Pilot Comparison of Adjustment Protocols of Personal Sound Amplification Products

Abstract

The Over-the-Counter Hearing Aid Act of 2017 was signed into law in August 2017 and facilitates the introduction of direct-to-consumer sales of hearing aids for adults with mild-to-moderate hearing loss. Among many questions surrounding over-the-counter sales is the ability of users to self-fit amplification. Many studies have conducted self-fitting procedures using guidance materials provided by audiologists. In this pilot, we explore the ability of users to self-adjust personal sound amplification devices using only materials provided by the manufacturer and contrast this with models that involve a hearing professional. Outcomes to assess adjustments included clinic-based speech-in-noise measures and ability to approximate NAL-NL2 prescriptive targets. We found that an audiologist-driven model provided the best outcomes. However, it is unknown if the difference is clinically meaningful.

In the United States, hearing loss impacts approximately 30 million adults and prevalence increases with age such that two-thirds of adults over the age of 70 years have a clinically meaningful hearing loss. ¹ Due to the aging population, the number of adults with hearing loss is expected to more than double to over 70 million by the year 2060. ² Importantly, hearing loss is associated with numerous negative healthcare outcomes including cognitive decline, incident dementia, falls, loneliness, depression, and decreased quality of life. ^{3 4 5 6 7} Moreover, the societal and personal economic impacts of hearing loss are considerable. ⁸ The positive impact of amplification on the quality of life has been established and it is possible that the use of amplification mitigates relationships with negative health outcomes noted earlier. ^{9 10 11}

Despite the known negative consequences of hearing loss and the likely benefits of amplification, hearing aid adoption remains relatively sparse at less than 20% of persons with hearing loss. ¹² Multiple factors contribute to this low uptake including affordability, accessibility, and awareness. ^{13 14 15 16} Lack of hearing aid usage has resulted in top-level attention from the White House President's Council of Advisors on Science and Technology; The National Academies of Sciences, Engineering, and Medicine; and the United States Congress. ^{13 17} Ultimately, this resulted in the passage of the Over-the-Counter Hearing Aid Act of 2017. ¹³ This act requires the Food and Drug Administration (FDA) to develop regulations for publicly available over-the-counter hearing aids (OTC HA) for consumers with mild to moderate hearing loss by the year 2020. The exact measures of regulation on OTC HA devices that will be applied by the FDA remain to be determined.

Currently, air-conduction hearing aids, those that amplify and deliver sound to the external ear canal via air conduction, are regulated by the FDA as class I medical devices. They are considered low risk and generally exempt from 510(k) premarket review. ¹⁶ Individual state laws regulate the delivery of hearing aids by licensing individuals (e.g., audiologists and hearing instrument specialists) who sell, rent, lease, dispense, and/or provide hearing aids to consumers. ¹⁶

Over-The-Counter Service Delivery Model

Alternative to the conventional model is the OTC HA delivery model whereby devices are obtained directly by the consumer rather than through a licensed dispenser. For an FDA-regulated hearing aid, this may entail circumnavigating the state dispensing laws via ecommerce or mail order sales. In addition, unregulated devices, such as a Personal Sound Amplification Product (PSAP), may be purchased directly in-store. ^{16 18} The OTC HA model removes the initial customization services provided by licensed dispensers and requires the user to self-adjust to meet their listening needs. Users adjust via pre-programmed generic or situation-specific (e.g., restaurants) settings or use volume and frequency tuning controls (e.g., buttons, wheels, smart phone applications) to customize the output. Notably, more recent devices allow for in situ hearing testing via a smart phone application which then programs the device to the user's hearing loss. OTC HA models also may preprogram the device based on a user-provided audiogram prior to delivery. ^{16 18} Beyond adjustment, the OTC HA model requires users to learn basics such as changing the batteries, insertion of the device, and cleaning the device.

