Objective and Subjective Benefit of Direct-to-Consumer Hearing Devices in Middle-Aged Adults

Abstract

Purpose:

The purpose of this project was to assess subjective and objective benefit provided by several direct-to-consumer hearing devices for middle-aged adults. The primary goal of this study was to determine the extent to which this type of device can yield benefit when users are listening in a range of acoustic conditions, rather than to compare benefit among devices.

Method:

Participants (M _age = 58 years, n = 40) completed a speech perception task with and without one of four direct-to-consumer devices. Stimuli were presented with three types of maskers (steady-state noise, modulated noise, and competing talkers) at two different signal-to-noise ratios and two target levels. Participants also rated the effort required to complete the task with and without the devices and completed a short questionnaire about device comfort and perceived effectiveness.

Results:

The amount of objective benefit (in terms of speech recognition accuracy) varied among the four products, with two of the four devices yielding statistically significant benefit with medium-to-large effect sizes. Reduction in self-rated listening effort was noted from the use of all four devices, with a moderate effect size. Degree of hearing loss (4-frequency pure-tone average) was not significantly associated with the amount of either subjective or objective benefit. Responses to the posttask questionnaire indicated that many of the participants would be willing to use these or similar devices in the “real world.”

Conclusions:

Our findings support the concept that direct-to-consumer hearing devices have the potential to improve objective and/or subjective speech recognition in middle-aged adults, at least when fit to prescriptive targets. Benefit from these devices was not related to degree of hearing loss in this sample of participants.

Evidence is building to support the idea that hearing problems beginning in midlife can have serious consequences for the health and vitality of individuals as they age. Data from a growing number of epidemiological studies have identified links between hearing loss and cognitive decline (e.g., Amieva et al., 2015; Armstrong et al., 2020; Davies et al., 2017; Gao et al., 2020; Rong et al., 2020). The degree to which remediation of hearing loss can modify this association has yet to be determined. If treating hearing loss is shown to be effective in helping to reduce or delay cognitive decline, one major roadblock will be to address the notoriously low acceptance rate of hearing aids.

The long-standing problem of low uptake of hearing aids has gained the attention of federal entities. Perhaps the most well-publicized outcome of this attention is the report from the President's Council on Advisors on Science and Technology, Executive Office of the President (2015), which concluded that adults with no more than moderate age-related hearing loss would be better served if hearing aids were sold over the counter. This led to the recommendation that a separate category of hearing aids be developed, with the Food and Drug Administration (FDA) providing guidance on the use of these devices. The FDA Reauthorization Act of 2017 included the Over-the-Counter Hearing Aid Act to implement these recommendations.

Concurrently, the past decade has seen an increase in the availability of direct-to-consumer hearing devices (DTCDs), which vary widely in terms of quality, cost, and sound processing capabilities. Several recent reviews have summarized existing research on the efficacy of DTCDs (Jilla et al., 2018; Maidment et al., 2018; Manchaiah et al., 2017; Tran & Manchaiah, 2018). This body of research suggests that many less expensive products are problematic in terms of electroacoustic performance (e.g., inadequate high-frequency gain, excessive low-frequency gain, high levels of distortion, and/or internal noise). However, higher end DTCDs typically offer noise reduction and directionality and some (but not all) of these devices can meet electroacoustic and/or real-ear targets, especially for people with mild hearing loss (Brody et al., 2018; Reed et al., 2019; Reed, Betz, Kendig, et al., 2017; Smith et al., 2016).

Some DTCDs can give objective and subjective benefit that is comparable to hearing aids. For example, Reed, Betz, Lin, and Mamo (2017) found that four out of the five DTCDs they tested provided similar benefit for speech perception in noise as a hearing aid in a sample of older adults (M _age = 72 years). Brody et al. (2018) demonstrated that three mid- to high-end DTCDs were capable of improving speech recognition performance and reducing listening effort in older adults with mild-to-moderate hearing loss, although, in general, to a lesser extent than hearing aids. Cho et al. (2019) compared benefit of hearing aids and a DTCD in middle-aged adults with hearing loss ranging from mild to severe. They found about equivalent benefit on the majority of their outcome measures (a Korean version of the Hearing in Noise Test [Nilsson et al., 1994], performance on a dual-task, and pupillometry) for a DTCD and a hearing aid in their participants with mild-to-moderate hearing loss. Recently, Seol et al. (2021) found a hearing aid, a DTCD, and commercially available earbuds all led to statistically significant improvement in speech perception in quiet for older adults with mild-to-moderate hearing loss, although none yielded significant benefit for a measure of speech understanding in noise.

