speech at 70 dB SPL.
|Figure 4. The speech screen in the NAL-NL1 program showing achieved coupler gains at three input levels, together with a display of the amount of audible speech spectrum for an input of 70 dB SPL.|
Case 2SV: SV is 21 months old, and she has a bilateral sensorineural moderate-to-severe hearing loss. Her hearing thresholds were determined using insert earphones and her own earmolds. RECD values were not available initially, and age-appropriate values were selected. Figure 5 shows the Audiological Input screen with hearing thresholds of SV and predicted RECDs.
|Figure 5. Audiological input screen showing that age-appropriate RECD values were applied when no measurements were available.|
RECDs were measured using insert earphones and the childs earmold at a subsequent appointment, and these values replaced the predicted values for calculating targets using the stand-alone software. In the Selection screen, BTE and Bilateral fitting were selected. We chose to fit the Siemens Prisma 2SP+ to this child, so 4 was selected under No. of channels.
For Target type (found in the Options menu), Coupler response matches REAG targets was selected. In the Target screen, we selected Coupler gain targets and 2cc OSPL 90 for display. In the Parameter table below the graphs, 2cc coupler speech levels were displayed for each channel, as shown in Figure 6.
|Figure 6. Displays of 2cc coupler gain targets and 2cc OSPL 90 for a four-channel hearing aid.|
Using the manufacturers fitting software (Connex on a NOAH platform), the childs audiogram was entered and the NAL-NL1 rule was selected. Speech noise was selected as test stimulus. Under Fitting, New was selected, and two hearing aids were connected via a HIPRO in an Aurical to a computer. First fit was ticked, and the settings of the hearing aids were set to approximate the targets. We chose to display G(ain) rather than LO(output). In the Frequency shaping field, the channel delineators that were closest to the cross-over frequencies prescribed by NAL-NL1 were selected. In Compression, the compression thresholds and ratios (CK and CR) for each band were selected to approximate those prescribed by the stand-alone NAL-NL1 software. Multi-channel AGC-O was selected in the Output_More screen.
The hearing aid gain was measured in a 2cc coupler, using speech-shaped noise as stimulus. Prior to measurement, all adaptive features of the hearing aid (voice activity detection, microphone noise reduction, and cross-channel coupling) were switched off by selecting Fitting, test settings, adaptive parameters. This enables the frequency response of the hearing aid to be measured for broadband signals at different input levels in quiet.
The measured gains were compared to the coupler gain values prescribed by the NAL-NL1 stand-alone software. To achieve a close match to the prescribed targets, the gains for individual frequency bands were adjusted in the Frequency shaping screen, and compression thresholds and ratios were modified in the Compression screen. Figure 7 shows the coupler gains prescribed by NAL-NL1 stand-alone software (broken lines) and the achieved gains (solid lines) for input levels of 55, 70, and 80 dB SPL, together with the hearing aid parameters used.
|Figure 7. Top: Coupler gain targets prescribed for Client SV by the NAL-NL1 stand-alone program (broken lines) together with the achieved gains. Bottom: Hearing aid parameter settings.|
At audiometric frequencies (250-4000 Hz), the fitting was within ±4 dB of the gains prescribed by the NAL-NL1 stand-alone software. Averaged across the audiometric frequencies over the same range, the rms error was 2.2 dB.
The maximum output of the hearing aid was measured using a pure-tone frequency sweep, and minimal adjustments were required. This was done in the Output_More screen.
After the targets were satisfactorily verified, all adaptive parameters were activated by simply clicking Yes to re-program the hearing aids when prompted at the close of the dialogue window for test settings.
How Well Can NAL-NL1 Targets Be Achieved?
The following summarizes results from fitting Bernafon Smile 110 BTE hearing aids to 22 children using the procedures described above. The hearing aid is dual-channel, and the gain, compression threshold, compression ratio, and maximum power output for each channel can be independently adjusted using the manufacturers software.
The children ranged in age from 7 to 18 years. Their pure-tone thresholds were measured using insert earphones coupled to their own earmolds using the Audiometry module of the Aurical system. Table 1 gives the mean, range, and standard deviation of the hearing loss of these children.
|Table 1. Mean hearing threshold level (dB HL), range and standard deviation for the 43 test ears.|
Individual RECDs were measured for each ear via the childs own earmolds using the Real-ear Measurement module of the Aurical system. The 6 kHz probe tone method was used to position the probe tube 9 mm from the eardrum for RECD measurement. The mean insertion depth from the inter-tragal notch was 24.4 mm (range: 21-29 mm; standard deviation: 1.55 mm). The average RECD for this group is shown in Figure 8.
|Figure 8. Mean real-ear-to-coupler difference (RECD) values for 43 ears, with error bars indicating ±1 standard deviation. The dotted line shows adult average RECD values.|
To derive NAL-NL1 coupler gain targets, the individual hearing thresholds and RECDs were input into the NAL-NL1 stand-alone software as described above. A total of 43 ears (21 bilateral fittings and 1 unilateral fitting) were fitted with hearing aids.