The OTC HA model makes numerous assumptions of the individual. This model requires the individual to have the health literacy capacity to learn to properly manipulate devices as described earlier. ¹⁹ As important as device manipulation, the OTC HA model assumes self-management capabilities of individuals to live and cope with hearing loss. That is, the OTC HA model is unlikely to include any guidance in emotional or psychosocial aspects of living with hearing loss. ²⁰ Moreover, communication skills that are critical to compliment amplification devices may not be provided via an OTC HA model. ^{16 18 21}

Over-The-Counter Hearing Aid Devices

OTC HAs under current regulations, not to be confused with the forthcoming newly created FDA regulations for OTC HA, are technically already available via ecommerce and mail-order sales that side-step the current state dispensing laws. ^{16 18} In addition, PSAPs, which are unregulated by the FDA and available direct to the consumer, represent a low-cost OTC amplification option. ¹⁸ Under current regulations by the Federal Trade Commission (FTC), PSAPs may not be advertised directly as a compensatory mechanism or treatment for hearing loss. ²² Nonetheless, by technologic standards, they could have the same features as an FDA-regulated hearing aid. Early research on PSAPs revealed they performed relatively poorly, specifically producing high levels of equivalent input noise (EIN) and total harmonic distortion (THD) as well as having limited frequency ranges. ^{14 23} However, more recent studies have demonstrated that some technologically advanced PSAPs perform well on electroacoustic measures. ^{24 25}

The literature on the application and usage of OTC amplification is currently emerging. ¹⁴ Previous studies have used pre-programmed hearing aids to examine efficacy and effectiveness of hearing-impaired users' ability to self-adjust devices free of audiologist-led fitting. Two small cohort studies of elderly participants using hearing aids have shown quality-of-life improvements from baseline when the devices were fit using an OTC HA delivery model without audiologist involvement. ^{26 27} Importantly, a recent large double-blind randomized control trail reported a consumer-driven approach to hearing aid selection and fitting was efficacious compared with audiology best practices, revealing similar effect sizes between the two cohorts based on the Profile of Hearing Aid Benefit. ²⁸ It is important to note that in all of these OTC HA studies, the participants with hearing loss were provided instructions on the proper usage of hearing aids, which were developed by hearing care professionals.

Literature limited to available PSAPs, rather than hearing aids, and their impact on outcomes is sparse. A recent pilot study found some PSAPs improved speech understanding as well as a hearing aid, while a less advanced PSAP actually degraded speech understanding, possibly due to high EIN and THD. ²⁹ However, these results were limited in their generalizability due to the fact they were obtained from a small convenience sample in a highly controlled setting. A study of self-fitting procedures using a pre-market PSAP (later classified as a hearing aid) found that most participants were able to competently follow written procedures for self-fitting the device. ³⁰

As noted earlier, per the recent legislation, ¹³ regulations for a new category of HA for mild to moderate loss are under preparation. These devices will enter the market by 2021 and provide an option available direct-to-consumers in-stores, online, or via mail order. The pending disruption to the current standard of care warrants new research in care delivery in audiology. Understanding the impact of OTC HA delivery models on hearing care outcomes, how users interact within the current OTC HA delivery models, and how to improve delivery models (including the integration of optimal self-adjustment procedures and self-management skills to address consequences of hearing loss and expectations of care) is vital to the future of hearing care.

Two questions that arise from the current research are (1) is a user with hearing impairment able to program OTC/PSAP devices on their own based only on the materials provided by the manufacturer without instruction or supplemental materials developed by an audiologist; (2) if so, how does this relate to professional services provided by an audiologist. In this pilot study, we aimed to explore the impact of two high end PSAPs adjusted with three tiers of device customization models with various levels of professional input. We assessed the performance with each PSAP using basic clinical outcome measures including speech understanding in noise and the ability to approximate prescriptive National Acoustics Laboratories' (NAL) NAL-NL2 targets.

Methods

Study protocol and procedures were approved by the Towson University Institutional Review Board (IRB). Written informed consent was obtained prior to participation.

Participants

From December 2016 to March 2017, thirteen adults were screened and a convenience sample of nine participants was recruited for the pilot study from the Towson University Institute for Well-Being Audiology Clinic and the local community. In accordance with the target for consumers of OTC HA technology, participants had symmetrical mild-to-moderate sensorineural hearing loss defined as four-frequency pure-tone average (PTA) of 0.5, 1, 2, and 4 kHz between 20 and 55 dB HL, no air–bone gaps of greater than 10 dB HL, no between-ear air conduction threshold gaps greater than 10 dB HL at a given frequency, and type-A tympanometry bilaterally. In addition, participants had no evidence of significant cognitive decline based on a score of ≥25 on the Department of Veterans Affairs St. Louis University Mental Status (VA-SLUMS) ³¹ examination. As an efficacy study under ideal conditions, measures were taken to limit the participation group to those with typical presbycusic hearing loss (i.e., the demographic for which new FDA-regulated OTC HA are intended). As such, participants were excluded if there was evidence of noise-induced hearing loss (based on self-report noise-exposure and researcher judgment of notch at 3, 4, and/or 6 kHz), hearing loss secondary to a medical condition, a self-report history of amplification use longer than 1 month in the past or reported inability to perform basic functions with a smart cell phone.

