What’s in a name? A systematic review and meta-analysis to assess the effectiveness of non-medical amplification devices in adults with mild and moderate hearing losses

Abstract

Objective

To assess non-medical amplification devices in adults with mild-to-moderate hearing loss, and the impact of device features on outcomes.

Design

A prospectively registered systematic review.

Study sample

Ten studies evaluating personal sound amplification products (PSAPs), and four evaluating smartphone amplification applications (or apps). Devices were classified as “premium” or “basic” based on the number of compression channels (≥16 or <16, respectively).

Results

Meta-analyses showed that premium PSAPs improved speech intelligibility in noise performance compared to unaided, whereas basic PSAPs and smartphone apps did not. Premium PSAPs performed better than basic hearing aids. Premium hearing aids performed better than premium and basic PSAPs, smartphone apps, and basic hearing aids. Although data could not be pooled, similar findings were also found for quality of life, listening ability, cognition, feasibility, and adverse effects.

Conclusions

Premium PSAPs appear to be an effective non-medical amplification device for adults with mild-to-moderate hearing loss. Given the overlap in features available, it may be that this is a key consideration when drawing comparisons between devices, rather than the device being named a PSAP or hearing aid. Nevertheless, the extent to which PSAPs are effective without audiological input remains to be determined.

Keywords:

• Adult aural rehabilitation
• amplification
• hearable
• personal sound amplification product (PSAP)
• smartphone application (app)

Introduction

Hearing aids are currently the primary clinical intervention for adults with mild-to-moderate hearing loss (MMHL), improving listening abilities, as well as hearing-specific and general health-related quality of life (Ferguson et al. 2017). Globally, hearing aid adoption rates have increased over the past decade. For example, in the United Kingdom (UK), with a publicly-funded hearing healthcare system, hearing aid adoption rates have risen from 37% in 2009 to 53% in 2022 (Anovum 2022). Similarly, in the United States of America (USA), where the cost of hearing healthcare is not always reimbursed by medical insurance, adoption rates rose from 25% in 2008 to 38% in 2022 (Powers and Bisgaard 2022). Despite this, it is apparent that most individuals worldwide who would benefit from hearing aids do not adopt them. As such, there is a need to identify suitable intervention strategies that will enable and empower individuals to manage their hearing difficulties successfully.

In recent years, numerous alternative amplification devices to conventional hearing aids have become commercially available, which potentially offer more affordable and accessible options to address hearing difficulties (Olson, Maidment, and Ferguson 2022). Alternatives to conventional hearing aids include over the counter (OTC) hearing aids, personal sound amplification products (PSAPs), hearables, and smartphone amplification applications (or apps). All alternatives can be obtained without consultation with a hearing healthcare professional (National Academies of Sciences, Engineering & Medicine 2016). Furthermore, some alternatives are deemed suitable for MMHL only (e.g. OTC hearing aids), whilst others can only be marketed to overcome listening difficulties in certain environments (e.g. PSAPs). Hearables are electronic in-ear devices that are not exclusively marketed for hearing loss but are designed for multiple purposes, such as activity tracking and music listening. To enable amplification, hearables typically need to be paired to a smartphone and used in conjunction with a sound amplification app (for characteristics, see Ferguson et al. 2023). More recently, some hearables now include additional, in-built sound amplification features (e.g. Apple AirPods, Samsung Galaxy Buds Pro).

Arguably, the proliferation of alternative devices to conventional hearing aids has largely been driven by legislative changes in the USA. In 2016, the National Academies of Sciences, Engineering and Medicine published a report outlining key priorities for improving access to and affordability of hearing healthcare for adults. One of these priorities was to expand options regarding hearing-based technologies, including OTC hearing devices. Subsequently, the OTC Hearing Aid Act, which was part of the US Food and Drug Administration (FDA) Reauthorization Act of 2017, was introduced. This Act aimed to enable adults with perceived MMHL to access OTC hearing aids – a new category of hearing aids that consumers can buy directly, circumventing the need for a medical examination or an appointment with an audiologist (Warren and Grassley 2017). In August 2022, the FDA finalised the regulations, and from October 2022, OTC hearing aids have been available for purchase in the USA. Critically, several trials have shown that, compared to audiologist-fitted hearing aids, OTC hearing aids can result in similar outcomes, including self-reported benefit and satisfaction, as well as speech-in-noise performance (De Sousa et al. 2023; Humes et al. 2017; Humes et al. 2019; Sabin et al. 2020).

