Auditory Brainstem and Middle Latency Responses Measured Pre- and Posttreatment for Hyperacusic Hearing-Impaired Persons Successfully Treated to Improve Sound Tolerance and to Expand the Dynamic Range for Loudness: Case Evidence

Abstract

In this report of three cases, we consider electrophysiologic measures from three hyperacusic hearing-impaired individuals who, prior to treatment to expand their dynamic ranges for loudness, were problematic hearing aid candidates because of their diminished sound tolerance and reduced dynamic ranges. Two of these individuals were treated with structured counseling combined with low-level broadband sound therapy from bilateral sound generators and the third case received structured counseling in combination with a short-acting placebo sound therapy. Each individual was highly responsive to his or her assigned treatment as revealed by expansion of the dynamic range by at least 20 dB at one or more frequencies posttreatment. Of specific interest in this report are their latency and amplitude measures taken from tone burst-evoked auditory brainstem response (ABR) and cortically derived middle latency response (MLR) recordings, measured as a function of increasing loudness at 500 and 2,000 Hz pre- and posttreatment. The resulting ABR and MLR latency and amplitude measures for each case are considered here in terms of pre- and posttreatment predictions. The respective pre- and posttreatment predictions anticipated larger pretreatment response amplitudes and shorter pretreatment response latencies relative to typical normal control values and smaller normative-like posttreatment response amplitudes and longer posttreatment response latencies relative to the corresponding pretreatment values for each individual. From these results and predictions, we conjecture about the neural origins of the hyperacusis conditions (i.e., brainstem versus cortical) and the neuronal sites responsive to treatment. The only consistent finding in support of the pre- and posttreatment predictions and, thus, the strongest index of hyperacusis and positive treatment-related effects was measured for MLR latency responses for wave Pa at 2,000 Hz. Other response indices, including ABR wave V latency and wave V-V′ amplitude and MLR wave Na-Pa amplitude for 500 and 2,000 Hz, appear either ambiguous across and/or within these individuals. Notwithstanding significant challenges for interpreting these findings, including associated confounding effects of their sensorineural hearing losses and differences in the presentation levels of the toneburst stimuli used to collect these measures for each individual, our limited analyses of three cases suggest measures of MLR wave Pa latency at 2,000 Hz (reflecting cortical contributions) may be a promising objective indicator of hyperacusis and dynamic range expansion treatment effects.

Learning Outcomes: As a result of this activity, the participant will be able to: (1) compare and contrast latency and amplitude measures estimated from tone-evoked auditory brainstem and middle latency responses recorded pre- and posttreatment from hyperacusic hearing-impaired persons who were successfully treated to expand their dynamic ranges for loudness; and (2) delineate central neuronal origins contributing to pretreatment hyperacusic conditions and those neuronal sites responsive to the intervention up to and including the auditory cortex.

Formby et al recently described a successful intervention to expand the dynamic range for loudness among failed hearing aid users and prospective hearing aid candidates who, prior to intervention, presented with reduced sound tolerance and diminished dynamic ranges.¹ An intriguing question at the onset of this dynamic range expansion trial was whether those study participants, who achieved large measurable treatment-related increases (≥20 dB) in their loudness judgments from their assigned interventions (i.e., structured counseling combined with either conventional or short-acting placebo sound generators), would demonstrate accompanying objective changes in electrophysiologic measures of auditory function. Specifically, one of the key secondary aims of the trial was to establish whether hallmark amplitude and latency measures recorded from the auditory brainstem response (ABR) and/or the middle latency response (MLR) changed concurrently with pronounced measurable changes in participants' loudness judgments from pre- to posttreatment. A more fundamental question of equal interest and importance was whether their pretreatment ABR and MLR measures would somehow be characteristically atypical, thus offering insights into the origins or nature of the measurable sound tolerance problems for the affected hearing-impaired participants.

A meager literature relevant to these questions presently exists. Most of what we know has been summarized in a recent review.² The reviewers concluded there is conflicting and insufficient evidence to establish relations between or among electrophysiologic indices and loudness judgments for normal-hearing listeners (see also Korczak et al in this issue).^{3 4 5 6} Moreover, there is virtually no evidence of electrophysiologic dysfunction and altered loudness perception as a consequence of hyperacusis, nor is there related evidence for the effects of treatments on the condition for hearing-impaired persons.

If we exclude unilateral treatment effects, which appear to give rise to a unique set of functional compensatory changes, and focus solely on bilateral treatments, which are of primary interest in this study, then based on the review of Fournier et al, the best available evidence of treatment effects on loudness judgments in hearing-impaired persons is that associated with a single study of binaural hearing aid use.^{2 7} In theory, binaural amplification should act as a source of environmental sound enrichment of the kind associated with the use of sound generators used in the treatment of tinnitus and hyperacusis.^{8 9} The objective of both forms of sound therapy is to reset (reduce) central auditory gain processes. Consistent with resetting of a central gain process, Philibert et al reported a reduction in the slope of the loudness growth function measured at a high audiometric frequency following 6 months of binaural amplification.⁷ However, no change in the loudness growth function was measured at a low frequency for which the amplification was weak for listeners with high-frequency sensorineural losses. These investigators reported a concurrent reduction in the click-evoked ABR wave V latency in the right ear, but not in the left ear.

The ambiguous evidence reported by Philibert et al of a relation between loudness and ABR change due to binaural hearing aid use is problematic on several levels.⁷ First, there is evidence that chronic bilateral hearing aid use has little or no effect (i.e., ≤3 dB) on loudness discomfort level (LDL) judgments,^{9 10 11 12} whereas sound therapy from bilateral low-level broadband sound generators, in combination with structured counseling (as described in this issue by Gold and Formby¹³), has a sizable incremental effect (≥10 dB) on LDL and categorical loudness judgments.^{1 14} Moreover, the main effects of long-term hearing aid exposure, in terms of change in categorical loudness judgments, are those usually associated with typical gain levels to which the aided user is primarily exposed.^{7 15 16} These gain levels correspond to judgments of moderately loud sounds, but not uncomfortably loud sounds. This aided exposure effect also is unlike the effects of low-level sound therapy from broadband sound generators. The latter characteristically reduces loudness the most for higher-level sounds, has systematically less effect on moderate sound levels, and has little or no effect on the loudness of lower-level sounds.¹ Furthermore, when prolonged changes in loudness judgments have been reported in response to long-term aided use, these effects have characteristically been specific to the frequencies of primary amplification exposure (i.e., usually restricted to higher frequencies where the hearing loss is often greatest and the most amplification is needed by the aided user).^{7 15 16} This aided frequency effect contrasts with that measured in response to bilateral low-level sound generators, which routinely affect loudness judgments for all audiometric frequencies more or less equally.¹ Accordingly, prolonged exposure effects to aided sound from bilateral hearing aids and those from bilateral sound generators appear to be characteristically different in terms of treatment-related changes in loudness judgments.⁹