Procedures

For screening and baseline purposes, all participants completed an audiologic evaluation in a double-walled sound booth (calibrated to American National Standards Institute specifications; ANSI S3.1–1999) with a calibrated (ANSI S3.6–1996) audiometer (Grason-Stadler, Eden-Prairie, MN; GSI-61) at the Towson University Audiology Clinic. This included bilateral otoscopic examination, tympanometric testing (Grason-Stadler, Tympstar V2), pure-tone air conduction (250–8,000 Hz), and bone-conduction (500–4,000 Hz) audiometry, and completion of the AzBio ³² sentence test. The AzBio sentence lists include 20 sentences, ranging from 4 to 12 words, of equal intelligibility. The AzBio sentence test was presented in sound field via a speaker at a 0-degree azimuth to the participant with concurrent noise presented via speaker at a 90-degree azimuth to the participant. Noise was directed at the right ear, which was the ear fit with a PSAP. Unilateral fittings are used because we believe it simulates a more realistic real-world use of PSAPs. Sentences were presented at +5 dB sensation level to the participant's PTA with a +5 signal-to-noise ratio to concurrent four-talker babble noise. The order of AzBio lists was randomized to control for order effects.

Following baseline measures, participants were asked to repeat the AzBio testing procedures described earlier with two different PSAPs in each three fitting conditions (total six conditions). Based on positive findings compared with a hearing aid from a previous study, ²⁴ the Sound World Solution CS-50+ (Sound World Solutions, Park Ridge, IL) and Soundhawk (Soundhawk, Cupertino, CA) PSAPs were selected based on how well these devices performed compared with a traditional hearing aid from a previous study. ²⁴ Device electroacoustic measurements, including high-frequency average (HFA) OSPL90, frequency range, EIN, and THD (recorded as a single average of 500, 800, and 1600 Hz) were obtained prior to each participant's session with Audioscan Verifit 1.0 test box (Audioscan, Dorchester, ON) to ensure reasonable device capability. In the event that a PSAP device failed or drastically changed (>25% difference from baseline measures) for the worse, a backup device was available. The three fitting conditions, described in detail later and used for each device, included out-of-the-box, advanced-user fit, and audiologist-fit. Lastly, aided real-ear measurements were obtained using the Audioscan Verifit 1.0 for each condition after all adjustments were made but prior to repeating the AzBio sentence test procedures.

The out-of-the-box condition was designed to simulate the most basic fitting procedures using a PSAP. This involved letting the participant choose the most comfortable dome and manipulate the PSAP using only the manufacturer's instructions to adjust the volume and/or select a preprogrammed acoustic setting. Participants were not offered any help nor allowed to reference other materials or sources for aide in self-fitting. Participants were allowed to listen to a practice AzBio sentence list (presented in the parameters described earlier) to further adjust to a level they perceived to be their optimal fit for listening in noise. Notably, the AzBio list used in practice was not repeated during the follow-up sessions.

The advanced-user condition, designed to simulate a fitting procedure based on advanced advice from a professional, involved using the most comfortable dome and instructing the participants to utilize each PSAP's accompanying customization software via a smart phone application (CS Customizer by Sound World Solutions and Soundhawk Smart Listening System). For the CS-50+, the software application customized gain at low-, mid-, and high-frequency bands based on the participant's degree of hearing loss from in-situ hearing test results using a proprietary algorithm for the “everyday” setting. Following this initial fitting, the participants were then instructed to use the software's “Equalize” software platform to further adjust intensity at low-, mid-, and high-frequency bands to their preferred setting while listening to the practice AzBio sentence list with concurrent noise (described earlier). For the Soundhawk, the application set the gain to the default “indoors” setting. Participants were then instructed to use the software's “Sound Scene” feature to customize fitting using a touch-based application. In this application, frequency occurred along the x -axis, while intensity of sound occurred along the y -axis and participants were instructed to move their finger along the screen to find their preferred listening combination of intensity and frequency emphasis. Similarly, participants utilized the practice AzBio list to customize to their preference for listening in noise. While the researcher did not instruct the participant on levels (frequency and intensity) to which they should adjust each device, they did provide complete instructions on how to manipulate the devices and were present to answer follow-up questions.