There also has been some research on the efficacy of fitting mild gain hearing aids on people with self-reported hearing problems and little or no pure-tone threshold elevation. Justification for this type of fitting is that noise reduction and/or directional microphone technology in these devices may be useful for people who report speech perception problems in noise but have minimal audiometric hearing loss. Roup et al. (2018) conducted a 4-week trial of hearing aid fitting with such individuals and found positive outcomes in terms of hearing handicap, self-reported auditory processing, and performance on the Speech Perception in Noise (SPIN) test (Kalikow et al., 1977) with target speech from the front and noise from the back. Notably, only three of their participants chose to purchase the hearing aids at the end of the trial. Similarly, Singh and Doherty (2020) reported a study of hearing handicap in middle-aged adults with normal pure-tone thresholds. Of those who had self-reported problems in noise, use of mild gain hearing aids led to a reduction in hearing handicap after 2 weeks. However, consistent with the results of Roup et al. (2018), only 20% of their participants with self-reported problems in noise indicated that they would consider purchasing a hearing device.

The work described in this article represents an exploratory study of the degree to which selected DTCDs can provide objective and subjective benefit for middle-aged adults. It departs from other studies of DTCDs in several ways: by focusing on middle-aged (rather than older) adults, who likely would constitute a large proportion of the target audience for these devices; by quantifying listening effort and obtaining subjective feedback about devices; and by measuring performance in a variety of acoustic conditions. Participants' understanding of sentences, presented at softer and louder levels, was tested in the presence of different types of maskers (steady-state noise, modulated noise, and competing speech) at two different signal-to-noise ratios (SNRs). Our primary goal was not to compare performance or benefit among devices; rather, we sought to determine the extent to which, when fit to prescriptive targets, this class of devices could improve speech recognition ability and decrease listening effort, and how this benefit was related to degree of hearing loss. Also of interest was examining participants' subjective impressions of the devices and determining whether they would consider using such devices in the real world.

Materials and Method

Participants

Forty-six middle-aged adults participated in this study. We placed announcements in community newsletters to identify people who had some trouble hearing but who felt “not quite ready” for hearing aids and did not currently use hearing aids. Preliminary exclusion criteria included not speaking English as a first language and a history of otologic or other health problems that impact hearing. This project was approved by the University of Massachusetts Amherst Institutional Review Board.

Audiometry and tympanometry were completed on all participants prior to administration of the experimental task. Participants also completed the Montreal Cognitive Assessment (MoCA; Nasreddine et al., 2005). One participant was excluded from the study based on asymmetric pure-tone thresholds, and five individuals were excluded as they scored below 26 on the MoCA. The age of the remaining 40 participants ranged from 50 to 64 years (M = 58.43 years). Participants' conventional 4-frequency pure-tone average (PTA; average of thresholds for 0.5 kHz – 4 kHz pure tones) was between 5 dB HL and 34 dB HL (M = 17.59 dB HL). Due to the large number of experimental conditions, participants were randomly assigned to be tested using one of four DTCDs (described below). Composite audiograms for each of these device groups are shown in Figure 1. One-way analyses of variance (ANOVAs) were conducted to identify whether the groups were equivalent in terms of age and PTA. Neither of these factors differed significantly among device groups (age: F(3, 36) = 0.58, p = .631; PTA: F(3, 36) = 0.18, p = .913).

Figure 1.

Composite audiograms for each direct-to-consumer hearing devices group. Dashed lines represent the highest and lowest pure-tone thresholds.