Given these recent legislative changes, it is perhaps unsurprising that there has been a growth in research assessing the effectiveness of other alternative amplification devices not regulated as medical devices. In 2018, the first systematic review and meta-analysis in this area was published (Maidment et al. 2018), broadly defining alternative amplification devices as non-medical standalone products (e.g. PSAP, smartphone amplification app) or assistive listening devices (ALDs) that provide additional functionality to a conventional hearing aid (e.g. remote microphone systems). Only two studies assessed non-medical standalone products (Reed et al., 2017; Sacco et al. 2016), with a meta-analysis suggesting that PSAPs improved speech intelligibility in noise performance relative to unaided conditions. Further pooled comparisons could not be made due to heterogeneity in outcomes, and the evidence was judged to be subject to bias due to limitations in the before-after study designs employed. A further systematic review comparing PSAPs and conventional hearing aids has been published (Chen et al. 2022). Data pooled across five studies (Brody, Wu, and Stangl 2018; Cho et al. 2019; Choi et al. 2020; Seol et al. 2021; Kim et al. 2022) showed no statistical differences between PSAPs and conventional hearing aids for speech intelligibility in noise, sound quality, and listening effort. As a result, Chen et al. (2022) concluded that PSAPs are potentially just as beneficial as hearing aids.

Nevertheless, the findings of previous systematic reviews should be interpreted with caution given that device features can impact outcomes. For example, mixed results have been found for a range of outcomes (e.g. speech understanding, listening effort, hearing-related quality of life, preference ratings, localisation) when hearing aids are classified as “premium” relative to “basic” based on the features available, such as the number of compression channels, type of directional microphone, binaural data streaming, and noise reduction (Cox, Johnson, and Xu 2016; Johnson, Xu, and Cox 2016, 2017; Prakash et al. 2022). In contrast, comparable speech intelligibility performance between PSAPs and hearing aids has only been found for PSAPs that retail ≥ US$299.99 (Reed et al. 2017). This latter finding likely reflects the fact that higher-priced PSAPs potentially offer superior amplification technology compared to those available at a much lower price-point (i.e. < US$299.99). While Chen et al. (2022) reported several sensitivity analyses comparing different hearing aids and PSAPs based on the features available, they did not provide specific comparisons between premium and basic devices based on the classifications established by Johnson, Xu, and Cox (2016). In addition, neither review included studies assessing sound amplification apps or hearables.

Based on the caveats of the existing systematic reviews, the current review had two aims. First, to provide an up-to-date review of the evidence assessing a broad range of non-medical amplification devices. Second, to assess the impact of device features between devices on outcomes.

Materials and methods

Prior to commencing the systematic review, the protocol was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO), registration number: CRD42022324916. Methods are reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) checklist (Page et al. 2021).

Eligibility criteria

The inclusion criteria were specified in the protocol in terms of participants/population, intervention(s), comparator(s)/control, outcomes, and study designs (PICOS) as follows.

Participants/population

Adults (≥18 years) with MMHL, defined as an average audiometric hearing threshold across octave frequencies 0.5–4 kHz in the better hearing ear ≥20 and <50 dB HL (World Health Organisation 2021). Studies that included moderate-severe, severe, or profound hearing losses were not included unless data for each grade was reported separately. Studies were included where the mean average hearing threshold fell within the specified range or if only qualitative descriptions of hearing threshold were provided with no audiometric data. Studies with bilateral, unilateral, conductive, and mixed hearing losses were included.

Intervention(s)

Any alternative listening device to conventional hearing aid was included. Alternative amplification devices were defined as standalone products that are not regulated medical devices (e.g. PSAPs, sound amplification apps, hearables). All devices should aim to improve hearing and communication outcomes in people with MMHL through the amplification of external sound sources. These devices do not include ALDs (e.g. remote microphone systems) or OTC hearing aids given that they are regulated as medical devices by the FDA.

Comparator(s)/control

Either passive/inactive (e.g. unaided, usual care, waiting list) or active (e.g. conventional hearing aid, non-medical amplification device). Hearing aids are defined as a regulated medical device that deliver electroacoustic amplification via air conduction, irrespective of where they are worn (e.g. behind the ear, in the ear), including hearing aids delivered through an OTC model. Studies with analogue hearing devices or bone conduction hearing aids were not included.

