A comparison of manufacturer-specific prescriptive procedures for infants By Richard Seewald, Jillian Mills, Marlene Bagatto, Susan Scollie, and Sheila Moodie Early hearing detection and communication development a function of frequency. Prescriptive algorithms provide (EHDCD) programs enable us to identify hearing loss in the foundation on which the hearing instrument perforearly infancy, making intervention by 6 months of age a mance characteristics are selected for the infant or child to reasonable goal. For infants and children fitted with hearbe fitted. Recent survey data indicate that most audioloing instruments, it is critical that the devices be adjusted gists fitting hearing aids to infants and young children use appropriately so that the pre-lingual listener has full and a published hearing aid prescriptive procedure.6,9 At preconsistent access to sound, without discomfort for loud sent, prescriptive procedures can be divided into two classounds. sifications: (1) evidence-based generic prescriptive algorithms Several published pediatric amplification guidelines and (2) manufacturer-specific proprietary algorithms. are available to assist clinicians with the hearing aid fitEvidence-based generic prescriptive algorithms for chilting process for infants and children.1-5 While the guidedren include the DSL m[i/o]10 and the NAL-NL1,11 both lines were developed in different jurisdictions, similar of which are intended for use with any hearing aid. recommendations are made for all stages of the hearing Proprietary prescriptive methods incorporate prescriptive instrument fitting process. For example, the use of a sysalgorithms developed by manufacturers for use with their tematic, evidence-based approach to hearing aid fitting specific hearing instruments. The lack of published inforthat accounts for an infant’s external ear acoustics at the mation regarding the development of pediatric manufacassessment, selection, and turer-specific proprietary verification stages and that algorithms makes it diffi“…an infant can end up with includes the measurement cult for the audiologist to of hearing aid output limmake an informed choice substantially different hearing iting is a requirement of about which procedure every current pediatric will be best for their young aid performance depending amplification protocol. patient.12 Some manufacWhile it is encouraging turers have decided to on which prescriptive to see consensus in pediforego the development of atric hearing aid fitting a proprietary pediatric algorithm is chosen..." guidelines, there continue algorithm and use either to be substantial differthe NAL or DSL prescripences among clinical practice behaviors. Surveys have inditive algorithms for fitting children and have implemented cated that the majority of pediatric audiologists use sound these formulas in their fitting software. These generic prefield-aided threshold measures to verify the electroacoustic scriptive algorithms may have been adapted by the manperformance of hearing instruments for young infants.6 ufacturer to account for the unique characteristics (e.g., Probe-microphone measures were used by roughly 20% compression kneepoint) of their hearing aids. of the audiologists fitting hearing aids to infants.6 It has Several investigators have compared proprietary prebeen well-documented that behavioral measures (i.e., aided scriptive algorithms for use with adults. Keidser and colthreshold measures) do not provide accurate estimates of leagues found a 10-dB variation in the amount of gain the aided audibility of soft, average, and loud speech or prescribed by NAL-NL1, DSL[i/o], and four proprietary the real-ear saturation response of the hearing aid and that algorithms for the same audiogram.12 A study by Hawkins infants 6 months of age and younger are not developmenand Cook compared simulated 2-cc versus actual 2-cc gain tally capable of performing this task.7,8 Failure to approas well as simulated insertion versus actual insertion gain priately verify the electroacoustic performance of the hearing measures from the fitting software for 28 hearing aids of aid in terms of predicted speech audibility and maximum various styles from four major manufacturers.13 Results hearing instrument output can result in obstructing the indicated a clear trend for the simulated values to overeslanguage benefits an infant would have otherwise received timate what was actually provided by the hearing aids for from being identified at an early age and optimally fitted. both the 2-cc coupler and insertion gain comparisons.13 It is also important to consider the prescriptive targets Differences of as much as 10 to 20 dB were noted for the used to determine the recommended levels for amplified various hearing aids and prescriptive algorithms in the speech and target limits for hearing instrument output as study. 