There appears to be no electrophysiologic evidence yet reported for hyperacusis in the absence of tinnitus. There is, however, some limited evidence in terms of electrophysiologic responses measured from persons with tinnitus, who often self-report hypersensitivity to suprathreshold sound.^{8 17 18 19 20} These complaints may be accompanied by measures of reduced LDLs and steeper-than-normal loudness growth functions typical of hyperacusis.^{18 21} Such suprathreshold hypersensitivity has been reported in roughly 40 to 50% of persons with tinnitus,^{17 19} which, in turn, has been associated with a reduced click-evoked wave I amplitude in the ABR response and normal or greater wave V amplitudes.^{22 23}

Thus, this scant and often conflicting evidence reflects the fact that we know little about relations between and among electrophysiologic auditory measures, reduced sound tolerance, and diminished dynamic ranges in persons with sensorineural hearing losses. This review sets the stage and begs the need for an electrophysiologic study of higher-level auditory function of the kind highlighted in the following case evidence.

HYPOTHESES

The hypotheses addressed in this case evidence are: (1) the sources of the physiological mechanisms giving rise to loudness complaints associated with reduced sound tolerance and the restricted dynamic ranges of the hearing-impaired participants, and the processes that remediate these problems and contribute to the underlying treatment effects, reflect underlying neuronal gain processes and their response dynamics within the central auditory pathways^{24 25 26 27 28}; (2) ABR and MLR measures that assess auditory function up to and including the auditory cortex will reveal the sites of these gain mechanisms and their dynamic processes if these mechanisms have origins within these neuronal pathways; and (3) the problem gain mechanisms will be revealed pretreatment by increased response amplitudes and reduced response latencies in prominent waveforms of the ABR and/or MLR corresponding to the presumed sources of these waveforms and gain processes; subsequent to remediation, these same posttreatment amplitude and latency measures will be reduced and increased, respectively, reflecting treatment-induced recalibration of these central processes toward typical normal gain settings.

METHODS

Subjects

The findings we showcase here were measured pretreatment and over the course of a novel sound therapy protocol for dynamic range expansion from the right ears of three sound-challenged hearing-impaired participants.¹ We detailed the case evidence and described comprehensive response profiles for the left ears of two of these participants, designated 5T and 25T, in a companion report in this issue.¹⁴ Generally, their audiometric hearing losses were symmetric and their response profiles were similar for the right and left ears. Their right-ear response profiles are presented anew in the Appendix. The third participant, whom we designate here as 30T, was not considered in that report. His right-ear behavioral response profile also is presented in the Appendix for review.

The ABR and MLR findings for these individuals were selected for presentation in this report because each of these participants was demonstrably hyperacusic pretreatment, both in terms of their abnormally reduced LDLs and their very narrow dynamic ranges. Moreover, all three individuals ultimately proved to be highly responsive to their interventions. Specifically, each of these individuals achieved large treatment effects exemplified by bilateral shifts in LDLs and expansion of their dynamic ranges by greater than 20 dB at one or more frequencies. Accordingly, we reasoned these were the ideal candidates for revealing the origins of the underlying neurophysiological mechanisms contributing to their hyperacusis conditions and to those neuronal processes contributing to treatment remediation within the central auditory pathways up to and including the auditory cortex.

Following is a brief audiometric description of each subject, including pure tone thresholds and salient pre- and posttreatment uncomfortable loudness levels (UCLs) and dynamic range estimates shown in Fig. 1. Again, a comprehensive response profile for the right ear of each subject is presented in the Appendix.

Figure 1.

Audiometric measures for subjects 5T (left), 25T (middle), and 30T (right). The top panels display the air-conduction thresholds from 250 to 8,000 Hz as well as changes between pre- and posttreatment UCLs (shown by downward pointing arrow) for each subject. The bottom panels display their pre- and posttreatment dynamic ranges at 500 to 8,000 Hz. HL, hearing level; UCL, “uncomfortable loudness”.

Subject 5T

Subject 5T was 69 years of age at the time of this study. She presented with a bilaterally symmetric mild-to-moderate sensorineural hearing loss. Her audiometric thresholds and pre- and posttreatment UCLs and dynamic range data are presented in Fig. 1. Her pretreatment UCLs were in the range of 70 to 80 dB hearing level (HL) across frequency from 500 to 8,000 Hz. After 11 months of low-level sound therapy, including structured counseling for dynamic range expansion, ¹³ subject 5T's dynamic range expanded by ∼20 to 35 dB from 500 to 4,000 Hz. Consequently, posttreatment, her UCLs were at or above 100-dB HL, which are within the normal range of sound tolerance.

Subject 25T

Subject 25T was 58 years of age at the time of her participation. She presented with a mild sensorineural hearing loss in her right ear and a mild to moderate sensorineural hearing loss in her left ear. Her pure tone thresholds and pre- and posttreatment UCLs and dynamic range values are presented in Fig. 1. Her pretreatment UCLs were in the range from 60 to 80 dB HL from 500 to 8,000 Hz. Her posttreatment UCLs were increased and her dynamic range was expanded by ∼15 to 25 dB across frequency after ∼6 to 7 months of her assigned treatment (i.e., counseling combined with placebo sound generators). Although not presented here, but described in a companion report in this issue,¹⁴ 25T subsequently was crossed over to the full treatment protocol, including conventional sound therapy from bilateral sound generators and recounseling. The resulting full treatment approach for 25T then yielded UCLs approaching or exceeding the expected normal range of sound tolerance, with further and dramatic expansion of her dynamic range; at some frequencies this expansion exceeded 40 dB.

Subject 30T

Subject 30T was 67 years of age at the onset of this study. He presented with a bilateral mild to moderate high-frequency sensorineural hearing loss. His pure tone thresholds and pre- and posttreatment UCL and dynamic range data are displayed in Fig. 1. His pretreatment UCLs were in the range from 65 to 80 dB HL between 500 to 8,000 Hz. After ∼7 to 8 months of low-level sound therapy, including structured counseling for dynamic range expansion,¹³ his dynamic range was increased across all test frequencies, with the largest expansion, up to 30 dB, achieved in the 500- to 2,000-Hz range.

