Comparing Hearing Aid Fitting Methods in Blast-exposed Veterans

Since 2000, at least 250,000 U.S. Service members have experienced a blast-related mild traumatic brain injury. A retrospective analysis of over 100,000 post-9/11 Veterans shows that blast injury more than doubles the risk of a diagnosed auditory problem. Many blast-exposed Veterans experience "functional hearing difficulties" (FHDs): problems in challenging listening environments despite clinically normal hearing as measured by the pure-tone audiogram. VA audiologists have begun using low-gain hearing aids to treat FHDs, but there are no concrete guidelines for this application given standard procedures rely on the pure-tone audiogram. This study proposes a data-driven approach called speech-based audiometry (SBA), which optimizes hearing aid gains from a patient's responses to speech stimuli in aided conditions. This trial will assess the behavioral (speech recognition in noise, subjective listening difficulty) and neurophysiological (functional neuroimaging during a speech recognition task) benefits of low-gain hearing aids programmed conventionally or with SBA among blast-exposed Veterans with FHDs.

The present study is a randomized trial of low-gain hearing aids in blast-exposed Veterans with normal or near-normal audiometric hearing and self-reported functional hearing difficulties. Hearing aids will be programmed using either conventional methods or a novel procedure called speech-based audiometry. Participants will be randomized into different groups receiving conventional or speech-based programming. Participants will be followed for a six-week intervention period and complete several outcome measures before, during, and after the intervention period. Each group will be further divided, at random, into equal subgroups assigned to (i) use their hearing aid daily during the six-week intervention period or (ii) use their hearing aid only to complete the outcomes assessments. The study will be broken into five periods:

1. Prescreening, screening, and enrollment
2. Preparatory period
3. Baseline
4. Treatment (first five weeks of hearing aid intervention)
5. Follow-up (final week of hearing aid intervention)

Prescreening, screening, enrollment (V0).

Prescreening includes identification of potential participants via medical record search, examination of patient rolls, clinician referral, or patient self-referral (e.g., from study advertisements). A prescreening phone call or in-person prescreening visit is scheduled for willing participants.

At the time of the in-person prescreening or prescreening phone call, a script will be used to learn if prospective participants meet basic eligibility criteria and are able to present for an in-person discussion of informed consent.

After prescreening is completed, potential participants meeting preliminary eligibility criteria will be scheduled for an initial study visit including informed consent procedures and, if consent is obtained, the following additional procedures.

Participants will be screened for inclusion and exclusion criteria. Screening includes performance of a routine audiological exam (otoscopy, tympanometry, air- and bone-conduction pure-tone audiometry) if a recent (< 6 months) exam is not available in the medical record. The subject also completes the following three questionnaires/interviews to determine eligibility, which are not part of routine audiological care:

1. Hearing Handicap Inventory for Adults (HHIA). The HHIA is a 25-item questionnaire that focuses on the emotional, social/situational, and occupational effects of hearing loss.
2. Mini-Mental State Exam (MMSE). The MMSE is an experimenter-led interview that assesses the global cognitive state of the subject including situational awareness, memory, language, and sensorimotor function.

3. Quantification of Cumulative Blast Exposure (QCuBE). The QCuBE is a semi-structured, experimenter-led interview that measures number and severity of exposures to blast, including date of exposure, distance from explosion, type of explosive, and protective gear worn, among other details. The QCuBE also obtains a brief history of non-blast head traumas.

   If a subject is not eligible (i.e. fails screening), an explanation is provided to the subject along with a full debriefing of the study. If a subject is eligible, they are randomized and asked to complete the following additional questionnaires:

4. Tinnitus Handicap Inventory (THI). The THI is a 25-item questionnaire that quantifies the impact of tinnitus on daily life.
5. Demographic data form (optional). This form obtains information on the subject's sex, ethnicity (Hispanic or Latino, not Hispanic or Latino), and race for reporting to the study sponsor using standard definitions.