The audiologist-fit condition was used to simulate fitting procedures completely driven by a professional. The researcher utilized real-ear measurements to fit PSAPs to the participant's NAL-NL2 fitting targets at 0.5, 1, 2, and 4 kHz for an average speech input (65 dB SPL). The researcher leveraged the customization applications and changed dome size to adjust devices to as close as possible to NAL-NL2 targets.

To account for the effects of learning and fatigue, the order of devices and AzBio sentence lists were randomized. To simulate a real-world “out-of-the-box” and “advanced-user” self-fitting procedure, the participants were not informed of the results of their audiologic testing until the end of the session after all conditions had been completed. The results of electroacoustic analysis, real-ear measurements, and speech-in-noise testing were recorded for each device and fitting protocol. All participants were reimbursed for their time with a $25 gift card.

Outcome Measures and Analyses

The primary outcome for this pilot study was the absolute change in AzBio score from unaided baseline to six aided experimental conditions. The AzBio was scored as the absolute percent of words repeated back correctly on the 20-sentence list. The secondary outcomes of the study included electroacoustic measurements (OSPL90, frequency range, EIN, and THD) compared with manufacturer specifications; real-ear measurement data presented as ability to meet NAL-NL2 targets within ± 5 dB at 0.5, 1, 2, and 4, kHz and as a root mean square (RMS) value; and an exploration of the relationship between real-ear measures and change in AzBio score. Given the pilot nature of this study and its small sample size, the ability to perform inferential statistics is limited. In lieu, descriptive statistical reporting drove analyses and emerging trends were noted. Confidence intervals were calculated for change measures for reference only but are not interpreted for inference. Correlative measures were explored but sample size and limited data collection limited modeled analyses and interpretation. Statistical Package for the Social Sciences (SPSS) was used to create figures and graphs for display of the collected data.

Results

Table 1 displays the sex, age, VA-SLUMS, and PTA results for the participants. The mean age of participants was 69.67 years (standard deviation [SD] = 11.45; range = 51–82) with a mean PTA of 28.89 dB HL (SD = 6.07; range = 23.75–40) in the right ear. Six of nine participants were male.

Table 1. Baseline Demographic Characteristics.

Participant	Sex	Age	VA SLUMS	Right Ear PTA
1	M	82	26	33.75
2	M	51	28	23.75
3	F	54	29	23.75
4	F	59	28	22.50
5	M	61	28	27.50
6	M	51	27	25.00
7	M	58	27	35.00
8	M	78	27	40.00
9	F	70	30	28.75
Mean (Standard Deviation)		62.67 (11.45)	27.78 (1.20)	28.88 (6.07)

Note. PTA = pure-tone average of 500, 1000, 2000, and 4000 Hz. VA SLUMS = Veterans Affairs Saint Louis University Mental Status (out of 30 points).

Table 2 displays absolute difference from AzBio score baseline for each participant as well as mean scores across conditions. The overall mean unaided baseline AzBio percent correct score was 57.16% (SD = 12.58; range = 28–72%). On average, participants improved from baseline in all aided conditions. For the Soundhawk, participants improved on average by +12.64% (SD = 14.99%), +14.44% (SD = 17.28%), and +17.93% (SD = 13.52%) in the out-of-the-box (mean = 69.80%; SD = 11.12%; range = 54–90%), advanced-user (mean = 71.60%; SD = 13.51%; range = 51–87%), and audiologist-fit (mean = 75.09%; SD = 7.66%; range = 64–84%) conditions, respectively. Similarly, when using the Sound World Solutions CS-50+ device, on average participants improved by +3.45% (SD = 13.68%), +12.22% (SD = 11.62%), and +14.81% (SD = 9.12%) in the out-of-the-box (mean = 60.61%; SD = 10.42%; range = 44–74%), advanced-user (mean = 69.38%; SD = 10.73%; range = 51–86%), and audiologist-fit (mean = 71.97%; SD = 10.83%; range = 52–85%) conditions, respectively.