Procedure

Each subject was seen for multiple test sessions over the course of the dynamic range expansion trial.¹ Each session lasted ∼2.5 hours, during which a battery of behavioral and electrophysiologic measurements was conducted. Test sessions 2 to 4 were scheduled approximately 2 weeks apart, subsequent to baseline testing just prior to treatment onset. The trial period ranged from ∼7 to 11 months based on the participant and his or her response dynamics to treatment.¹

All electrophysiologic testing was conducted in a sound booth (Industrial Acoustics Corporation, Bronx, NY; model 1400ATT) using a commercially available evoked potential unit (Biologic, model Navigator SE; Natus Medical Corporation, Pleasonton, CA) to generate the stimuli and record the ABR and MLR responses. Responses for the ABR and MLR were recorded simultaneously during each test session to brief 500- and 2,000-Hz tones, which were presented at several stimulus intensities.⁶ The presentation order of the test frequencies was randomized across test sessions.

At the first electrophysiologic test session, each subject was asked to judge the loudness of the brief 500- and 2,000-Hz tone bursts as a function of presentation level. This first session served as an opportunity for the subjects to judge the loudness of these brief stimuli across the categorical continuum for the Contour Test of Loudness.²⁹ Presentation levels (in decibels re normal hearing level [nHL]) were recorded that corresponded to each subject's loudness judgments for each of the seven Contour categories ranging from very soft to uncomfortably loud.²⁹ At subsequent sessions, the ABR and MLR were recorded at the same physical intensities that were established in the first test session for the comfortable, but slightly soft; comfortable; comfortable, but slightly loud; and loud, but OK categories. Because all of these subjects either had a documented history or presented with complaints of reduced sound tolerance, ABRs and MLRs were not recorded at intensities that corresponded to the uncomfortably loud category, thus, avoiding levels of discomfort for these subjects. ABRs and MLRs also were not recorded at sound levels that corresponded to categorical judgments of very soft and soft categories because waveforms collected at these low stimulus intensities were not replicable and, therefore, the ABR/MLR waveform features could not be clearly identified. Their loudness judgments for the toneburst stimuli were recorded only at the first test session because we had multiple other measures of perceptual change in loudness for them, as evidenced by increased LDLs, shallow posttreatment loudness growth functions, and expanded dynamic ranges (see outcome profiles for subjects 5T, 25T, and 30T in the Appendix).

The procedure for measuring the categorical judgments of loudness is discussed in detail in the Methods section of a companion report in this issue.⁶ Thus, only a brief synopsis of these procedures is presented here. The categorical loudness judgments were measured separately for the 500- and 2,000-Hz tone bursts. A series of tone bursts was presented for ∼1,000 milliseconds. Each test stimulus was presented in ascending level, starting at a low initial level at or just above each subject's audibility threshold. The level (in decibels re nHL) was increased in 5-dB steps for each stimulus frequency. Each series of brief tone bursts was presented three times at each stimulus level. The median presentation level for each loudness category was determined from the three-trial sequences, and this value was used for final analysis.

Stimuli

The stimuli, described by Korczak et al, were 500- and 2,000-Hz Blackman gated tone bursts that were presented to each subject's right ear via insert earphones (Etymotic, model ER3A; Etymotic Research Inc., Elk Grove Village, IL).⁶ The 500-Hz stimuli had rise and fall times of 5 milliseconds, and the 2,000-Hz stimuli had rise and fall times of 1.25 milliseconds. The stimuli were presented at a rate of 10.9/s using an alternating-onset polarity. Both stimuli were presented at four stimulus levels. These presentation levels corresponded to the median presentation levels (in decibels re nHL) judged by each of the subjects for the comfortable, but slightly soft; comfortable; comfortable, but slightly loud; and loud, but OK Contour test categories for a given test frequency. The presentation levels for each frequency were randomized within the test session. The presentation levels, which were the same for the pre- and posttreatment sessions, were characteristically at the low-level end of the normal range of presentation levels, which is consistent with the hyperacusis-related conditions of each individual (see Figs. 3, 4, 5, and 6).

Figure 3.

Pre- and posttreatment ABR wave V and MLR wave Pa latencies in milliseconds measured as a function of increasing loudness category at 500 Hz for subjects 5T, 25T, and 30T. The specific levels (in dB nHL) used to collect these latency measures for each individual are shown in the corresponding lower panels for each loudness category. The response latencies for the three hyperacusic subjects are compared with those measured for 10 age-matched control subjects. The range of the control mean ABR and MLR latencies (±1 standard deviation SD), from Korczak et al,⁶ is shown by shading in each panel. The range of minimum and maximum levels judged within each loudness category across the normal control group is displayed in the corresponding lower panels. The organization of Figures 4–6 follows the same pattern. Abbreviations: ABR, auditory brainstem response; MLR, middle latency response; dB nHL, decibels re normal hearing level.

Figure 4.

Shown are the pre- and posttreatment ABR and MLR wave V and wave Pa latencies as a function of increasing loudness category measured at 2,000 Hz for subjects 5T, 25T, and 30T. The levels (in dB nHL) used to measure these latency values for each individual are shown in the corresponding lower panels for each loudness category. The range of the mean response latencies (±1 standard deviation) for the normal control group as a function of loudness category is displayed in shading in each panel. Shown in the designated lower panels is the minimum and maximum range of levels judged within a given category by the control subjects. Abbreviations: ABR, auditory brainstem response; MLR, middle latency response; dB nHL, decibels re normal hearing level.

Figure 5.

Shown are the pre- and posttreatment ABR wave V-V′ and MLR wave Na-Pa amplitude values in microvolts measured as a function of increasing loudness category at 500 Hz for subjects 5T, 25T, and 30T. The specific levels (in dB nHL) used to collect these amplitude measures for each individual are shown in the corresponding lower panels for each loudness category. The range of mean response amplitudes (±1 standard deviation) as a function of loudness category for the control group is displayed in shading in each panel. The range of minimum and maximum levels judged within a given category by the normal control subjects is shown in the corresponding lower panels. Abbreviations: ABR, auditory brainstem response; MLR, middle latency response; dB nHL, decibels re normal hearing level.

Figure 6.

Shown are the pre- and posttreatment ABR and MLR amplitudes (in microvolts) measured as a function of increasing loudness category at 2,000 Hz for subjects 5T, 25T, and 30T. The specific levels (in dB nHL) used to collect these amplitude measures for each individual are shown in the corresponding lower panels for each loudness category. The range of mean response amplitudes (±1 standard deviation) as a function of loudness category for the normal control groups is shown by shading in each panel. The range of minimum and maximum levels judged within each category by the normal-hearing control group is shown in the designated lower panels. Abbreviations: ABR, auditory brainstem response; MLR, middle latency response; dB nHL, decibels re normal hearing level.