DTC Devices and Device Fitting

The devices chosen for this study were Sound World Solutions CS50+, Nuheara IQ Buds, Tweak Focus, and Bose Hearphones.¹ These four devices were selected based on the availability of directional microphones and volume control and/or frequency response fine-tuning capability. Measurement of each device's directional microphone function was performed with an Audioscan Verifit 2 by fitting the devices on a KEMAR mannequin and comparing the level of a 65 dB SPL speech signal presented from a front loudspeaker with a 65 dB SPL speech-spectrum noise presented from a loudspeaker at 180°. All devices chosen for inclusion in this study had directional microphones that provided some level of reduction in noise presented from the rear. A summary of the devices and their settings used in this study is shown in Table 1.

Table 1.

Devices and settings used in this study.

Device	Fit, form factor	In-situ hearing test	Settings
IQbuds	Occluding, ITE	Yes	“Focus” – directional mode enabled
			“SINC” (digital noise reduction) and “World EQ” graphic equalizer disabled
			Volume – N (default setting)
Tweak Focus	Open, thin-tube BTE	No	“Noisy” – directional mode enabled
CS50+	Occluding, BTE	Yes	“Restaurant” – directional mode enabled
Hearphones	Occluding, ITE	No	“Focused” – directional mode enabled

Note. ITE = in the ear; BTE = behind the ear.

DTCD Fitting and Verification Procedure

Each participant was randomly assigned to one of the four device groups. Participants were fit binaurally, with suitable dome sizes and wire length selected to ensure an appropriate physical fit. Participants assigned to the IQ Buds or CS50+ then performed an in-situ hearing test using the device's smartphone app. All devices were set to directional microphone mode (called “focus,” “noisy,” “restaurant,” and “focused” for the IQ Buds, Tweak Focus, CS50+, and Hearphones, respectively) for the remainder of the experiment.

Verification of each DTCD was performed with the Audioscan Verifit 2 Speechmap system using an on-ear probe microphone to measure output. The International Speech Test Signal (ISTS) was used to obtain real-ear unaided responses for 50 dB SPL input, followed by aided responses at 50 dB SPL and 65 dB SPL. The output of each device was adjusted so that the long-term average speech spectrum of the ISTS would approximate National Acoustic Laboratories–Non-Linear 2 (NAL-NL2) targets within ± 5 dB SPL where possible. For devices with limited fine-tuning ability, targets in the 1- to 2-kHz range were given preference. To confirm loudness comfort, swept tones were presented from 0.25 to 8 kHz at 85 dB SPL, followed by pink noise at 80 dB SPL. Participants rated the above stimuli on a 7-point loudness scale (7 = uncomfortably loud). If participants rated either loudness test as a 7, gain was reduced and the loudness test was repeated until a comfortable loudness level was reached. In cases where gain was reduced following the loudness comfort test, aided responses to ISTS were remeasured with the final device settings to be used in the speech perception task.

Speech Recognition and Listening Effort

Participants completed a listening task where they heard and repeated sentences from the name Theo, Victor or Michael (TVM) Colors corpus (Helfer et al., 2016). These stimuli were developed specifically to assess speech perception in the presence of competing speech, one type of condition included in this study. Sentences in this corpus that all begin with the name Theo, Victor or Michael have the same syntactic structure: Name found the color noun and the adjective noun here. A sample sentence is Theo found the pink menu and the true item here, with key words used for scoring denoted in bold. All sentences were recorded from a female talker. Each participant completed the task in both aided and unaided conditions, with half of participants in each device group beginning with unaided testing and the other half with aided testing; this factor was randomized within device group. Participants were instructed to repeat back the target sentence. Their responses were audio-recorded and analyzed off-line.

