Prescriptive procedures are used in hearing aid fittings to select an appropriate amplification characteristic based on measurements of the auditory system. The advantages of using a prescriptive procedure as opposed to an evaluative or other approach are (1) they can be used with all populations in a time-efficient way, (2) they help the clinician select a suitable parameter combination from among an almost unlimited number possible in modern hearing aids and settings, and (3) they can be verified. On the negative side, there is little interaction with the client when fitting hearing aids according to a prescriptive procedure, and any two people with the same type of loss may have different preferences for the loudness and the tone of sounds.

More than fifty prescriptive procedures for fitting hearing aids have been presented. The procedures vary with respect to the type of amplification characteristic that is prescribed, the type of data the procedure is based on, and the aim of the procedure.

The parameter most commonly prescribed in hearing aids is gain as a function of frequency. In linear devices, only one gain/frequency response is prescribed, which applies to all input levels that do not cause the hearing aid to saturate. If the hearing aid is nonlinear (contains compressor amplifiers), gain varies as a function of both frequency and input level. In that case, gain/frequency curves are prescribed for different input levels, or the static compression parameters are prescribed for selected frequencies. To avoid excessive loudness when listening to high-intensity input levels, the maximum output of the hearing device must also be prescribed.

Some procedures prescribe the amplification characteristic based on threshold levels only. Others use suprathreshold loudness judgments, such as the most comfortable level (MCL), the loudness discomfort level (LDL), or the entire loudness scale. Supporters of threshold-based procedures argue that loudness data are difficult to measure and unreliable, especially in children and special populations, and that preferred gain and maximum output can be adequately predicted from threshold levels. The argument for using individually measured loudness data is that the fitting will be more accurate because hearing aid users with the same audiogram can perceive loudness differently. Table 1 lists some of the most widely used prescription procedures developed to date. They are categorized according to which parameters are prescribed and the data used.

Most procedures for fitting linear devices share the general aim of amplifying speech presented at an average level to a comfortable level situated approximately halfway between threshold and LDL. The rationale is that such a response provides optimum speech understanding and comfortable listening in general situations. Despite this common rationale, the assumptions and underlying operational principles behind each procedure vary, producing very different formulas. The assumptions presented include the following: (1) The audibility of all speech components is important for speech understanding (e.g., DSL; Seewald, Ross, and Spiro, 1985). (2) Speech discrimination is highest when speech is presented at levels above MCL (e.g., MSU; Cox, 1988). (3) Speech is best understood when speech bands at different frequencies have equal loudness (e.g., NAL-R; Byrne and Dillon, 1986; and CID; Skinner et al., 1982). (4) For hearing aid users with mild to moderate losses, speech presented at average input levels is restored to the MCL when providing gain equal to about half the amount of threshold loss (e.g., Berger, Hagberg, and Rane, 1977; and POGO; McCandless and Lyregaard, 1983).

Some of these procedures take the shape of the speech spectrum into consideration when prescribing gain at each frequency (DSL, MSU, CID, and NAL-R), whereas others introduce a reduction in the low-frequency gain to avoid upward spread of masking from low-frequency ambient noise (Berger, POGO). Either way, the net result is that, even for a flat hearing loss, less gain is prescribed in the low than in the high frequencies. The NAL-R procedure differs from the other linear procedures in two respects. First, the gain prescribed at any frequency is affected by the degree of loss at other frequencies. Second, it is the only procedure that is well supported by direct empirical data (e.g., Byrne and Cotton, 1988).

One procedure for fitting nonlinear devices, NAL-NL1 (Dillon, 1999), follows the common rationale of procedures for fitting linear devices by aiming at maximizing speech intelligibility for any input level. To avoid amplifying all input levels to a most comfortable level, which probably would make the loudness of environmental sounds unacceptable, the rationale uses the constraint that for any input level, the overall loudness of speech must not exceed normal loudness. This procedure prescribes gain/frequency responses that make the speech bands approximately equal in loudness (Fig. 1), which is in agreement with several procedures for fitting linear devices. As hearing loss at any frequency becomes severe or profound, the ear becomes less able to extract information, even when the signal in that frequency region is audible (Ching, Dillon, and Byrne, 1998). Consequently, the goal of achieving equal loudness is progressively relaxed within the NAL-NL1 rule as hearing loss increases.

Figure 1..  

Graph illustrating loudness perception of speech when the interfrequency variation of intensity for speech is maintained (loudness normalization) and when the intensity of speech bands has been equalized to maximize speech intelligibility.