Electroencephalographic Recordings

The ABR and MLR were recorded simultaneously from the right ear only using disposable electrodes placed at Cz (noninverting), A2 (inverting), and Fpz (ground). All electrode impedances were less than 5,000 Ohms, with interelectrode impedance values less than 2,000 Ohms. The electroencephalographic channel was amplified (50,000), filtered (10 to 1,500 Hz), and digitized (512 points) using a poststimulus analysis window of 0 to 80 milliseconds. Four replications of 1,000 trials each were obtained at each stimulus intensity, yielding a total of 4,000 trials. Grand average waveforms, representing the overall average of the 4,000 sweeps, were calculated for each test condition.

Response Measurements and Analyses

Peak-to-peak amplitude measurements of ABR wave V-V′ and MLR wave Na-Pa, as well as absolute latency measurements for ABR wave V and MLR wave Pa, were taken from each subject's grand average waveform for each test condition. The respective response measurements shown in Fig. 2 for the ABR and MLR were selected mainly because these are typically the most prominent and replicable waveform features measured in response to toneburst stimulation.⁶ Specifically, wave V is the primary peak in the ABR in response to toneburst stimuli. Wave V also is commonly believed to arise from neuronal activity at the level of the lateral lemniscus as the neural fibers enter the area of the inferior colliculus.^{30 31 32} The latter, is the initial level in the auditory pathways for binaural interaction and, therefore, is a plausible site for controlling or affecting auditory pathway gain as a consequence of bilateral sound therapy.²⁴ Similarly, wave Pa is the most robust and stable peak in the MLR. Wave Pa is widely believed to have cortical origins.^{33 34 35} It therefore is an index that, more than ABR wave V, can be expected to align with judgments of loudness, which are widely assumed to require higher-level cortical function.^{36 37} The criteria applied for selection and analyses of these features are detailed in Korczak et al in this issue.⁶

Figure 2.

Electrophysiologic response measures and the study hypotheses for an idealized person with hyperacusis pre- and posttreatment to expand the auditory dynamic range for loudness. The dark solid waveform for an adult with normal-hearing sensitivity shows a simultaneously recorded ABR and MLR waveform in response to a 2,000-Hz tone burst. ABR wave V and MLR waves Na, Pa, and Nb are labeled along the waveform. Magnified in the inset box is ABR wave V′, which was used to obtain the wave V-V′ amplitude values in this study. Wave V′ is defined as the primary negativity that occurs within 8 milliseconds following the latency of wave V.³⁹ The dashed waveform shows the predicted effect of hyperacusis on the simultaneously recorded pretreatment ABR and MLR. The hypothetical pretreatment waveform is characterized by larger response amplitudes and shorter latencies relative to the normal control response. The solid red waveform shows the predicted posttreatment response subsequent to a successful intervention to expand the dynamic range for loudness. The predicted posttreatment waveform reflects diminished response amplitudes and increased response latencies, with these measures returning toward normal values. Abbreviations: ABR, auditory brainstem response; MLR, middle latency response.

Of specific interest in the analyses were responses patterns that suggest or reveal evidence of hyperacusis in the pretreatment responses (i.e., larger response amplitudes and shorter response latencies than normal) and evidence of response trends in the posttreatment measures consistent with diminished hyperacusis conditions (i.e., smaller normative-like response amplitudes and longer response latencies relative to their corresponding pretreatment values). Ostensibly, such evidence, if found, would lend support for the working hypotheses introduced earlier in terms of underlying neuronal gain processes, while providing insights into the neural origins of the underlying hyperacusis conditions and the neuronal sites responsive to treatment.

RESULTS

Overview

The case results presented here are organized to test predictions delineated in the hypotheses. An idealized illustration of these hypotheses for a hearing-impaired individual with hyperacusis is presented in Fig. 2, pre- and posttreatment, relative to a typical response waveform for an idealized control subject with normal-hearing sensitivity and average sound tolerance. The response indices for deriving our primary measures of amplitude and latency for the ABR and MLR, respectively, are identified in the control response waveform and in the inset. The specific predictions, highlighted in Fig. 2, are that hearing-impaired individuals with hyperacusis will exhibit (1) larger than normal response amplitudes and shorter than normal response latencies pretreatment and (2) reduced response amplitudes and lengthened response latencies (relative to corresponding pretreatment values) posttreatment. The latter prediction further posits that these posttreatment measures will approach normal values or more typical values consistent with the amount of hearing loss for the individual.

To facilitate comparison of a given set of amplitude or latency data for a given frequency condition, the pre- and posttreatment ABR and MLR response values for each hearing-impaired individual are shown in all presentations of the findings in the form of amplitude (in microvolts) or latency (in milliseconds) functions with respect to increasing loudness category. In Figs. 3, 4, 5, and 6, the specific levels (in decibels re nHL) used in collecting these response measures for each individual are shown in the lower inset panel for each loudness category. These median presentation levels correspond to the respective categorical judgments of loudness reported by each individual. For comparison, corresponding average amplitude or latency functions (with shading of the ± 1 standard deviation ranges) are shown for the common set of ABR or MLR measures for a control group of 10 typical middle-age normal-hearing persons described in this issue by Korczak et al.⁶ The range of minimum and maximum levels judged within a given loudness category across the group of 10 control listeners is shown for them in the lower inset panels in Figs. 3, 4, 5, and 6. The range of levels within category corresponds to those used to measure the respective ABR and MLR data from the individuals in the control group. The loudness judgments for this control group are described in greater detail in companion reports in this issue.^{6 38}

Latency Responses

Latency Trends for 500 Hz

Because the reliability and stability of absolute latency response measures generally tend to be better than those reported for corresponding amplitude response measures,³⁹ we present first in Fig. 3 the ABR wave V and MLR wave Pa latency functions for subjects 5T, 25T, and 30 T for 500 Hz.

Consider first the ABR wave V pre- and posttreatment latency functions for subject 5T. The latency values in each function declined systematically for 5T with increasing loudness across category from comfortable, but slightly soft sounds (45 dB nHL) to loud, but OK sounds (70 dB nHL). The functions connecting the corresponding pre- and posttreatment ABR wave V latency values for 5T interweaved across the loudness continuum. Both the pre- and posttreatment ABR wave V latency functions for 5T reflected values that were consistently longer than the normal range of latency values across the measured loudness continuum at 500 Hz. Thus, these latency results were markedly inconsistent with predictions that 5T's pretreatment ABR wave V latencies should be shorter than normal and also with the prediction that her posttreatment latencies should be lengthened relative to pretreatment values and, in turn, in better agreement with the control range of latency values. By contrast, her pretreatment latency function for MLR wave Pa at 500 Hz consistently reflected latency values measured across the loudness continuum that were somewhat shorter than the average control latency values; these values fell just beneath or on the lower boundary of the control range of latencies. These pretreatment values were consistently shorter than 5T's corresponding posttreatment values. The resulting posttreatment latency function closely approximated the average control latency function and fell entirely within the control latency range. These MLR latency results for 5T therefore were consistent with the predictions that: (1) her pretreatment latencies should be shorter than normal and (2) her posttreatment latencies should be prolonged (relative to pretreatment latencies) and should approach those of the control listeners. Thus, for 5T at 500 Hz, the ABR wave V latencies did not agree with predictions, whereas her MLR wave Pa latencies were consistent with predictions.