Target sentences were played from a loudspeaker located directly in front of the participant, and maskers were played from loudspeakers located 60° to the right and left. Maskers were either speech-shaped steady-state noise (SSN), speech-shaped envelope-modulated noise (SEM), or sentences from the TVM corpus spoken by other female talkers. Masking sentences began at a random point in the sentence, with the beginning of the sentence appended to the end, in order to assist participants in segregating the target sentence (which always began with a name) from the masking sentences. Target presentation level and SNR were both varied as well: Target sentences were presented at either 50 dB A or 65 dB A, and maskers were presented at levels corresponding to 0 dB SNR and −6 dB SNR; these SNRs have been shown to produce an appropriate range of performance in previous work from our lab (e.g., Helfer et al., 2016). There was a total of 24 test conditions (3 masker types × 2 target levels × 2 SNRs × 2 aided conditions), with nine trials (45 scoring words) per condition. Within each aided condition, other variables (masker, target level, and SNR) were blocked, and blocks were presented in random order. Testing began with four practice trials using each of the three maskers. At the end of each block of trials (aided and unaided), participants were asked to indicate how much effort they needed to expend to complete the task on a scale of 1 (very little effort) to 10 (a great deal of effort).

Data and Analyses

Initial statistical analyses (repeated-measures ANOVA) were conducted to characterize baseline speech perception ability in our participants across the acoustic conditions. Even though comparing benefit among the four devices was not a primary purpose of this study, we intended to present data broken out by device. In order for those data to be meaningful, we first needed to determine whether there were significant differences in unaided speech recognition between the four DTCD groups. We analyzed unaided and aided data (speech perception and effort ratings) averaged across acoustic conditions to identify the effect of DTCD, using repeated-measures ANOVA. Effect size was calculated as partial eta squared (η_p ²). Finally, in an attempt to explain some of the large degree of individual variability in the data, we examined the association between average pure-tone thresholds and benefit, and the association between average pure-tone thresholds and willingness to use a DTCD in the real world, via Pearson r correlation analyses.

Results

Electroacoustic Analysis

Mean deviation from real-ear NAL-N2 targets for each device is shown in Figure 2. All four devices were able to reach targets within ± 10 dB for both softer (50 dB SPL) and louder (65 dB SPL) input levels. The output of three of the devices was consistently within 5 dB of targets between 750 Hz and 3000 Hz.

Figure 2.

Mean deviations from National Acoustic Laboratories–Non-Linear 2 (NAL-NL2) real-ear targets for each direct-to-consumer hearing device. Upper panel: 50 dB SPL input; bottom panel: 65 dB SPL input.

Unaided Speech Recognition

Unaided objective performance (percent correct recognition of target words) and aided objective performance by device can be seen in Figure 3. First, we analyzed unaided speech recognition ability to characterize baseline performance among participants in each device group. Repeated-measures ANOVA with input level (50 dB A vs. 65 dB A), SNR (−6 dB or 0 dB), and masker (SSN, SEM, or competing speech) as within-subject factors and device as the between-subjects factor found significant main effects of input level, F(1, 36) = 15.96, p < .001, η_p ² = .31; SNR, F(1, 36) = 570.37, p < .001, η_p ² = .94; and masker, F(2, 35) = 15.03, p < .001, η_p ² = .46. The Level × Masker interaction was significant as well, F(2, 35) = 4.66, p = .016, η_p ² = .21. Post hoc tests (one-way ANOVA with Bonferroni correction) indicated that the perception of softer target speech was significantly poorer in the presence of competing speech than in the presence of either of the noise maskers. For louder speech targets, recognition scores were lower in the presence of competing speech than in the SEM noise. Moreover, perception of louder versus softer speech only differed significantly when the masker was competing speech.

Figure 3.

Percent correct speech recognition for each direct-to-consumer hearing devices group for softer (50 dB SPL) and louder (65 dB SPL) target speech stimuli by signal-to-noise ratio (−6 dB and 0 dB) and masker type. Error bars represent the standard error. SSN = steady-state noise; SEM = speech-shaped envelope-modulated noise.

Recall that participants were assigned randomly to one of the four DTCD groups. There were small differences in unaided performance among these groups, as shown by a significant Level × Device interaction, F(3, 36) = 4.68, p = .007, η_p ² = .28. After correction for multiple tests, there were no significant post hoc comparisons. Nevertheless, since the groups of participants assigned to each device did not have equivalent performance for unaided speech recognition, differences between devices in aided benefit (discussed below) should be interpreted with caution.