The rationale behind most nonlinear prescription procedures, however, is loudness normalization. Examples are LGOB (Humes et al., 1996), FIG6 (Killion and Fikret-Pasa, 1993), IHAFF (Cox, 1995), DSL[i/o] (Cornelisse, Seewald, and Jamieson, 1995), and ScalAdapt (Kiessling, Schubert, and Archut, 1996). The assumption behind this rationale is that “normal hearing” is best for speech understanding and for listening to environmental sounds. Loudness normalization is achieved by applying the gain needed to make narrow-band stimuli of any input level just as loud for the impaired ear as they are for normal ears. This rationale maintains the interfrequency variation of loudness that normally occurs for speech (Fig. 1). Consequently, loudness normalization is not consistent with the principles of equalizing loudness across frequency and deemphasizing loudness at those frequencies where loss is greatest, principles that have emerged from research into linear amplification.

Because of the different assumptions and principles used by the various procedures, different procedures prescribe different amplification characteristics for the same type of hearing loss (Figs. 2–5). The differences are more pronounced for flat and reverse sloping loss than for the more common high-frequency sloping loss.

Figure 2..  

Audiogram and prescribed insertion gain curves for a person with a moderate flat hearing loss. For IHAFF, the targets are calculated based on average threshold-dependent loudness data (Cox, personal communication). Note that the NAL-NL1 rule does not prescribe insertion gain at frequencies where it is doubtful that amplification will contribute to speech intelligibility.

Figure 3..  

Audiogram and prescribed insertion gain curves for a person with a moderate to severe low-frequency hearing loss. For IHAFF, the targets are calculated based on average threshold-dependent loudness data (Cox, personal communication). Note that the NAL-NL1 rule does not prescribe insertion gain at frequencies where it is doubtful that amplification will contribute to speech intelligibility.

Figure 4..  

Audiogram and prescribed insertion gain curves for a person with a gently sloping high-frequency hearing loss. For IHAFF, the targets are calculated based on average threshold-dependent loudness data (Cox, personal communication). Note that the NAL-NL1 rule does not prescribe insertion gain at frequencies where it is doubtful that amplification will contribute to speech intelligibility.

Figure 5..  

Audiogram and prescribed insertion gain curves for a person with a steeply sloping high-frequency hearing loss. For IHAFF, the targets are calculated based on average threshold-dependent loudness data (Cox, personal communication). Note that the NAL-NL1 rule does not prescribe insertion gain at frequencies where it is doubtful that amplification will contribute to speech intelligibility.

A recent evaluation of loudness normalization versus speech intelligibility maximization suggests that when the difference in prescription between the two rationales is substantial, hearing aid users prefer and perform better with the speech intelligibility maximization rationale (Keidser and Grant, 2001).

The gain/frequency curves may be prescribed according to the acoustic input the client is likely to experience. Simple variations applied to the amplification characteristic prescribed to compensate for the hearing loss have proved useful for compensating for defined changes in the acoustic input (Keidser, Dillon, and Byrne, 1996). Such variations can be programmed into different memories in a multimemory hearing aid.

Some procedures also prescribe the maximum output of the hearing aid known as the saturation sound pressure level (SSPL). It is important to have the output level of the hearing aid correctly adjusted. If the SSPL is too high, the hearing aid can cause discomfort or damage to the hearing aid user. On the other hand, if the SSPL is too low, the hearing aid user may experience insufficient loudness and excessive saturation. Most procedures that prescribe SSPL aim at avoiding discomfort. In those cases the SSPL is set equal to or just below the hearing aid user's discomfort level; examples are CID, MSU, POGO, and IHAFF. Only one procedure, NAL-SSPL, considers both the maximum output level and the minimum output level and prescribes a level halfway between these two extremes (Dillon and Storey, 1998). This is also the only procedure for prescribing the output level that has been experimentally evaluated. It was found to provide an SSPL that did not require fine-tuning for 80% of clients (Storey et al., 1998).

Many prescription procedures target a sensorineural loss of mild to moderate degree. Appropriate adjustments to the prescriptions may be needed if prescribing amplification for clients with a conductive component (Lybarger, 1963), and a severe to profound loss (POGO II: Schwartz, Lyregaard, and Lundt, 1988; NAL-RP: Byrne, Parkinson, and Newall, 1990). Some adjustments are also needed for clients who are fitted with one versus two hearing aids (Dillon, 2001).

When the hearing aid has been adjusted according to the prescriptive procedures, the setting can be verified against the prescribed target, either in a hearing aid test box or in the real ear. Verifying the prescriptive parameters in the real ear allows individual configurations of the ear canal and the acoustic coupling between ear and hearing aid to be taken into consideration. For some clients fine-tuning may be needed after the client has tried the hearing aid in everyday listening environments.

The most commonly used prescription procedures are readily available in an electronic format, either as specifically designed computer programs, in programs provided by hearing aid manufacturers for fitting their programmable devices, or in equipment for measuring real-ear gain.

See also hearing aid fitting: evalution of outcomes; hearing aids: sound quality.