The corresponding pretreatment ABR wave V latency function for subject 25T at 500 Hz also reflected values higher than or equal to the control values across the loudness continuum from the comfortable to the loud, but OK judgments. These ABR latency findings were generally inconsistent with pretreatment predictions for 25T. The posttreatment latency function reflected values higher than those measured for either the pretreatment function or for most of the control range of latencies. These posttreatment ABR latency results for 25T also were inconsistent with predictions. The MLR wave Pa pre- and posttreatment latency functions for 25T at 500 Hz mostly interweaved within the control range of latency values; the former result was inconsistent with prediction. Accordingly, neither the ABR nor MLR latency results for 25T provided substantive support for our predictions at 500 Hz.

Similarly, neither the ABR nor MLR latency data for 30T at 500 Hz provided consistent support for either of the predictions in as much as both the pre- and posttreatment ABR latency functions overlaid the upper boundary of the control range. This may, in fact, be a sensible finding for 30T because his audiometric threshold at 500 Hz was within normal limits. However, his MLR pretreatment function reflected latency values that were obviously longer than the corresponding control values across the loudness continuum, and his posttreatment MLR latency function mostly ran along the upper boundary of the control range of latencies. The latter finding offers some weak support for the prediction that the MLR latency response should be more like the control function posttreatment, which again makes sense because he had a hearing threshold at 500 Hz within normal limits.

Latency Trends for 2,000 Hz

The corresponding latency results measured at 2,000 Hz for subjects 5T, 25T, and 30T are shown in Fig. 4 for ABR wave V and MLR wave Pa. For the most part, the pre- and posttreatment ABR wave V latency functions for 5T and 30T superimposed on the control range. The only exception is that for the pretreatment latencies for the comfortable, but soft categorical judgments, which tended to be slightly elevated relative to either the posttreatment or control ranges of latencies. The pre- and posttreatment ABR wave V latency functions for subject 25T interweaved just above the control range. In general, these results do not support the prediction of briefer-than-normal pretreatment ABR wave V response latencies at 2,000 Hz nor corresponding posttreatment latencies appreciably lengthened relative to the pretreatment latencies.

In contrast, the MLR Pa pre- and posttreatment latency functions for all three subjects appear entirely consistent with predictions. Specifically, pretreatment latencies for all three individuals were shorter than those of the control average latencies. Also, their posttreatment latencies, relative to their pretreatment latencies, typically were prolonged and were within the control range across most or all of the loudness continuum for each individual.

Amplitude Responses

Amplitude Trends for 500 Hz

Consider now the amplitude response measures for each individual. The corresponding ABR wave V-V′ and MLR wave Na-Pa peak-to-peak amplitude functions for subjects 5T, 25T, and 30T are shown for 500 Hz in Fig. 5. The respective control average amplitude function from Korczak et al and the shaded ± 1 standard deviation range around the control mean function is superimposed in each panel.⁶ Shown again in the lower inset panel are the control ranges of levels (judged for each loudness category) used in the ABR and MLR measurements and the corresponding measurement levels for subjects 5T, 25T, and 30T. To reiterate, the pretreatment prediction is that their ABR and MLR amplitude values will be greater than those measured on average for the control listeners or those measured posttreatment for the individual listeners.

The pretreatment ABR amplitude function for 5T increased dramatically above both the control function and her posttreatment function as loudness increased across the comfortable, but slightly loud and loud, but OK categories. Her posttreatment amplitude function, except for the loud, but OK category (for which the amplitude response remained above the control function), otherwise interweaved with the control function. This set of findings is generally consistent with our predictions for both her pre- and posttreatment responses. 5T's MLR pre- and posttreatment amplitude functions were almost identical, reflecting dramatic increases in amplitude with increasing loudness category. 5T's increased pretreatment amplitude response was entirely consistent with prediction, but her posttreatment response was essentially unchanged from that measured pretreatment and, therefore, was not in agreement with prediction.

Subject 25T's ABR amplitude functions, both pre- and posttreatment, closely approximated the control amplitude function across the loudness continuum at 500 Hz. Her MLR pre- and posttreatment amplitude functions at 500 Hz were only slightly higher than the corresponding range for the control function, offering weak evidence that her pretreatment amplitude response may provide some support for the prediction. Her companion posttreatment MLR function reflected little difference from the pretreatment function, which is contrary to prediction.

At 500 Hz, the pretreatment ABR amplitude function for subject 30T was mostly separated from, and obviously higher than, his posttreatment ABR function. The latter, in turn, was higher than the control function, except for the loud, but OK category at which the two functions merged. These trends for 30T at 500 Hz were consistent with predictions for both the ABR amplitude pre- and posttreatment functions. In contrast, 30T's MLR pre- and posttreatment functions virtually superimposed along the upper boundary of the control range across the loudness continuum for 500 Hz. The latter posttreatment findings for 30T do not support prediction in as much as his pretreatment MLR amplitude function did not obviously differ from his posttreatment amplitude function, which again is not entirely surprising given his normal audiometric threshold at 500 Hz. Nonetheless, it is challenging to reconcile the apparent discrepancies in these ABR and MLR findings for 30T at 500 Hz.

Amplitude Trends for 2,000 Hz

The amplitude functions measured for 5T, 25T, and 30T at 2,000 Hz are shown in Fig. 6. The ABR amplitude functions for 5T at 2,000 Hz were generally similar to those reported for her at 500 Hz, albeit not as distinctively different as measured for her pre- and posttreatment nor as large as measured pretreatment at 2,000 Hz. These results, however, were mostly consistent with positive predictions for 5T. Her MLR pre- and posttreatment amplitude functions at 2,000 Hz mirrored her 500-Hz amplitude measures, reflecting increasing strength of similar magnitude across the loudness continuum. The latter evidence for 5T of an unchanged posttreatment MLR amplitude function (relative to that measured pretreatment) does not support the posttreatment prediction at 2,000 Hz.

The ABR pre- and posttreatment amplitude functions at 2,000 Hz for 25T generally interweaved across the loudness continuum within the control range, whereas her MLR pre- and posttreatment amplitude functions interweaved almost uniformly at higher amplitudes above the control range. These findings for subject 25T, other than her elevated pretreatment MLR amplitude function, do not support predictions.