Outcomes

Primary outcomes were: (i) behavioural measures of speech intelligibility in noise; (ii) hearing-specific health-related quality of life (QoL), where participation is the key domain, measured using any self-report questionnaire; and (iii) adverse events (e.g. related to device fitting, such as pain, discomfort, tenderness, skin irritation, or ear infection; related to over-amplification, such as noise induced hearing loss). Secondary outcomes included: (i) health-related QoL; (ii) listening ability; (iii) cognition (e.g. working memory); and (iv) feasibility (e.g. usability, adherence).

Study designs

Retrospective or prospective studies, randomised or non-randomised controlled trials, and before-after studies were included. Articles reporting expert opinions, practice guidelines, case reports, case series, conference abstracts, book chapters, and non-peer-reviewed research (i.e. grey literature) were excluded.

Search strategy

KN searched the following databases on 27th April 2022: CINAHL (via EBSCOhost), ClinicalTrials.gov, Cochrane Library, EMBASE (via Ovid SP), ISRCTN Registry, MEDLINE (via Ovid SP), PubMed, Scopus, Web of Science, and WHO International Clinical Trials Registry Platform (ICTRP). The searches were updated on 12th June 2023, to ensure that recently published papers were considered before the final analysis. Full electronic search strategies are provided in Supplementary Materials 1. Database searches were completed in one day, with the following date/time restriction, March 2018 to current, to capture publications released after the final search conducted in a previous systematic review (Maidment et al. 2018). There were no language, document type, or publication status limitations. Hand searching the last six months of publications from key audiology journals, snowballing of the reference lists from included studies, and screening of related articles by shortlisted authors, were all employed to identify any relevant articles that may not have been returned by the database searches.

Study selection

Two investigators independently assessed the identified references to decide eligibility by reading the title and/or abstract using Rayyan (www.rayyan.ai). KN assessed all references, with three co-authors (DWM, MAF, RJB) assessing one-third each. The full text was obtained for articles that appeared to meet eligibility or where there was any uncertainty (i.e. insufficient information to make a clear decision). Contact with study authors was not needed to resolve questions concerning eligibility. Discrepancies were resolved through discussion between all co-authors.

Data collection process

A standardised data collection form constructed via Covidence (www.covidence.org) was piloted, which included study details (e.g. sponsorship source, country, setting), author’s contact details (name, institution, email, postal address), study design, population (inclusion/exclusion criteria, baseline characteristics), interventions, comparators, and outcomes. Data extraction was conducted by KN and RJB independently but in duplicate for every included record. If data were presented in graphical form, results were estimated from figures using WebPlotDigitizer (automeris.io/WebPlotDigitizer). Disagreements about numerical data extracted from figures were resolved by averaging.

Risk of bias in individual studies

DWM and KN independently assessed the risk of bias of each included study with the Cochrane risk of bias tool (Higgins and Green 2011), which rates studies as high risk, low risk or unclear risk in relation to generation and concealment of the allocation sequence, blinding of participants and study personnel (performance bias), blinding of outcome assessors (detection bias), incomplete outcome data, selective outcome reporting, and other sources of bias (e.g. influence of funders). Studies were also assessed for study quality using the Downs and Black (1998) checklist, which consists of 27 criteria. Criterion 27 was adapted to consider whether a power calculation was performed rather than whether there was sufficient power to detect a clinically meaningful change. All criteria were scored 0 (no; unable to determine) or 1 (yes), except for criterion 5, which was scored 0 (no), 1 (partially), or 2 (yes). The total maximum score was 28, with study quality rated as excellent (26–28), good (20–25), fair (15–19) or poor (≤14) (Hooper et al. 2008).

Data synthesis

Meta-analyses were only performed when included studies were broadly comparable in terms of study design, interventions, and outcomes. For continuous data, as different outcome measures were employed across studies, effect sizes (ES) were calculated as standardised mean differences (SMDs); values of 0.2–0.5 are considered small, 0.5–0.8 medium, and > 0.8 large (Cohen 1988). Heterogeneity in effect sizes across studies was examined using the I² statistic and its significance was tested using a Q test. Heterogeneity ranged was from 0 to 100%, with low (0–40%), medium (41–60%) and high (61–100%) ranges (Higgins and Green 2011). All meta-analyses were performed in R studio software using the metafor package (Viechtbauer 2010). Where studies only reported the median and interquartile range, these were estimated using the estmeansd package (McGrath et al. 2020). For each meta-analysis, studies were weighted according to their contributions to the pooled estimate by calculating the inverse variances of their effect estimates. This statistical method gives studies with more precise results (narrower confidence intervals) more weight (Higgins and Green 2011). In the absence of meta-analysis, primary and secondary outcomes were assessed at the individual study level through narrative synthesis (Popay et al. 2006).