Table 2. Baseline Unaided AzBio Score and Change in AzBio Score Across Conditions.

		Change from Unaided Baseline, Percentage Points
		Soundhawk			Soundwolrd Solutions CS-50+
Participant	Baseline AzBio, Percentage Points	Out-of-the-box	Advanced-user	Audiologist-fit	Out-of-the-box	Advanced-user	Audiologist-fit
1	62.60	+12.20	+24.80	+24.80	-1.10	+19.10	+22.2
2	55.17	−1.16	+20.16	+20.16	−1.71	+5.79	+3.77
3	62.25	+18.04	+19.13	+19.13	+11.72	+23.28	+19.5
4	64.38	+7.47	+4.59	+4.59	+7.15	+4.23	+9.83
5	54.79	+15.55	+22.57	+22.57	+5.47	+16.01	+20.39
6	64.15	+2.51	−12.78	−12.78	+3.00	−1.39	+2.74
7	71.53	−9.27	−11.94	−11.94	−27.73	−5.53	+5.95
8	27.74	+30.48	+27.82	+27.82	+20.06	+23.59	+23.92
9	51.82	+37.96	+35.62	+35.62	+14.22	+24.89	+25.00
Mean (Standard Error; 95% Confidence Interval)	57.16 (4.19; 47.49–66.83)	+12.64 (5.00; 1.12–24.16)	+14.44 (5.76; 1.16–27.73)	+17.93 (4.50; 7.54–28.32)	+3.45 (4.56; −7.06–13.97)	+12.21 (3.87; 3.29–21.15)	+14.81 (3.04; 7.80–21.82)

On an individual level, several interesting trends were seen. First, four out of nine participants performed poorer with amplification compared with unaided baseline in at least one test condition. Decreased AzBio scores from baseline were more likely to be in the out-of-the box and advanced-user conditions (nine out of ten occurrences). In one instance, no difference from baseline was seen (participant 006 in the audiologist-fit condition with the Soundhawk). Second, some participants saw greater improvement with the two user-driven approaches compared with the audiologist-led approach. Participants 001, 005, and 009 had their largest improvement with amplification using either out-of-the-box or advanced-user fitting condition with the Soundhawk, while participants 003, 008, and 009 experienced their greatest improvement score with a user-fit condition with the CS-50+.

Table 3 displays the mean electroacoustic measurements for devices from the study and the manufacturer's reported specifications. Notably no device completely failed during the pilot study nor did electroacoustic measurements vary greatly between sessions. The same devices were used for all participants. For both PSAPs, the HFA OSPL90, frequency range, and THD mean measures were able to relatively approximate manufacturer specifications when available (OSPL90 was unspecified for CS-50+). However, the mean EIN values for the Soundhawk (mean = 41.57 dB; SD = 1.40 dB) and the CS-50+ (mean = 36.29 dB; SD = 5.59 dB) were notably higher than their manufacture specifications of less than 34 dB and less than 26 dB, respectively.

Table 3. Mean Electroacoustic Measurements Compared with Manufacturer Specifications.

	Soundhawk		Soundworld Solutions CS-50+
	Mean (Standard Deviation)	Manufacturer Specifications	Mean (Standard Deviation)	Manufacturer Specifications
Average OSPL90	93.71 dB SPL (2.06)	96 dB SPL	102.43 dB SPL (5.32)	Not Available
Frequency Range	< 200–6871 Hz	< 200–6720 Hz	< 200–7807 Hz	200–8000 Hz
Equivalent Input Noise	41.57 dB SPL (1.40)	< 34 dB SPL	36.29 dB SPL (5.59)	26 dB SPL
Total Harmonic Distortion	1.00% (0.58)	< 2%	1.29% (0.49)	≤1%

Total Harmonic Distortion is presented as an average of 500, 800, and 1600 Hz.