At 2,000 Hz, subject 30T exhibited increasing ABR amplitudes as a function of increasing loudness category in both his pre- and posttreatment functions, and both were appreciably elevated above the control range. However, in contrast to prediction, his posttreatment ABR amplitude function was greater than that measured pretreatment at the two highest loudness categories. His pre- and posttreatment MLR amplitude functions were indistinguishable within the control range across the full loudness continuum. Thus, other than his elevated pretreatment ABR amplitudes, there was little evidence to support the predictions at 2,000 Hz for 30T.

Summary of Results Relative to Predictions

To facilitate a comparison of the results across measured indices, frequencies, and subjects, a summary of the trends is presented in Table 1 relative to the predicted results. Each “minus” entry in the table denotes the specific outcome was inconsistent with the corresponding pre- or posttreatment prediction for the associated ABR or MLR latency or amplitude measure for the identified individual. Conversely, a “plus” entry indicates an outcome consistent with prediction.

Table 1. Summary of Response Patterns Across ABR and MLR Latency and Amplitude Measures, Frequency, and Subjects Relative to Predicted Response Patterns.

	500 Hz			2,000 Hz
	5T	25T	30T	5T	25T	30T
Latency (ms)
ABR (wave V)
Pretreatment	−	−	−	−	−	−
Posttreatment	−	−	−	−	−	−
MLR (wave Pa)
Pretreatment	+	−	−	+	+	+
Posttreatment	+	−	+ (weak)	+	+	+
Amplitude (μV)
ABR (wave V-V′)
Pretreatment	+	−	+	+	−	+
Posttreatment	+	−	+	+ (weak)	−	−
MLR (wave Na-Pa)
Pretreatment	+	+ (weak)	+ (weak)	+	+	−
Posttreatment	−	−	−	−	−	−

Minus (-) indicates an outcome inconsistent with prediction, and plus (+) indicates an outcome consistent with prediction. The red colored symbols in the table denote outcomes for all three subjects are consistent with pre- and posttreatment predictions for the 2,000-Hz MLR latency measures. Abbreviations: ABR, auditory brainstem response; MLR, middle latency response.

Suffice to say, all of the ABR wave V latency measures pre- and posttreatment at both 500 and 2,000 Hz yielded negative outcomes in terms of the respective predictions for all three individuals. In contrast, the pre- and posttreatment MLR wave Pa latency values for 5T and the corresponding posttreatment MLR latencies for 30T, albeit weak, supported predictions at 500 Hz; however, neither of 25T's pre- or posttreatment MLR latency functions nor the pretreatment MLR latencies for 30T were consistent with predictions at 500 Hz. Most significantly in this analysis, all of the MLR latency outcomes, pre- and posttreatment for each individual, were strongly consistent with predictions at 2,000 Hz.

The corresponding ABR wave V-V′ peak-to-peak amplitude outcomes were consistent with pretreatment predictions for 5T and 30T at both 500 and 2,000 Hz, but neither amplitude outcome at 500 nor at 2,000 Hz was consistent with the pretreatment prediction for 25T. Similarly, the trends of the posttreatment amplitudes at 500 Hz for both 5T and 30T agreed with predictions, as did that at 2,000 Hz for 5T (albeit the prediction was supported only weakly). Conversely, the posttreatment ABR amplitudes for 25T at 500 Hz and those for 25T and 30T at 2,000 Hz were inconsistent with predictions. The pretreatment MLR amplitudes for all individuals, excluding that for subject 30T at 2,000 Hz, were consistent with prediction (albeit the MLR pretreatment amplitudes for 25T and 30T at 500 Hz were only very weakly positive). In contrast, none of the posttreatment MLR amplitude outcomes for either 500 or 2,000 Hz supported predictions for any of the three individuals.

Thus, the only positive outcomes entirely consistent with predictions pre- and posttreatment across all individuals in this analysis were the MLR latency findings at 2,000 Hz. These consistently positive outcomes are denoted by the red symbols in Table 1. Otherwise, pretreatment amplitude values for both the ABR and MLR were suggestive of positive outcomes consistent with predictions for at least two of the individuals at both frequencies. Also, posttreatment MLR latency values for all individuals, except 25T, were positive with respect to the predictions at both frequencies.

DISCUSSION

Predictions and Conflicts

On one hand, the response trends summarized in Table 1 are highly encouraging and, on the other hand, are disappointing. That the pre- and posttreatment MLR latencies for all three subjects were consistent with predictions at 2,000 Hz is indeed a positive result. Specifically, these findings indicate pretreatment latencies were consistently briefer than control values, and posttreatment latencies were consistently prolonged relative to pretreatment values. Moreover, the posttreatment MLR latencies closely approximated the control latencies at 2,000 Hz, which is an outcome that further supports the predictions. The corresponding MLR latency results were either mixed or inconsistent at 500 Hz for 25T and 30T, but were in agreement with both pre- and posttreatment predictions for subject 5T.

Perhaps the most troubling conflict in our findings arises when one considers the MLR amplitude values, which we anticipated would inversely mirror the MLR latency trends. Whereas the pretreatment MLR amplitude values at 500 Hz for all three subjects and at 2,000 Hz for 5T and 25T were consistent with the predictions (i.e., pretreatment amplitudes larger than control amplitudes), none of the corresponding posttreatment MLR amplitude outcomes for any subject or frequency supported the prediction of diminished posttreatment amplitudes relative to those measured pretreatment. Specifically, the posttreatment MLR amplitudes were mostly unchanged relative to the pretreatment MLR amplitudes. These posttreatment amplitudes therefore were not reduced relative to the corresponding pretreatment amplitudes when, simultaneously, the associated posttreatment MLR latencies (most notably at 2,000 Hz) were increased (into the control range) with respect to the corresponding briefer pretreatment MLR latencies.

Similarly, there is conflict between the ABR latency values and the ABR amplitude values. The former were prolonged rather than diminished relative to the control ABR latencies or they were similar to the control values pre- and posttreatment (none of these outcomes was anticipated a priori). In contrast, the ABR amplitudes pretreatment for 5T and 30T were greater than control values at both 500 and 2,000 Hz and also greater for them at 500 Hz posttreatment (predicted a priori). There was little or no change from pre- to posttreatment in the ABR amplitudes at 2,000 Hz for any of the three subjects, which agrees with the finding of a contrary negative prediction for ABR latency change posttreatment at 2,000 Hz for all subjects. Thus, these latter results, albeit internally consistent, do not support the expected inverse relations between the corresponding ABR amplitude and latency values.