Results

A total of 1,741 records were identified for screening. Following the removal of 625 duplicate publications, 1116 records were subjected to a three-stage screening process (Supplementary Materials 2). The full texts of 57 articles that passed the initial title and abstract screen were retrieved. Forty-five articles were not judged to have met the inclusion criteria and were excluded. Twelve studies were included in the review, plus two studies (Reed et al. 2017; Sacco et al. 2016) assessing PSAPs carried-over from a previous systematic review (Maidment et al. 2018).

Summaries of the characteristics of the 14 studies and devices are provided in Supplementary Materials 3 and 4, respectively. Two categories of non-medical devices were evaluated: (i) PSAPs (n = 10), and/or (ii) sound amplification apps coupled with hearable or wired in-ear headphones (n = 4). Comparators included unaided, conventional (air conduction) hearing aids, and/or another non-medical amplification device. Four studies assessed outcomes after trialling the intervention in everyday life. Nine studies tested devices in the laboratory, with no real-world listening for acclimatisation.

PSAPs

Speech intelligibility in noise

Eight studies tested speech intelligibility in noise (Brody, Wu, and Stangl 2018; Cho et al. 2019; Choi et al. 2020; Kim et al. 2022; Reed et al. 2017, 2019; Sacco et al. 2016; Seol et al. 2021). Two studies (Cho et al. 2019; Kim et al. 2022) presented data for mild hearing loss (MHL) and moderate hearing loss (MDHL) separately.

Hearing aids and PSAPs were classified as either “premium” or “basic” based on the number of compression channels (≥16 or <16, respectively) (Johnson, Xu, and Cox 2016). For some PSAPs, the number of channels was not specified and could not be obtained from publicly available manufacturer materials (see Supplementary Materials 4). As a result, these devices were not included in any meta-analysis.

All devices were fitted by either a hearing healthcare professional or a researcher. However, one study (Reed et al. 2019) assessed the same PSAPs across three different conditions (i.e. “out of the box”, “advanced users”, “audiologist fitted”). For data to be comparable across studies, only the outcomes for the “audiologist fitted” devices were included. For studies evaluating several devices, performance across premium and basic devices was initially pooled to provide conservative estimates of effect. Where possible, separate comparisons between premium and basic devices were also undertaken. Where studies evaluated several, similar devices, performance was first pooled across devices, with only one value entered in the meta-analysis. Summary effects are provided in Supplementary Materials 5.

PSAP (pooled) vs. Unaided

Across eight comparisons, speech performance favoured PSAPs compared to unaided (Figure 1(A)). The overall effect (n = 176 participants) was statistically significant (p = 0.013; ES = −0.31), and heterogeneity was low (I² = 24%).

Figure 1. Summary of the random effects meta-analyses for speech intelligibility in noise: (A) PSAP (pooled) vs. Unaided; (B) PSAP (premium) vs. Unaided; and (C) PSAP (basic) vs. Unaided. As different outcome measures were employed across studies, effect sizes were calculated as standardised mean differences (SMDs). Black squares = summery effect size of each study. Error bars = 95% confidence intervals (CI) for the summery effects. Diamond = overall effect size, lateral points indicate 95% CI for overall effect estimate.

PSAP (premium) vs. Unaided

Across five comparisons, performance mostly favoured premium PSAPs compared to unaided (Figure 1(B)). The overall effect (n = 118) was statistically significant (p = 0.032; ES = −0.48), and heterogeneity was high (I² = 61%).

PSAP (basic) vs. unaided

Results were mixed; three comparisons favoured basic PSAPs, whereas two comparisons showed the opposite result (Figure 1(C)). The overall effect (n = 125) was not statistically significant (p = 0.099; ES = −0.21), and heterogeneity was low (I² = 0%).

PSAP (pooled) vs. hearing aid (pooled)

Out of seven comparisons, five favoured PSAPs and two favoured hearing aids (Figure 2(A)). The overall effect (n = 167) was not statistically significant (p = 0.319; ES = −0.14), and heterogeneity was medium (I² = 35%).