Speech Recognition Benefit

To determine the extent to which the devices led to an improvement in speech recognition, repeated-measures ANOVA was conducted on the percent correct speech recognition scores averaged across all acoustic conditions with aided versus unaided condition as the within-subject variable and device as a between-subjects variable. Both the main effect of aided condition, F(1, 36) = 20.79, p < .001, η_p ² = .37, and the interaction of aided condition with device, F(3, 36) = 4.22, p = .012, η_p ² = .26, were statistically significant. Post hoc ANOVA indicated that the difference between aided and unaided speech recognition was statistically significant for only two of the devices (IQ Buds and Hearphones) with medium-to-large effect sizes, IQ Buds: F(1, 9) = 14.87, η_p ² = .62; Hearphones: F(1, 9) = 15.47, p = .003, η_p ² = .63.

Benefit in speech intelligibility (percent correct aided – percent correct unaided) for individual participants is shown in Figure 4. It is clear that there is a large amount of individual variability in these data. With only 10 participants per group, it is feasible that we did not have a sufficient number of participants to identify statistically significant benefit for all of the devices. However, benefit for the CS50+ and Tweak devices averaged only 1%–2%, so it is unlikely that increasing the number of participants would have changed the outcome. Mean benefit for the other two devices was 7.53% for the IQ Buds and 10.24% for the Hearphones, with some participants receiving more than 20% benefit.

Figure 4.

Objective and subjective benefit from individual participants. Boxes correspond to the interquartile range, with the horizontal line within the box indicating the median. Whiskers show the distribution of the nonoutlier values. Note that greater effort benefit = larger reduction of benefit ratings/more negative values.

Subjective Effort

Participants rated the effort needed to complete the task at the end of each of the two major blocks of trials (unaided and aided) using a scale of 1 (very little effort) to 10 (a great deal of effort). Benefit derived from the DTCDs for effort ratings (aided effort ratings – unaided effort ratings) is shown in Figure 4. Repeated-measures ANOVA on the effort ratings with aided versus unaided condition as the within-subject factor and device as the between-subjects factor showed a significant main effect of aided condition, F(1, 36) = 30.49, p < .001, η_p ² = .46, with nonsignificant main and interaction effects involving device, main effect: F(3, 36) = 0.77, p = .519, η_p ² = .06; interaction: F(3, 36) = 0.55, p = .653, η_p ² = .05. Participants rated the speech perception task as being less effortful when using any of the four DTCDs, as compared to unaided listening.

Subjective Evaluation of Devices

After completing the listening task, participants filled out a brief survey asking about how useful and comfortable they found the device. Responses to this survey are summarized in Table 2.

Table 2.

Responses to the posttesting survey.

Note. Each asterisk represents one participant’s data, colorcoded by response (green = positive, black = neutral/maybe, red = negative, purple = no response). Q1: Do you think this device would be helpful in the “real world”? Q2: How willing would you be to use this device in the “real world”? Q3: Would you recommend this device to others? Q4: How comfortable did you find wearing this device?

Even though benefit was relatively modest for many of our participants in terms of percent correct performance on our speech recognition task, feedback about the devices was generally positive. Notably, more than half of the participants indicated willingness to use the device in the real world (i.e., they responded to, “How willing would you be to use this device in the real world?” with “very willing” or “somewhat willing”). Table 3 displays each participant's specific response to this question, sorted by PTA, along with open-ended comments (prompted by, “Please let us know any other opinions you have about this device”).

Table 3.

Open-ended comments and participants' willingness to use the devices in the “real world,” sorted by 4-frequency (0.5 kHz–4 kHz) pure-tone average.