To verify the fittings, the hearing aids were measured in a 2cc coupler using the Hearing Instrument Testing module of the Aurical system. The test signal was a speech-shaped noise. Figure 9 shows the mean coupler gain targets prescribed by NAL-NL1 stand-alone software together with the mean coupler gain values achieved for all fittings for input levels of 55 dB, 70 dB, and 80 dB SPL.
|Figure 9. Coupler gain targets prescribed by the NAL-NL1 stand-alone program (circles) and achieved gains (triangles) averaged over 43 ears for low, medium, and high input levels.|
The maximum fitting error (ie, prescribed minus achieved gains) for a 70 dB input level for any child at any audiometric frequency (250-4000 Hz) was ±5 dB. Across all frequencies (250-4000 Hz), the root-mean-square error between the prescribed and the achieved gains was 2.3 dB.
Advances in hearing aid technology and increased knowledge in assessment and fitting methods have no doubt increased our ability to prescribe appropriate amplification targets and then meet those targets. There is, however, still much to learn about prescribing the optimal amplification characteristics for any child.12 Whereas we have direct evidence to support the use of NAL-NL1 for adults5, the evidence from children is less direct and more uncertain. The uncertainties are greater for children with severe hearing losses than those with milder losses.13,14 In a study funded by the Oticon Foundation, researchers at the NAL of Australia and the University of Western Ontario of Canada are jointly investigating the efficacy of alternative prescriptions, with the aim of improving prescriptions for children. It is expected that new information will be presented in this area over the next 12 months.
|Teresa YC Ching, PhD, (pictured) is senior research scientist, and Louise Britton, Harvey Dillon, PhD, and Katrina Agung are hearing scientists at the National Acoustic Laboratories (NAL) in Sydney, Australia.|
The authors would like to acknowledge the contributions of Susan D. Scollie and M. Jane Joyce, National Centre for Audiology, University of Western Ontario, in the development of the electroacoustic fitting protocols described in this paper.
Correspondence can be addressed to HR or Teresa Y C Ching, National Acoustic Laboratories, 126 Greville St, Chatswood, NSW 2067, Australia; email: [email protected].
1. Dillon H. Fitting a wide dynamic range of speech into a narrow dynamic range of hearing. In: Seewald RC, ed. A Sound Foundation Through Early Amplification. Warrenville, Ill: Phonak Inc, 2000:65-76.
2. Dillon H. NAL-NL1: A new procedure for fitting non-linear hearing aids. Hear Jour 1999;52(4):10-16.
3. Byrne D, Dillon H, Ching TYC, et al. The NAL-NL1 procedure for fitting non-linear hearing aids: Characteristics and comparisons with other procedures. JAAA. 2001; 12:37-51.
4. Ching TYC, Dillon H, Katsch R, Byrne D. Maximising effective audibility in hearing aid fitting. Ear Hear. 2001;22(3): 212-224.
5. Keidser G, Grant F. Comparing loudness normalization (IHAFF) with speech intelligibility maximization (NAL-NL1) when implemented in a two-channel device. Ear Hear. 2001;22(6):501-515.
6. Seewald R. The Desired Sensation Level (DSL) method for hearing aid fitting in infants and children. Phonak Focus. 1995; 20:3-18.
7. Dillon H. Hearing Aids. New York City: Thieme; 2001.
8. Seewald RC, Moodie KS, Sinclair ST, Scollie S. Predictive validity for a procedure for paediatric hearing instrument fitting. AJA. 1999;8(2): 143-152.
9. Moodie KS, Seewald RC, Sinclair ST. Procedure for predicting real-ear hearing aid performance in young children. American Journal of Audiology. 1994;3(1):23-31.
10. Scollie SD, Seewald RC, Cornelisse LE, Jenstad LM. Validity and repeatability of level-independent HL to SPL transforms. Ear Hear. 1998;19:407-413.
11. Seewald RC, Scollie S. Infants are not average adults: Implications for audiometric testing. Hear Jour 1999;52(10): 64-72.
12. Ching TYC. Effective amplification for hearing-impaired children. Hear Jour 2002;55(4):10-18.
13. Ching TYC, Dillon H, Katsch R. Do children require more high-frequency audibility than adults with similar hearing losses? In: Seewald RC, Gravel JS, eds. A Sound Foundation through Early Amplification 2001. Proc. 2nd International Conference. Phonak Inc. Warrenville, Ill, 2001:141-152.
14. Stelmachowicz PG. The importance of high frequency amplification for young children. In: Seewald RC, Gravel JS, eds. A Sound Foundation through Early Amplification 2001. Proc. 2nd International Conference. Phonak AG. Warrenville, Ill, 2001: 167-175.