Figure 2. Summary of the random effects meta-analyses for speech intelligibility in noise: (A) PSAP (pooled) vs. Hearing aids (pooled); (B) PSAP (premium) vs. Hearing aids (basic); and (C) PSAP (basic) vs. Hearing aids (premium); (D) PSAP (basic) vs. Hearing aids (basic).

PSAP (premium) vs. hearing aid (premium)

Only one study by Reed et al. (2017) (n = 42) included this comparison, showing that performance favoured a premium hearing aid (M = 88.4; SD = 12.52) compared to premium PSAPs (M = 86.07; SD = 12.61).

PSAP (premium) vs. hearing aid (basic)

Across three comparisons, performance favoured premium PSAPs compared to basic hearing aids (Figure 2(B)). The overall effect (n = 67) was statistically significant (p = 0.018; ES = −0.41), and heterogeneity was low (I² = 0%).

PSAP (basic) vs. hearing aid (premium)

Three comparisons favoured premium hearing aids compared to basic PSAPs (Figure 2(C)). The overall effect (n = 81) was statistically significant (p = 0.006; ES = 0.44), and heterogeneity was low (I² = 9%).

PSAP (basic) vs. hearing aid (basic)

Out of four comparisons, three showed that performance favoured basic PSAPs, with one marginally favouring basic hearing aids (Figure 2(D)). The overall effect (n = 83) was not statistically significant (p = 0.205; ES = −0.20) and heterogeneity was low (I² = 0%).

A synthesis for other outcomes (i.e. hearing-specific health-related QoL, adverse events, health-related QoL, listening abilities, cognition, feasibility) is presented in Supplementary Materials 6, as meta-analysis was not possible.

Sound amplification apps (used with wired in-ear headphones or hearable)

Speech intelligibility in noise

Four studies tested speech intelligibility in noise (Han et al. 2022; Lin et al. 2022; Martinez-Beneyto et al. 2020; Seol et al. 2021), with mixed results. Where meta-analyses were possible, summary effects and forest plots are provided in Supplementary Materials 5 and Figure 3, respectively.

Figure 3. Summary of the random effects meta-analysis for speech intelligibility in noise: (A) Sound amplification apps vs. Unaided, and (B) Sound amplification apps vs. Hearing aids (basic).

Sound amplification apps vs. unaided

Two studies (Martinez-Beneyto et al. 2020; Seol et al. 2021) favoured sound amplification apps compared to unaided, with a further two studies either showing no difference when an in-built app was used with Galaxy Buds Pro (Han et al. 2022), or the opposite result (i.e. poorer performance) with Apple AirPods (Lin et al. 2022) (Figure 3(A)). The overall effect (n = 108) was not statistically significant (p = 0.432; ES = −0.27), and heterogeneity was high (I² = 83%).

Sound amplification apps vs. PSAP (premium)

Seol et al. (2021) (n = 18) compared a Galaxy Buds Pro (Med = −0.7 dB signal-to-noise ratio [SNR]; IQR = 1.7) with a premium PSAP (Med = −0.6 dB SNR; IQR = 2.1), showing that performance did not differ between devices.

Sound amplification apps vs. hearing aid (premium)

Only assessed by Lin et al. (2022) (n = 21), showing that speech intelligibility favoured a premium hearing aid compared to Apple AirPods (p < 0.05).

Sound amplification apps vs. hearing aid (basic)

Two studies (Lin et al. 2022; Seol et al. 2021) compared hearables plus app with basic hearing aids, with one (Seol et al. 2021) favouring the Galaxy Buds Pro, and the other (Lin et al. 2022) favouring the basic hearing aid relative to AirPods (Figure 3(B)). The overall effect (n = 39) was not statistically significant (p = 0.973; ES = 0.02), and heterogeneity was high (I² = 86%).

A narrative synthesis for listening ability and feasibility is presented in Supplementary Materials 6, as meta-analysis was not possible due to the limited number of studies.

Hearing aids

Speech intelligibility in noise

Hearing aid (basic) vs. hearing aid (premium)

Three comparisons showed that performance favoured premium hearing aids compared to basic hearing aids (Figure 4). The overall effect (n = 60) was statistically significant (p = 0.013; ES = 0.46), and heterogeneity was low (I² = 0%).