Table 4 displays accuracy of fittings, defined as matching target within ± 5 dB of NAL-NL2 prescribed target, at 0.5, 1, 2, and 4 kHz and the RMS value for three frequencies (0.5, 1, and 2 kHz) and four frequencies (0.5, 1, 2, and 4 kHz) across conditions. Notably, the audiologist-fit condition was the most accurate in meeting NAL-NL2 targets across all conditions using either the count of within ± 5 dB or RMS method. Fittings were most accurate at meeting target 0.5 kHz (83%), while fittings were least accurate at 4 kHz (28%). The four-frequency RMS values were highly influenced by the poor accuracy at 4 kHz. For this reason, we chose to explore the relationship between three-frequency RMS values and AzBio-aided improvement. Fig. 1 presents each participants' three-frequency RMS value as a function of aided improvement on the AzBio across conditions for the Soundhawk (left panel) and CS-50+ (right panel). Linear regression lines were fit for each condition and Pearson's correlation measures ( R ² ) are displayed within figure. The highest correlation, albeit relatively weak, was between the audiologist-fit three-frequency RMS values and aided improvement in AzBio sentence testing.

Table 4. NAL-NL2 Targets Met within 5 dB at 500, 1000, 2000, and 4000 Hz by PSAPs Across Conditions.

		NAL-NL2 Targets met within 5 dB
		Soundhawk			Soundwolrd Solutions CS-50+
Target Frequency	Total Targets Met by Frequency (%)	Out-of-the-box (± 5 dB/total)	Advanced-user (± 5 dB/total)	Audiologist-fit (± 5 dB/total)	Out-of-the-box (± 5 dB/total)	Advanced-user (± 5 dB/total)	Audiologist-fit (± 5 dB/total)
500 Hz	45/54 (83%)	6/9	9/9	7/9	7/9	8/9	8/9
1000 Hz	37/54 (69%)	5/9	6/9	5/9	6/9	7/9	8/9
2000 Hz	33/54 (61%)	7/9	5/9	7/9	4/9	4/9	6/9
4000 Hz	15/54 (28%)	4/9	1/9	4/9	2/9	1/9	3/9
Total Targets Met by Condition (%)		22/36 (61%)	21/36 (58%)	23/36 (64%)	19/36 (53%)	20/36 (56%)	25/36 (69%)

Figure 1.

The 3-frequency (500 Hz, 1000 Hz, 2000 Hz) root-mean-square (RMS) values were calculated for each participant, PSAP device: Soundhawk (top panel) and CS 50+ (lower panel), and fitting condition: out-of-the-box (in red), advanced-user (in blue), gold-standard (in black). Separate linear regression lines were fit to the data for each fitting protocol. Individual participants are represented by symbols to the right of each figure.

Discussion

In a pilot study of nine participants, an audiologist-driven fitting protocol provided the largest aided improvement in AzBio sentence test score from the unaided baseline score compared with two user-driven protocols for both PSAPs. In addition, the audiologist protocol produced the most accurate fittings based on meeting NAL-NL2 targets (±5 dB). Given the nature of this pilot study, inference is limited, and our results should be interpreted cautiously, as this study was not designed to test for inferential results. Nonetheless, these data contribute to the changing landscape and our growing knowledge concerning over-the-counter devices within the field of audiology.

In the present pilot study, perhaps the most important finding is the emergence of a trend whereby the involvement of the audiologist in the fitting of the PSAPs resulted in greater improvement in AzBio scores and a more accurate fitting based on NAL-NL2 targets. For both PSAPs, the out-of-the-box condition, which relied on the user's ability to interpret and adjust the device with only the basic materials provided by the manufacturer, resulted in the smallest improvements in AzBio scores. In requiring users to essentially “figure it out” on their own, this condition likely taxed the user's health literacy and practical self-management skills. This highlights the need to take self-management skills into account within the OTC HA model. In contrast, the audiologist-fit condition resulted in the largest improvements and the advanced-user condition sat in between them. At present, we are unsure if the differences seen in the current study represent clinically meaningful difference such that they would impact overall perception of amplification in the real world. It is possible that the out-of-the-box model represents an efficacious approach.