26 THE HEARING JOURNAL Pe d i a t r i c A m p l i f i c a t i o n : A s p e c i a l i s s u e NOVEMBER 2008 • VOL. 61 • NO. 11 A more recent study examined behindthe-ear (BTE) hearing aids from six major manufacturers programmed using their “default” fitting procedure.14 For some this was a generic fitting formula and others applied their own proprietary prescriptive method. The OSPL90 was measured at the default program settings with all special features deactivated. Results indicated maximum output values ranging from 90 to 109 dB SPL (re: 2-cc coupler) at 2000 Hz across the six hearing aids for a flat 50-dB-HL hearing loss.14 RATIONALE Despite the unexpected degree of variation among prescriptive methods, for everyday listening the average adult listener may be able to adjust the volume control of the hearing instruments to compensate for the 10- or 20-dB variation provided from one algorithm to the next. However, in the case of a 6-month-old infant who can neither manipulate the gain of the hearing aid nor verbalize concerns, these recent findings would seem to be clinically important. Infants and young children must live and learn with the hearing instrument characteristics provided (or not provided) to them. They cannot accommodate a 10-dB variation in prescribed overall gain by adjusting a volume control wheel as an adult would. They also cannot “fill in the blanks” when words or speech sounds are inaudible the way adults do because they have not yet developed sufficient knowledge of spoken language. The results reported above make it apparent that for the same audiometric hearing loss there may be differences in the prescribed gain from various prescriptive procedures. In addition, it may be that there is some variation in the gain prescribed from manufacturer-specific implementations of the same generic prescriptive procedure. As a result, an infant can end up with substantially different electroacoustic characteristics in hearing aid performance depending on which prescriptive algorithm is chosen by the audiologist within the hearing aid fitting software. It is important to systematically quantify the differences that may exist among manufacturer-specific implementations of prescriptive algorithms that potentially affect audibility and comfort of amplified speech. 28 THE HEARING JOURNAL hearing aids’ responses in the ear canal.15 Nine audiograms, ranging from mild to profound, were used to program each hearing aid (see Table 1). Simulated real-ear measurements (S-REM) were performed for soft (55 dB SPL), average (65 dB SPL), and loud (75 dB SPL) speech and maximum power output (MPO), using the Audioscan® Verifit VF-1 real-ear measurement system (Software Version 2.0.18). Each BTE instrument was coupled to the standard HA-2 coupler available on the VF-1 and placed inside the test chamber. The test chamber was closed prior to making all electroacoustic measurements. The equipment (i.e., cables, programming boots) supplied by the manufacturers specific to each BTE instrument was connected to a HI-PRO box for programming each hearing instrument. For each BTE instrument programming, the 1/3-octave band measurements obtained from the VF-1 were collapsed into a single Speech Intelligibility Index (SII) value, based on the 1/3-octave band SII calculation procedure.16 The SII value ranges from 0 to 1, and was converted to a value between 0% and 100% for the purposes of this study. A Microsoft Excel spreadsheet was used for all SII calculations.17 Separate SII values were calculated for each BTE instrument across the nine audiograms and for each of the three input levels. The nine audiograms were converted to equivalent adult hearing levels18 before SII calculations were performed. It has been documented that young children are not able to use RESEARCH QUESTIONS This study was designed to address three questions: (1) Will the prescribed output-limiting levels differ for the same 6-month-old infant depending on the manufacturerspecific prescriptive algorithm chosen? (2) Will the prescribed gain by frequency characteristics differ for the same 6-monthold infant, as a function of input level, depending on the