Figure 4. Summary of the random effects meta-analysis for speech intelligibility in noise for Hearing aids (basic) vs. Hearing aids (premium).

Risk of bias & quality assessment

Using the Cochrane risk of bias tool, all studies, except for Nieman et al. (2022) and Reed et al. (2017), were judged to be a high or unclear risk with regard to selection bias due to before-after study designs used (Supplementary Materials 7). For the five studies judged as unclear risk, while participants were randomised to different device conditions (between-subjects design) or the order with which devices were trialled (within-subjects design), the randomisation procedure was not provided in sufficient detail. The risk of performance and detection bias was judged to be high for 10 and 14 studies, respectively, as blinding procedures were not reported. The risk of attrition bias due to incomplete outcome data was judged to be low for 13 studies as there was no attrition. However, for Han et al. (2022) this was judged to be high as outcome measures were administered to only 14 out of the 40 participants with no reasons given for attrition. The risk of reporting bias was high for two studies, as deviations from the pre-registered protocol were not documented. Eleven studies were judged unclear as no study protocol was pre-registered and selective outcome reporting could not be determined. For risk of other bias, four studies were judged to be unclear if the study was funded by manufacturers of hearing devices or if there were other potential conflicts of interest which could have posed a threat to validity.

Scores on the Downs and Black (1998) checklist ranged from 13 to 24 out of a possible total score of 28, indicative of poor to good overall quality (Supplementary Materials 7). In terms of “reporting”, except for Sacco et al. (2016) and Nieman et al. (2022), adverse effects were not reported by any study. Whether participants were representative of the target population from which they were recruited (“external validity”) could not be determined for most studies, as insufficient detail was reported to make a clear judgement. In terms of “internal validity” (e.g. randomisation, blinding), except for two, no studies reported attempts to blind participants to the intervention they received. Furthermore, no studies reported attempts to blind those measuring the main outcomes of the intervention. Only two studies randomised participants to intervention groups. Lastly, only Han et al. (2022) and Nieman et al. (2022) reported undertaking a power calculation to determine sample size.

Discussion

The current systematic review updates the findings of two previous reviews (Chen et al. 2022; Maidment et al. 2018) assessing the effectiveness of non-medical amplification devices in adults with MMHL. In total, 14 studies were eligible for inclusion, with most evaluating PSAPs (n = 10), and four evaluating sound amplification apps with hearable or wired in-ear headphones (Amlani et al. 2019; Han et al. 2022; Lin et al. 2022; Martinez-Beneyto et al. 2020; Maidment and Ferguson 2018; Seol et al. 2021). Most studies compared speech intelligibility in noise between devices, whereas other outcomes (i.e. QoL, adverse effects, listening ability, cognition, feasibility) were assessed but to a lesser extent.

The current review provides novel and robust evidence that the features of all types of amplification device, including hearing aids, is important when making comparisons. Specifically, we showed that, for speech intelligibility in noise, premium PSAPs were better than unaided, whereas basic PSAPs and smartphone amplification apps used with a hearable or wired in-ear headphones were not. Premium hearing aids were better than both premium and basic PSAPs, as well as smartphone amplification apps and basic hearing aids. Although premium PSAPs were better than basic hearing aids, there was no difference between basic PSAPs and basic hearing aids or smartphone amplification apps. A similar pattern of findings was also found for QoL, adverse effects, listening ability, and feasibility. Heterogeneity in the outcome measures and/or devices did not permit meta-analyses for these outcomes. Based on these results, for speech intelligibility in noise, premium hearing aids appear to be the best management option for adults with MMHL. Nevertheless, premium PSAPs, but not basic PSAPs or smartphone amplification apps, seem to be a suitable alternative for individuals who cannot or do not wish to use premium hearing aids.

Non-medical amplification devices are likely to be more affordable than conventional hearing aids, therefore reducing the out-of-pocket cost to the individual, are accessible via online retailers, and require no consultation with a hearing healthcare professional. As such, and based on the current findings, premium PSAPs could serve as a “gateway” product to facilitate the earlier uptake of hearing aids. Subsequently, non-medical alternatives could act as a “gate” between doing nothing about hearing loss and getting a hearing aid (Olson, Maidment, and Ferguson 2022; Seol and Moon 2022). However, this systematic review further highlights that not all non-medical amplification devices are suitable gateway products. Namely, for speech-intelligibility in noise, performance for both basic PSAPs and sound amplification apps used with either a hearable (Galaxy Buds Pro, Apple AirPods) or wired in-ear headphones was equivalent to unaided. As a result, it appears that these devices are not suitable management options, and could even deter individuals from seeking hearing aids in the future, given they do not ameliorate the speech-in-noise difficulties experienced by adults with MMHL.