Notwithstanding the above conflict with the companion MLR amplitude results, most notably those measured posttreatment, and the limited evidence presented here for three subjects, the very strong and positive MLR latency trends at 2,000 Hz pre- and posttreatment for all three subjects suggests this measure may be the best electrophysiologic index of pretreatment function and posttreatment change for future consideration. This measure, among those considered here, was the single measure that consistently agreed with the predicted trends obtained in this study. Moreover, the measured trends for the MLR latencies pre- and posttreatment at 2,000 Hz were prominent and robust across the loudness continuum, as well as obvious across all three subjects. The pretreatment MLR amplitudes, indicating the anticipated greater than control amplitudes for all subjects except 30T at 2,000 Hz, also lend further support for the MLR as a measure of interest for the future.

The only consistently tenable evidence to suggest the tone burst-evoked ABR as a measure of future interest was the finding that the pretreatment ABR amplitude values for 5T and 30T were consistently greater than control values at both 500 and 2,000 Hz. However, their posttreatment ABR amplitude values changed ambiguously or were little changed. This mixed finding, combined with the negative outcomes for ABR latency, which were prolonged pretreatment and were little changed or unchanged posttreatment at either frequency, diminishes enthusiasm for use of toneburst ABR measures in subsequent studies of this kind.

Differential Gain Processes

Perhaps the ambiguous outcomes for the ABR measures pre- and posttreatment reflect different gain processes at the brainstem level from those under investigation by the MLR, which ostensibly assayed cortical function. This idea is supported by unrelated evidence for a gain process operating at a lower-brainstem level, manifested in measures of the acoustic reflex, which appears to act separately and independently from a second gain process, probably with longer response dynamics.^{40 41} The latter appears to operate at a higher level in the auditory pathways to affect loudness judgments. Our tentative and limited results suggest the patterns of the loudness judgments and the positive effects measured for 5T, 25T, and 30T, revealed by incremental changes in their loudness judgments, sound tolerance, and expansion of their auditory dynamic ranges (see outcome profiles in the Appendix), may therefore be revealed only in the cortically derived MLR, but not in the ABR. This would seem sensible in as much as most investigators posit that judgments of loudness represent higher-order auditory processing that requires cortical function.^{36 37}

Dual (or even multiple) independent neuronal gain processes operating at the brainstem and cortical levels might also account for some of the key differences in the posttreatment response measures for 25T relative to the response patterns measured for 5T and 30T. Subject 25T effectively received only the counseling treatment,¹³ which was combined with placebo sound generators in her original treatment assignment (of specific interest in this report).¹⁴ In contrast, 5T and 30T received counseling and used conventional sound generators to achieve the low-level sound therapy. Also, as noted previously, the posttreatment ABR amplitude measures for 5T and 30T changed ambiguously or were little changed relative to their pretreatment measures, and their ABR latency values were prolonged pretreatment, but were little changed or possibly prolonged posttreatment. These negative ABR outcomes for 5T and 30T are consistent with posttreatment effects arising at levels above the brainstem and with positive MLR findings measured at the cortical level. The only strongly positive predicted outcome realized for 25T was measured for her MLR pre- and posttreatment latencies; seemingly this result for 25T follows sensibly if the treatment effect of counseling alone is, as expected, a cortical-level effect.

Challenges for Interpretation of the Auditory Brainstem Response/Middle Latency Response Measures

There are at least two primary challenges that need to be considered in the interpretation of the data from these case studies. These challenges are to account for (1) the influence of sensorineural hearing loss on the ABR and MLR response measures of our three subjects in the face of hyperacusis and (2) the differences in the ranges of presentation levels of the tone-burst stimuli used to measure the ABR and MLR for the hyperacusic and normal control subjects (i.e., typically lower levels, 40 to 75 dB nHL, for the former and mostly higher levels, 35 to 95 dB nHL, for the latter).

Consider first the interpretation challenge arising from the effects of sensorineural hearing loss. Nousak found differential effects of sensorineural hearing loss on response measures for simultaneously recorded ABRs and MLRs to 1,000-Hz toneburst stimulation.⁴² Specifically, Nousak found for typical persons with sensorineural hearing losses of mild-to-moderate degree, ABR wave V latencies were delayed and wave V-V′ amplitudes were reduced in comparison to the corresponding values measured for normal age-matched control subjects. These comparisons are shown in Table 2. This effect of sensorineural hearing loss appears to be largely independent of stimulus intensity. In contrast, Nousak found sensorineural hearing loss had no appreciable effect on MLR wave Pa latencies across stimulus intensity, but wave Na-Pa amplitudes were increased at most stimulus intensities by the effects of the hearing loss.

Table 2. Mean Absolute Latencies for ABR Wave V and MLR Wave Pa and Mean Peak-to-Peak Amplitudes for ABR Wave V-V′ and MLR Wave Na-Pa Recorded in a Group of 12 Normal-Hearing Adults and a Group of 12 Adults with Mild-to-Moderate Sensorineural Hearing Losses.

Stimulus Intensity	ABR Wave V		MLR Wave Pa
	Normal Hearing	Hearing Impaired	Normal Hearing	Hearing Impaired
Mean absolute latencies (ms)
90 dB nHL	6.10	6.36	28.1	28.1
80 dB nHL	6.37	6.54	28.03	28.17
70 dB nHL	6.48	7.22	29.07	28.91
55 dB nHL	7.10	8.52	30.26	30.03
	ABR Wave V-V′		MLR Wave Na-Pa
Mean peak-to-peak amplitudes (μV)
90 dB nHL	1.01	1.21	1.21	0.81
80 dB nHL	0.96	0.79	1.06	1.22
70 dB nHL	0.78	0.70	0.91	1.02
55 dB nHL	0.57	0.36	0.73	0.56

Note: These response measures were calculated from ABR and MLR data collected in response to a 1,000-Hz tone burst by Nousak.⁴²

Abbreviations: dB nHL, decibels re normal hearing level; ABR, auditory brainstem response; MLR, middle latency response.