The aided improvement in AzBio scores seen in the current study was expected based on previous findings, which compared aided and unaided speech understanding scores using PSAPs and a hearing aid. ²⁹ In this earlier study, the devices were fitted using an audiologist-fitting protocol similar to that employed in the present study. It is interesting to note, however, that while the sample population in the current study was similar in nature (i.e., mild-to-moderate hearing loss and similar age group) to those individuals who participated in the earlier study, the mean improvement scores in the present sample were slightly larger than those found in the previous study. Specifically, there were mean aided improvements of 11.0 and 10.2% for the CS-50+ and the Soundhawk devices, respectively, in the earlier study versus improvements of 14.81 and 14.44% for these same two devices using an audiologist-fit protocol in the current study. We attribute these differences to directing the background noise immediately toward the PSAP rather than behind the user and the use of a sensation level approach for the presentation of the AzBio sentences in the current study rather than a fixed presentation level, which may have contributed to ceiling effects in the previous study. While speech understanding is a possible verification tool for amplification use, its generalizability is limited, and it does not reflect self-perceived validation of amplification. It is also important to note that there was heterogeneity in the amount of aided improvement in the AzBio scores seen in the current study. There were several persons who did not show any improvement in the aided condition, and there were others who were negatively impacted by the provision of a PSAP. This pattern of findings tended to occur most frequently for the out-of-the box as well as the advanced user protocols.

The electroacoustic results in the current study mostly agreed with those published by the manufacturer; however, EIN results were notably higher. This finding is in agreement with the variability in EIN measures with PSAPs reported in the earlier literature. Specifically, Smith et al ²⁵ reported EIN values of 39.04 and 34.4 dB for the Soundhawk and CS-50+, respectively, which are in agreement with those reported in the current study. However, another recent study reported much lower EIN values of 20 and 27 dB for the Soundhawk and CS-50++, respectively. ²⁴ Numerous factors, including the device measurement environment and the researchers (i.e., inter-rater reliability) may account for these differences. Electroacoustic analyses should be interpreted with caution for these reasons.

Based on the design of the study, it was expected that NAL-NL2 targets would be met most often by the audiologist-fit condition. In the audiologist-fit condition, approximating prescriptive targets was the goal, while in the other conditions, personal preference based on adjustment by the user drove the fitting. However, while no inference can be made from these results, an efficacy signal may be present, as closer approximation of NAL-NL2 targets resulted in the greatest AzBio sentence score improvement from baseline. This may be, in part, explained by the audiologist-fit condition most often approximating the 4-kHz target. Participants may have been less inclined to increase high frequencies given that low-frequency changes may have been more immediately noticeable compared with higher frequency changes.

The current pilot study is limited by its nature as it has a small, convenience sample. Due to this, we refrained from making inferential conclusions based on these data. Another possible limitation of this study is that all test conditions were performed in an audiometric test booth and it is unknown if results would carry over to a real-world setting using other verification and validation measures. While the order of AzBio sentence lists and test conditions were randomized, the unaided baseline was performed prior to any aided measure and it is possible practice effects impact the degree of improvement witnessed. Underlining the volatile nature of the OTC market, at the time of the conception and design of this study, the Soundhawk was commercially available; however, this device is no longer available. This finding somewhat limits the generalizability of the results of this study.

Future research in the area of OTC amplification is needed and could focus on comparative effectiveness of OTC HAs versus traditional hearing aids in more varied listening conditions, applied novel delivery models, and determining variables that contribute to precisely determining who is a good fit for OTC HA models versus traditional models. Furthermore, much of the previous literature has understandably relied on controlled OTC HA delivery models with input from the professional; however, it is important to elucidate the real-world variables that may affect clinical practice. Research should work to identify and characterize the different self-management characteristics of users associated with positive and negative OTC HA outcomes. Clinical research is highly difficult to execute given the numerous uncontrolled variables and outcome interpretation. However, future studies would benefit from adopting novel study methodologies such as equivalency and noninferiority studies powered on clinically meaningful differences to inform applied delivery service models.

Overall, the present pilot study offers insight and lays the groundwork for our group to design future studies exploring the impact of OTC devices and delivery models on important real-world outcomes. Audiologist-led delivery procedures appear to offer the most beneficial approach based on a speech understanding measure. It is unknown whether these trends would repeat in a larger sample size and whether the difference in improvement across conditions is large enough to make a clinical or perceptive impact in a real-world setting. Further research examining real-world condition and the ability of the consumer to fit without any influence from professionals is necessary to inform future audiologic care models.