manufacturer-specific algorithm chosen? (3) Will predicted speech audibility differ across manufacturer-specific prescriptive algorithms evaluated? METHOD BTE instruments recommended for use with the pediatric population were chosen from five manufacturers. The hearing aids were programmed using the most current version of each of the manufacturer’s proprietary fitting software, installed as modules in the NOAH® client module software system. Where proprietary prescriptive algorithms for pediatric fittings were not available, three of the BTE instruments were programmed using the manufacturer’s default prescriptive method. In two cases this was the manufacturer’s implementation of the NAL-NL1 procedure and in another case this was the manufacturer’s implementation of the DSL method. Advanced features (i.e., noise reduction, directional microphone) were deactivated in all devices. Average realear-to-coupler difference (RECD) values for a 6-month-old infant were applied to 2-cc coupler measures to predict the Table 1. Nine audiograms (A-I) in dB HL, used in this study. Frequency (Hz) Audiogram 250 500 750 1000 1500 2000 3000 4000 6000 A 30 30 35 35 40 40 45 50 50 B 30 30 35 40 45 50 60 70 70 C 30 30 35 45 50 60 80 90 90 D 50 50 50 50 50 50 50 50 50 E 50 50 50 55 55 60 65 70 70 F 50 50 55 55 65 70 80 90 90 G 70 70 70 70 70 70 70 70 70 H 70 70 70 75 75 80 85 90 90 I 90 90 90 90 90 90 90 90 90 Pe d i a t r i c A m p l i f i c a t i o n : A s p e c i a l i s s u e NOVEMBER 2008 • VOL. 61 • NO. 11 contextual information in the same way as adults.19 Thus, an equal band importance function was used in calculating the SII values. 150 140 Simulated real-ear saturation response (S-RESR) measurements differed significantly across the five manufacturerspecific prescriptive algorithms (F[4,32] = 47.66, p<.001). Post hoc analyses indicated that this effect was present per frequency (Tukey’s HSD, p<0.05). The results of the S-RESR measurements for the five manufacturer-specific algorithms at 3000 Hz for the nine audiograms are presented in Figure 1. Substantial variation is evident in the simulated RESR measurements across manufacturers for the nine audiograms used in this study. A review of the data presented by individual audiograms further revealed the significant variation in the output being produced by some of the manufacturer-specific prescriptive algorithms. For example, for audiogram A, values range from 100 dB SPL (Manufacturer 4) to 130 dB SPL (Manufacturer 1) in the higher frequencies. In the lower frequencies, values range from 90 dB SPL (Manufacturer 3) to 105 dB SPL (Manufacturer 4) (see Figure 2). Moreover, the shapes of the output limiting responses are different for each manufacturer. Some responses are relatively flat across frequencies (Manufacturer 4), while others are not (Manufacturer 1) (see Figure 2). Determining the basis for manufacturer-specific prescriptive algorithm output limiting targets is difficult, as the derivation of these algorithms is often neither clearly described nor accessible. However, from the present study, it can be seen that there is a large range of variation among the S-RESR for the manufacturerspecific prescriptive algorithms evaluated in this study. Clearly the output limits of some manufacturer-specific prescriptive algorithms may be inappropriate for pediatric hearing aid fittings. Prescribed gain by frequency for different input levels Simulated real-ear aided response (SREAR) measurements for soft, average, and loud speech inputs differed 30 THE HEARING JOURNAL 120 Manufacturer 1 110 Manufacturer 2 100 Manufacturer 3 Manufacturer 4 90 Manufacturer 5 80 70 60 Audiogram An example of S-RESR measurements for five manufacturer-specific algorithms at 3000 Hz for the nine audiograms used in this study. Figure 1. 150 140 130 S-RESR (dB SPL) Output limiting S-RESR (dB SPL) 130 RESULTS AND DISCUSSION Manufacturer 1 120 Manufacturer 2 110 Manufacturer 3 100 Manufacturer 4 Manufacturer 5 90 80 70 60 100 1000 10000 Frequency (Hz) Figure 2. S-RESR measurements obtained for the five manufacturer-specific algorithms using Audiogram A (gently sloping, mild to moderate hearing loss). significantly across manufacturers: soft inputs (F[4,32]=18.11, p<0.001); average inputs (F[4,32]=24.67, p<0.001); loud inputs (F[4,32]=31.89, p<0.001). Post hoc analyses indicated that these differences were significant at most test frequencies (Tukey’s HSD, p<0.05). An example is shown for Audiogram E (Figure 3, A-C). It can be seen that, in this example, S-REAR