If we assume for our hyperacusic hearing-impaired subjects a pattern of findings for tone-evoked ABRs and MLRs similar to that measured by Nousak for typical persons with sensorineural hearing losses (shown in Table 2), then we would expect ABR wave V-V′ to be decreased in amplitude and wave V latency to be prolonged among our hyperacusic subjects.⁴² However, we measured elevated pretreatment V-V′ amplitudes at 500 and 2,000 Hz for subjects 5T and 30T, relative to our control group values, and their wave V latencies were essentially in the normal range at both stimulus frequencies (with the exception of those measured at 500 Hz for subject 5T). A second intriguing finding is evident when we compare subject 5T's pretreatment ABR amplitude values to the values reported by Nousak for her hearing-impaired subjects for the same 70-dB nHL presentation level. Specifically, 5T's wave V-V′ amplitude values were two- to fivefold larger than mean V-V′ amplitude values measured for Nousak's typical individuals with sensorineural losses (i.e., subject 5T's pretreatment ABR wave V-V′ amplitude values were 3.3 μV at 500 Hz and 1.4 μV at 2,000 Hz, which are values substantially larger than the group mean wave V-V′ value of 0.70 μV measured by Nousak at 1,000 Hz for her hearing-impaired subjects). This same pattern of findings also was measured for subject 30T. Thus, it appears for our three subjects their processing of auditory signals, at the lower/mid brainstem level is not typical of that measured on average for Nousak's hearing-impaired participants. We acknowledge, however, that we do not know whether any of Nousak's hearing-impaired subjects had hyperacusis related problems because no measures of LDL/UCLs were reported for them. Her hearing-impaired subjects also were somewhat younger in age (mean age = 42 year) than our three subjects (age 58, 67, and 69 years), but they had similar degrees of sensorineural hearing impairments to those measured for 5T, 25T, and 30T.

A problem in comparing the case data in this study to the results from Nousak's study also arises from the different presentation levels employed in these response measurements.⁴² This difference in the presentation levels used by Nousak and the levels investigated here presents the second substantive challenge for interpreting and comparing the ABR and MLR measures for the hyperacusic subjects in this three-case study with corresponding measures from Nousak or with the measures from our control subjects. As previously noted, the ABR and MLR measurements for our hyperacusis subjects were conducted at presentation levels at the lower end of our control range (∼40 to 75 dB nHL) and at typically lower levels than were used by Nousak,⁴² who presented toneburst stimuli between 55 to 90 dB nHL to her hearing-impaired subjects. This use of lower presentation levels to measure the ABR and MLR for our three cases is consistent with their hyperacusis problems. In light of the fact that we were testing them at lower stimulus presentation levels relative to those used in the ABR/MLR measurements for many of the normal controls (see the corresponding levels of the categorical loudness judgments in Figs. 3, 4, 5, and 6), we would have anticipated the hyperacusic response amplitudes to be smaller and their response latencies to be prolonged relative to those for our control subjects. Again, for subjects 5T and 30T, their pretreatment ABR wave V-V′ amplitudes were substantially larger than those of the normal controls at both stimulus frequencies. Moreover, their pretreatment wave V latencies were essentially the same as those measured for our control group. Thus, perhaps there is some information in the tone burst-evoked ABR that may be diagnostic for hyperacusis. Accordingly, there appears to be an inherent difference in electrophysiologic measures of auditory function among our three hyperacusic cases when compared with either normal-hearing persons with average sound tolerance in this study or Nousak's typical persons with sensorineural hearing losses.⁴²

Future Research Directions

Obviously, these are curious cases for which we would be remiss if we attempted to generalize the findings based on results from three hyperacusic hearing-impaired individuals. This is especially so because there were so many discrepancies within their data and challenges for interpretation of the findings. Nonetheless, the hyperacusis and associated treatment findings are intriguing and raise more questions than answers. Some obvious areas for subsequent research include consideration of click-evoked ABR and MLR measures, which, despite mixed evidence (see Fournier et al²), might be more revealing of the origins of the hyperacusis problem(s) and the effects of treatment than were revealed by our toneburst measures. A limitation of the latter in this study was the primary focus on the specific response measures reported here to the exclusion of measures of neuronal activity arising from other potentially revealing brainstem sources reflected, for example, in ABR waves I and III.³¹ These latter response indices could not be discerned in our toneburst ABR measures. Other alternative auditory electrophysiologic measures that have been used to investigate typical loudness growth, which might also provide insights for understanding loudness growth in individuals with hyperacusis and the neural origins of this growth, include the Auditory Steady State Response (ASSR).⁴³ Evaluation of the ASSR across a range of stimulus rates, including low (<20/s), middle (20 to 60/s), and high rates (70 to 110/s), allows investigators to gain insights into contributions from the cortical, subcortical, and brainstem levels of the auditory system, respectively, with a common measurement tool.⁴⁴ Contrary to most of the trends reported here for our three cases, at least two ASSR reports suggest a possible association between brainstem function and loudness judgments in typical listeners.^{45 46}Accordingly, the ASSR may hold promise for discerning the origins of hyperacusis and related loudness-based problems of persons with sensorineural hearing loss. Another potentially promising line of inquiry is evident for the late or slow cortical responses, especially because the task of performing loudness judgments is widely believed to require cortical function. Indeed, Hoppe et al demonstrated a correlation between categorical loudness judgments and late cortical responses in cochlear implants patients.³⁶ Thus, notwithstanding the challenges and complexity of interpreting the electrophysiologic measures in this study, the need for relevant studies of hyperacusis and conditions associated with reduced sound tolerance are apparent and now await future investigation to resolve discrepancies and answer questions of the kind we raised and sought to address in this research.

APPENDIX

Right-ear outcome profiles for hyperacusic hearing-impaired subjects 5T, 25T, and 30T as followed across the assigned treatment period and posttreatment for unaided listening after transition to HAs. Shown for each subject are the repeated-measures audiometric pure tone thresholds (A) and loudness discomfort levels measured at 1,000 and 8000 Hz and for white-noise stimulation (B); repeated-measures categorical loudness judgments for spondee speech stimuli and for warble tones presented at 500, 2,000 and 4,000 Hz for each response category designated in the inset legend (these results reveal the dynamics of treatment change in terms of the presentation level corresponding to respective categorical judgments versus months of treatment) (C); repeated-measures NU-6 word recognition scores measured at presentation levels judged to be comfortable and loud, but OK, respectively, for spondee speech stimuli (D); word recognition scores (in rationalized arcsine units) measured pre- and posttreatment for comfortable and loud, but OK presentation levels, respectively, superimposed on the corresponding Articulation Index prediction (shown as a performance-intensity function) from the STEPS algorithm (see details in Formby et al^{1 14}) (E); pretreatment audiometric pure tone thresholds shown together with corresponding UCL judgments measured pre- and posttreatment (F); and corresponding pre- and posttreatment dynamic ranges (G). The number in panels A to D above the HA label denotes months postonset of treatment at the time of the measurement. The shaded symbols in the panels for subject 25T denote outcomes measured during a crossover-treatment period, which is not considered in this report. Abbreviations: dB nHL, decibels re normal hearing level; HA, hearing aid; UCL, uncomfortable loudness.

Appendix Fig 1.

Appendix Fig 2.

Appendix Fig 3.