Preparatory Period (V1, V2)

Two initial study visits will be conducted to prepare the subject for the hearing aid intervention. At the first visit (V1), the following procedures will be performed:

1. Speech-based audiometry. At the core of speech-based audiometry is the Open Speech Platform (OSP), an open-source hearing aid and speech processing platform with co-designed hardware and software elements. For the proposed work, the crucial component of OSP is a real-time master hearing aid, a modular software environment that abstracts the details of a digital hearing aid processor (e.g., block resampling, sub-band filtering, wide dynamic range compression, and feedback cancellation). That is, the master hearing aid is a software simulation of a hearing aid that performs all the same functions (e.g., sound amplification and compression). The flexible nature of the master hearing aid allows any arbitrary hearing aid prescription to be simulated in real time. The logic behind speech-based audiometry is as follows: (1) there is a library of potential hearing aid prescriptions (~1500 NAL-NL2 gain profiles) determined by unsupervised clustering performed on a large library of pure-tone audiograms from patients with a range of hearing loss levels and configurations; (2) there is also a library of speech stimuli (e.g., audiobook excerpts) that can be amplified with any prescription from the gain library in real time by OSP and presented to the patient over headphones; (3) this facilitates a non-exhaustive "search" of the possible prescriptions based on a patient's subjective ratings of speech amplified with different prescriptions; (4) in practice this takes the form of a series of A|B preference judgments made by the patient (much as in fitting for eyeglasses, "do you prefer A or B?"); (5) machine learning is used to select the A and B prescriptions for each presentation to optimize the search through the library of prescriptions for efficiency; (6) the procedure terminates when the search is finished.
2. Real ear measurement of realized gains. The real ear aided response (REAR) of the master hearing aid is obtained for the gain settings selected during speech-based audiometry following standard clinical procedures.
3. A hearing aid is ordered for the subject with preliminary gain settings established during speech-based audiometry or from the pure-tone audiogram (depending on group assignment).

At the second visit of the preparatory period (V2), real ear verification of hearing aid gain prescription is performed. That is, the preliminary gain settings initially programmed into the hearing aids are based on a set of "gain targets," which may differ from the actual gains realized when the hearing aids are worn by the subject. Thus, REAR measurements are again obtained for soft, moderate, and loud inputs separately for each ear. The hearing aid gains are manually adjusted by the fitting audiologist until the realized gains are within an acceptable range around the gain targets according to routine clinical standards. For subjects assigned to the speech-based audiometry group, the gain targets are the REAR measurements obtained during V1. For subjects assigned to the conventional fitting group, the gain targets are the output of the NAL-NL2 prescriptive formula applied to the subject's pure-tone audiogram; gains will be further adjusted based on the subject's subjective feedback, typically between 5-10 dB additional insertion gain at some or all frequencies between 500 and 4000 Hz. Once the final hearing aid gains are set, the hearing aids are placed in a hearing aid test box and their gain responses for soft, medium, and loud sounds are recorded. This will allow the study team to later use the hearing aid test box to verify that the hearing aid gains have not changed from prescribed settings. Finally, the subject will receive counseling from the fitting audiologist about how to use their hearing aids for the remainder of the study.

Baseline Period (V3, V4)

Study outcome measures will be obtained at baseline (i.e., before the daily use group has begun its six-week period of daily hearing aid usage). The first visit of the baseline period (V3) is an initial fMRI scan (see description of outcome measures). The fMRI scan session includes: (i) metal screening; (ii) counseling by a trained MRI technician about what to expect, how to remain safe, and how to trigger early termination of the MRI scan; (iii) a practice session conducted outside the scanner; (iv) familiarizing the subject with equipment to be used inside the scanner (e.g., button box, MR-compatible headphones) and placing the subject inside the MRI scanner; (v) acquisition of a T1 scout image and alignment of the MRI slices to be acquired in subsequent scans (axial slices aligned with the anterior-posterior commissure); (vi) acquisition of a high-resolution T1 anatomical scan and a gradient field mapping scan; (vii) eight fMRI scans; (viii) removal of the subject from the scanner; and (ix) 5-10 minute observation of the subject during a "cool off" period.

The second visit of the baseline period (V4) is a baseline speech-in-noise evaluation including the modified Quick Speech-In-Noise (mQuickSIN) test and the Digits in Noise (DIN) test in aided and unaided conditions. For subjects assigned to the daily hearing-aid-use group, hearing aids are dispensed to the subject for daily use over the next six weeks of the study.