values differ by as much as 20 dB at some frequencies for soft, average, and loud inputs. Predicted speech audibility Substantial differences among SII values were noted from the various manufacturer-specific prescriptive algorithms’ responses for soft speech input levels. Relatively less difference in these values was Pe d i a t r i c A m p l i f i c a t i o n : A s p e c i a l i s s u e noted for average level inputs and even smaller differences for loud input levels (Figure 4 A-C). It is evident that the observed differences in frequency response characteristics of the manufacturer-specific algorithms resulted in differences in SII values. Currently, there is no way to predict real-life outcome differences from SII value differences for a 6-month-old infant. However, from a habilitative audibility perspective, increased audibility should result in more languagelearning opportunities. Research has shown that children require and prefer audibility of the entire speech range and at higher sensation levels than adults in order to understand speech.20,21 Therefore it is prudent to consider that more NOVEMBER 2008 • VOL. 61 • NO. 11 100 1000 Frequency (Hz) 10000 Manufacturer 1 104 103 102 101 100 90 80 70 60 50 40 100 b) S-REAR (dB SPL) a) S-REAR (dB SPL) S-REAR (dB SPL) 140 130 120 110 100 90 80 70 60 50 40 1000 Frequency (Hz) 10000 Manufacturer 3 Manufacturer 2 140 130 120 110 100 90 80 70 60 50 40 c) 100 Manufacturer 4 1000 Frequency (Hz) 10000 Manufacturer 5 S-REAR measurements obtained for (a) soft (55 dB SPL), (b) average (65 dB SPL), and (c) loud (75 dB SPL) input levels across the five manufacturer-specific algorithms using Audiogram E (gently sloping moderate to moderately severe). Figure 3. 100 100 100 b) c) 80 80 60 60 60 40 SII (%) 80 SII (%) SII (%) a) 40 20 20 20 0 40 0 A B C D E F Audiogram G H A I Manufacturer 1 B Manufacturer 2 C D E F Audiogram Manufacturer 3 G H I Manufacturer 4 0 A B C D E F Audiogram G H I Manufacturer 5 Predicted SII values for (a) soft (55 dB SPL), (b) average (65 dB SPL), and (c) loud (75 dB SPL) input levels for the five manufacturer-specific algorithms for the nine audiograms used in this study. Figure 4. audibility is desirable for the pediatric population. Clinical implications Despite the published recommendations for pediatric amplification, real-ear or simulated real-ear verification of hearing instrument performance is not routine clinical practice for many pediatric audiologists.6,9 With the recent proliferation in EHDCD programs, more infants are being fitted with hearing aids by 6 months of age. The current study has demonstrated the substantial variation that is generated among manufacturer-specific prescriptive algorithms for the same 6-month-old infant. Some manufacturer-specific algorithms resulted in unexpectedly high output limiting values, generating concern about the potential for discomfort in a 6-month-old fitted with these algorithms. Many of the manufacturer-specific prescriptive algorithm output-limiting values were substantially higher than 32 THE HEARING JOURNAL output-limiting values recommended by published evidence-based generic procedures for output limiting in pediatric hearing aid fittings.10,11 Additionally, the variation in manufacturer-specific prescriptive algorithm hearing aid response characteristics to speech inputs resulted in substantial differences in predicted audibility as measured by the SII. It is important in any pediatric hearing aid fitting to ensure that speech inputs are audible in order to facilitate speech and language learning. The results of this study suggest that the manufacturer-specific prescriptive algorithms examined provide very different simulated real-ear responses even for the same infant. This research highlights the importance of verifying and comparing the output limiting and responses to speech inputs provided by various manufacturer-specific implementations of prescriptive algorithms. Regardless of the approach taken for electroacoustic prescription, the responsible clinician should want to know the Pe d i a t r i c A m p l i f i c a t i o n : A s p e c i a l i s s u e levels of amplified sound that hearing aids deliver to the child’s ear. Therefore, pediatric audiologists must apply comprehensive and evidence-based electroacoustic verification strategies that are compatible with the population they are working with. Acknowledgment This work was supported by the Masonic Help-2Hear Foundation of Ontario. Richard C. Seewald, PhD, is Distinguished