Mean pure-tone average (dB HL)		Willingness to try the device in the real world	Comments
Normal hearing listeners	5.0	somewhat willing	Noticed circuit noise
	5.0	very willing
	6.5	very unwilling	Heard better without; not willing to use this device since they didn't experience benefit, but in principle another device might be useful
	6.5	somewhat willing
	7.0	somewhat willing	Task definitely easier with device
	7.0	very unwilling	Maybe would work for other people
	9.0	somewhat unwilling	Not sure it helps
	9.0	somewhat willing	Depends on what it looks like and the price; don't know if they would get used to it; “surprised by how much it cleared something away”
	9.5	somewhat willing
	10.0	very unwilling	Objected to cosmetics; didn't notice benefit
	11.0	somewhat willing	Issues with retention in ear
	11.5	somewhat willing	Would be more willing if hearing was worse
	11.5	very willing	Generally averse to headphones, but would use if serious benefit seen
	13.0	maybe	Sound was natural; if beneficial, slight discomfort would be worth it
	13.0	very willing
	13.5	somewhat willing
	14.5	very willing	Seems useful; wouldn't wear all the time
Slight hearing loss listeners	15.5	somewhat willing	Noticed circuit noise; wouldn't use in all situations (“maybe at a party”)
	17.5	somewhat willing	Moving made it uncomfortable; easier to hear without it in
	17.5	somewhat willing	Task slightly easier with device
	18.0	somewhat unwilling	Noticed circuit noise; a little clearer with them in
	18.0	very unwilling	Anticipated task would be easier with device but didn't find it so; may be different in natural conversation
	18.5	very unwilling	Didn't feel secure in ear
	20.0	somewhat unwilling
	20.0	somewhat willing	Helpful in a way, but couldn't get directionality; would see if it helped; might get used to it; “It's not like they're not there”
	21.5	somewhat willing	Noticed circuit noise; would use if it was helpful and comfortable after getting used to it
	22.0	very willing
	22.5	very willing	Definitely made task easier
	24.0	somewhat willing
	24.5	somewhat unwilling
	24.5	somewhat unwilling	Sounded artificial and tinny, like a TV
	25.0	very willing	Made a huge difference
Mild hearing loss listeners	25.5	somewhat unwilling	Comfort vs. benefit not great; wish they were smaller
	26.5	somewhat unwilling	Didn't notice a big difference; task didn't seem any easier; device clunky and bulky
	27.0	somewhat willing	Was OK in controlled setting but wouldn't use it for long periods; would use if more comfortable; more help on one side than the other
	27.5	somewhat unwilling	Probably not useful because of size
	27.5	somewhat unwilling	Don't feel like they need it
	32.0	very willing	Could barely tell they were in; might become more useful with more use
	32.5	very willing	Didn't think it made a difference; may work better in the real world
	33.5	very willing	Very willing to try; seems different from regular hearing; seemed to increase “ability to distinguish [different sounds]”; gave “clarity”

Influence of Degree of Hearing Loss

We anticipated that individuals with more hearing loss might derive greater benefit from DTCDs. We completed a Pearson r correlation analysis to assess the associations between PTA, percent correct benefit averaged across all acoustic conditions, and effort benefit (unaided effort ratings – aided effort ratings) averaged across all conditions. Neither objective benefit (r = .16, p = .33) nor effort benefit (r = −.06, p = .73) was significantly associated with pure-tone thresholds.

Also of interest was determining the degree to which amount of hearing loss influenced participants' willingness to use a DTCD in the real world. In order to address this question, we conducted a point–biserial correlation analysis to analyze the association between participants' PTA and their responses to the prompt, “How willing would you be to use this device in the real world?” This analysis revealed a nonsignificant correlation between these two variables (r = .25, p = .14). With data from 40 participants included in these two analyses, we should have been able to detect any clinically meaningful correlation effect sizes.

Discussion

The results of this study support the concept of DTCDs' potential for middle-aged adults with no more than a mild hearing loss. All four of the devices tested were able to get within 10 dB (and, in many cases, 5 dB) of NAL-NL2 targets, and the use of all four devices led to a reduction in self-rated listening effort. Our results extend the outcomes of prior research showing the potential effectiveness of DTCDs (Brody et al., 2018; Cho et al., 2019; Reed, Betz, Kendig, et al., 2017; Seol et al., 2021) for middle-aged adults listening in a broader range of acoustic conditions. In this study, only two of the four devices were able to provide statistically significant benefit (with medium-to-large effect sizes) for objectively measured speech recognition ability in noise. Taken with results from other studies (Brody et al., 2018; Seol et al., 2021), it appears that some DTCDs may provide only limited objective benefit in the presence of noise.