Treatment Period (V5-V9)

Subjects complete weekly visits during the first five weeks of the six-week hearing aid intervention period (V5-V9). At each visit, the gain response of the subject's hearing aids will be verified in a hearing aid test box to ensure the final gain settings determined in V2 have not been altered. For subjects assigned to the daily hearing aid use group, the hearing aids' data loggers will also be checked to determine the number of hours the subject used the hearing aid each day since the preceding visit. The subject will complete the DIN as in V4.

Follow-Up Period (V10, V11)

At the end of the final week of the six-week hearing aid intervention period, subjects will complete two final visits (V10, V11) at which baseline testing is repeated. At V10, the subject completes a second fMRI scan with procedures just as in V3. At V11, the subject completes the same procedures as in V5-V9 with the following exceptions:

1. The subject also performs the mQuickSIN, as in V4.
2. The subject completes the THI, as in V0.
3. Subjects in the daily use group complete the Abbreviated Profile of Hearing Aid Benefit (APHAB). The APHAB is a validated questionnaire that measures subjective levels of listening difficulty pre and post hearing aid fitting in persons with hearing loss.

Once all the study procedures are completed, the subject receives a full debriefing. At the end of the debriefing, a study audiologist assumes the responsibility for transitioning the subject back to routine clinical care.

Instrumentation

Audiometric and speech-in-noise testing will be carried out in a double-walled sound-isolated audiometric chamber adjoined to a control room. Audiometric testing will be carried out using typical audiology equipment: A Madsen Astera2 or Grason-Stadler Audiostar Pro clinical audiometer calibrated to current standards (ANSI S3.6-2010) will be used to deliver the pure-tone signals for threshold testing through Etymotic Research ER-3A insert earphones. A Madsen Otoflex 100, Zodiac or Grasen-Stadler Tympstar Pro acoustic-immittance meter will be used for immittance measurements to verify normal middle ear function.

For speech-based audiometry, listeners will be seated comfortably in a sound-treated booth and wearing the same ER-3A earphones used in pure-tone audiometry. The procedure is driven by prototype commercial software (SpeechFit; Nadi, Inc.). Visual stimuli are displayed on a touchscreen monitor located in the sound booth. The same touchscreen monitor is used to collect responses from the subject. Acoustic stimuli are generated digitally by SpeechFit, output through a digital-to-analog converter and pre-amplifier (Motu M4), and amplified for presentation through ER-3A earphones. The equipment used for sound presentation is calibrated in Audcal software (Larson Davis, Inc.) to the same standard as routine speech recognition testing on a clinical audiometer (ANSI S3.6-2010). Calibration uses a Larson Davis 831C level meter and model AEC304 occluded ear simulator with calibrated 1/2-inch microphone.

An Auricle Freefit or Audioscan Verifit2 real ear measurement system connected to a VA-furnished and networked PC with installations of Otosuite and NOAH software will be used to collect real ear measurements. The same PC is used to program and order hearing aids.

For fMRI testing, speech stimuli will be presented over MR-compatible active noise cancelling headphones (Optoacoustics, Inc.). Amplification will be provided via the OSP master hearing aid running on the same PC used to generate and deliver the acoustic stimuli, with overall and band-specific levels verified by ear-canal probe microphone. In the aided condition, REAR will be matched as closely as possible to that recorded from the participant's fitted hearing aid. The MRI scanner is a 3 Tesla Siemens Skyra equipped with a 32-channel Siemens head coil, Philips SensaVue system for presentation of visual stimuli via MR-compatible LCD display, and Cedrus Lumina control box to collect button presses (Cedrus LS-LINE response box) and trigger signals from the scanner. Control of stimulus presentation is accomplished by custom MATLAB software running on a VA-owned PC that is permanently installed in the MRI control room.

The mQuickSIN is administered using software furnished by Walter Reed National Military Medical Center with sounds presented through a free field speaker array. The DIN is administered via custom Python software.

For all speech-in-noise tests (mQuickSIN, DIN, fMRI), unaided speech will be delivered at a moderate/conversational input level of 65 dB SPL.