University Professor at the Child Amplification Laboratory, National Centre for Audiology, University of Western Ontario. Jillian Mills, MSc, is an Audiologist at ListenUP! Canada. Marlene Bagatto, AuD, is a Research Audiologist; Susan Scollie, PhD, is Assistant Professor; and Sheila Moodie, MClSc, is a Research Audiologist, all at the Child Amplification Laboratory, National Centre for Audiology. Readers may contact Dr. Seewald at seewald@nca.uwo.ca or at National Center of Audiology, Elborn College Building, Western Road, University of Western Ontario, London, Ontario, Canada N6G 1H1. REFERENCES 1. American Academy of Audiology: Pediatric Amplification Protocol. Reston, VA: American Academy of Audiology, 2003. Retrieved July 25, 2008 from NOVEMBER 2008 • VOL. 61 • NO. 11 www.audiology.org/NR/rdonlyres/53D26792-E32141AF-850FCC253310F9DB/0/pedamp.pdf. 2. College of Audiologists and Speech-Language Pathologists of Ontario. Preferred practice guideline for the prescription of hearing aids to children. 2002. Retrieved July 25, 2008, from www.caslpo.com/Portals/0/ppg/ preshearingaidschild.pdf. 3. Ministry of Children and Youth Services, Ontario, Canada: Ontario Infant Hearing Program provision of amplification protocol and guidelines. 2007. Retrieved July 25, 2008, from www.mountsinai.on.ca/ care/infant-hearing program/documents/amplification revision2007_006.pdf 4. The Pediatric Working Group of the Conference on Amplification for Children with Auditory Deficits: Amplification for infants and children with hearing loss. AJA 1996;5(1):53-68. 5. Guidelines for the Fitting, Verification and Evaluation of Digital Signal Processing Hearing Aids within a Children’s Hearing Aid Service. Modernising Children’s Hearing Aid Services, University of Manchester, England; 2005; 1-3. Retrieved August 1, 2008, from www.psych-sci.manchester.ac.uk/mchas/guidelines/ fittingguidelines.doc. 6. Bamford J, Beresford, D Mencher G, et al.: Provision and fitting of new technology hearing aids: Implications from a survey of some “good practice services” in UK and USA. In Seewald RC, Gravel JS, eds. A Sound Foundation Through Early Amplification: Proceedings of an International Conference. Stäfa, Switzerland: Phonak AG: 2001:213-219. 7. Stelmachowicz P, Hoover B, Lewis D, Brennan M: Is functional gain really functional? Hear J 2002;55(11): 38-42. 8. Hawkins D: Limitations and uses of the aided audiogram. Sem Hear 2004;25(1):51-62. 9. Tharpe AM: Service delivery for children with multiple impairments: How are we doing? In Seewald RC, ed., A Sound Foundation Through Early Amplification: Proceedings of an International Conference, Stäfa, Switzerland: Phonak AG, 2000: 175-190. 10. Scollie S, Seewald R, Cornelisse L, et al.: The Desired Sensation Level Multistage Input/Output Algorithm. Trends Amplif 2005;9(4):159-197. 11. Dillon H, Byrne D, Brewer S: NAL Non-linear (NAL NL-1) User Manual, version 1.01. National Acoustic Laboratories: Chatswood NSW, Australia, 1998. 12. Keidser G, Brew C, Peck A: How proprietary fitting algorithms compare to each other and to some generic algorithms. Hear J 2003;56(3):28-38. 13. Hawkins D, Cook J: Hearing aid software predictive gain values: How accurate are they? Hear J 2003;56(7): 26-34. 14. Mueller HG, Bentler R, Wu Y: Prescribing maximum hearing aid output: Differences among manufacturers found. Hear J 2008;61(3):30-36. 15. Bagatto M, Scollie S, Seewald R, et al.: Real-ear-tocoupler difference predictions as a function of age for two coupling procedures. JAAA 2002;13:407-415. 16. American National Standards Institute: American National Standard Methods for Calculation of the Speech Intelligibility Index (ANSI S3.5-1997). New York; ANSI, 1997. 17. Scollie S: Children’s speech recognition scores: The speech intelligibility index and proficiency factors for age and hearing level. Ear Hear 2008;29(4):543-556. 18. Dillon H: Hearing Aids. New York: Thieme, 2001. 19. Nittrouer S, Boothroyd A: Context effects in phoneme and word recognition by young children and older adults. J Acoust Soc Amer 1990;87(6):2705-2715. 20. Nozza R, Rossman R, Bond L: Infant-adult differences in unmasked thresholds for the discrimination of consonant-vowel syllable pairs. Audiology 1991;30: 102-112. 21. Bagatto MP, Moodie ST, Scollie SD, et al.: Clinical protocols for hearing instrument fitting in the Desired Sensation Level Method. Trends Amplif 2005;9(4):199-226. 34 THE HEARING JOURNAL Pe d i a t r i c A m p l i f i c a t i o n : A s p e c i a l i s s u e NOVEMBER 2008 • VOL. 61 • NO. 11