Correlation analyses suggested that there was not a linear relationship between degree of hearing loss and benefit. However, the four participants who obtained the greatest amount of objective benefit (≥ 15% benefit in speech intelligibility averaged across all acoustic conditions) did have at least one thing in common: They all had pure-tone averages of between 7 dB HL and 18 dB HL. All four of these participants (one from the IQ Buds group and three from the Hearphones group) indicated that they were either somewhat willing or very willing to try their device in the real world. This finding suggests that DTCDs (at least the specific devices tested in this study) may give maximum objective benefit in noise to participants with very mild hearing loss. Nevertheless, it could be that factors besides (or in addition to) pure-tone thresholds contributed to the objective benefit obtained by these individuals and their willingness to try DTCDs.

It should be kept in mind that the acoustic conditions sampled in this study represent a miniscule proportion of the types of listening situations individuals find themselves in during their daily lives. Each of the devices used in this study had its own idiosyncratic pattern regarding the acoustic conditions in which it yielded the most benefit, likely related to differences in underlying processing/technology. Consequently, finding the optimal device for an individual listener is likely to involve some degree of trial and error, depending on the nature of the listening environments that individuals find most challenging, and how frequently they are in those environments.

Open-ended comments from participants yielded both positive and negative feelings about the devices. One consistent theme in these responses was that the devices might be more beneficial once the individual gets used to them. Considering our finding that different devices work best under different acoustical conditions, it seems essential to promote the idea that DTCDs be sold with generous return policies.

Given the very modest average benefit (1%–2%) in terms of performance on the speech recognition task for two of the devices, it is perhaps surprising that at least half of the participants in each device group stated that they would be willing to try them in the real world. Recall that participants in all four device groups also found listening less effortful when using the device (although it should be kept in mind that participants' aided effort ratings may have been influenced by the placebo effect). As with objective benefit, willingness to try a device was not linearly related to amount of hearing loss. In fact, participants who indicated that they were very willing to try the devices in the real world varied widely in terms of hearing (PTA between 5 dB HL and 34 dB HL). These findings emphasize the importance of considering factors other than degree of pure-tone hearing loss and objective improvement in speech perception when making clinical decisions regarding recommendations.

Several limitations to this study should be kept in mind. The large number of experimental conditions necessitated a between-subjects design, so it is not appropriate to compare results of the specific devices used here, as there were small differences in unaided speech recognition among the device groups. The number of participants in each device group was relatively small and it is possible that significant differences between aided and unaided performance would have been obtained for all four devices if we had a larger n (although this seems unlikely, given the very modest objective benefit found for two of the devices). All devices were fit by a researcher with hearing aid fitting experience and were fine-tuned using real-ear measurement. Hence, it is likely that results noted here are best-case scenarios, compared to what would be obtained by an unexperienced user tasked with self-fitting the devices (Reed et al., 2019). On the other hand, participants were not allowed any time to acclimate to the devices and greater benefit might be seen if they had been tested after a trial period of use (e.g., Glick & Sharma, 2020; Ng et al., 2014).

In summary, results of this study suggest that some DTCDs can give substantial objective and/or subjective benefit for middle-aged adults listening in noise, at least when they are customized to prescriptive targets. Moreover, many of our participants indicated a willingness to try these devices in the real world. There does not appear to be a linear relationship between amount of hearing loss and benefit; rather, our data suggest that these devices might be maximally beneficial in noise for individuals with slight or mild hearing loss.

Funding Statement

This work was funded by National Institutes of Health (NIH) Grant R01 DC01257 (awarded to Karen S. Helfer), and NIH Grant K23 DC016855 (awarded to Sara K. Mamo).