Although the findings of the current review appear to suggest that premium PSAPs improve speech intelligibility in noise relative to basic but not premium hearing aids, albeit with a small-to-moderate effect sizes, there are several considerations when drawing direct comparisons between devices. In the current study, both hearing aids and PSAPs were classified as premium based on the number of compression channels (≥16). As such, despite their cheaper price point, premium PSAPs often offered equivalent, or, in some cases, superior amplification technology compared to basic hearing aids (<16 channels). This differential between devices may explain why speech intelligibility in noise performance favoured premium PSAPs compared to basic hearing aids, as well as why performance between basic PSAPs and basic hearing aids did not differ statistically.

Additionally, all devices were fitted by a trained individual, such as an audiologist, community-based health worker, or researcher, to NAL-NL2 targets where possible. These “gold standard” fitting procedures are in contrast with the real-world service delivery model offered by non-medical amplification devices, where consumers are expected to buy, fit, and adjust a device themselves with no audiological input, thereby reducing the financial cost to the individual. When comparing audiologist-fitted PSAPs to real-world “out-of-the-box” (i.e. use manufacturer’s instructions to fit and adjust) and “advance-user” (i.e. fitting procedure based on advice from a professional) self-fitting procedures, Reed et al. (2019) found that the audiologist-fitted PSAPs provided the largest aided improvement in speech intelligibility performance. In another study, when the same self-fitting amplification devices were fitted by an audiologist, both new and experienced hearing aid users reported significantly higher device-related benefit and satisfaction compared to self-fitting by the user (Keidser and Convery 2018). The authors attributed these findings to the negative issues experienced by the user-driven fitting group in relation to the physical design and implementation of the self-fitting process. Thus, while device fitting undertaken by a trained hearing healthcare professional will likely result in the most optimal hearing outcomes, the provision of user-centred information to assist with the self-fitting and use of non-medical amplification devices will likely result in improved outcomes (Ferguson et al. 2019; Keidser and Convery 2016; Olson, Maidment, and Ferguson 2022).

A further methodological consideration is that most studies included in this review assessed outcomes under highly controlled, laboratory-based conditions, with no real-world listening for acclimatisation. In terms of speech intelligibility, for example, while all studies used validated measures, they varied in the type of background noise employed, where the background noise was presented relative to the target speech, as well as whether an adaptive or fixed SNR procedure was utilised. Such variation requires scrutiny, given that the degree of similarity between the target speech and background noise can have an impact on intelligibility performance (Rosen et al. 2013). Moreover, laboratory derived measures of speech intelligibility in noise do not necessarily reflect the complexities of real-world listening that can influence performance, such as the availability of audio-visual cues, binaural cues, and room reverberation (Brungart, Sheffield, and Kubli 2014; Phatak et al. 2018). As such, it has been suggested that laboratory-based measures may be insufficiently sensitive to detect change and that alternative measures collected in real-world listening situations, such as ecological momentary assessment, may be more appropriate when assessing the benefits of amplification devices (Timmer et al. 2018).

It should also be acknowledged that there are several limitations of the current review. For example, studies were only included if the sample consisted of participants with mild or moderate hearing losses in accordance with the grades specified by the World Health Organisation (2021), where the mean average hearing threshold fell within the specified range, or if only qualitative descriptions were provided with no audiometric data. Consequently, some studies may have been excluded where other audiometric descriptors for mild or moderate hearing losses have been used (e.g. British Society of Audiology 2011). Nevertheless, there remains a lack of consensus in terms of audiometric descriptors across different countries and organisations. For this review, we made the pragmatic decision to adopt the World Health Organisation (2021) grades of hearing loss so that our findings could be globally generalisable. Furthermore, we felt it appropriate to only include studies where the level of hearing loss could be verified audiometrically to allow for suitable comparisons between studies, reducing potential heterogeneity that could otherwise be introduced by including populations where hearing loss more severe and/or could not be substantiated, such as via self-report.

A further limitation of this review is that most included studies were before-after designs and subject to bias. Furthermore, the overall quality of the included studies ranged from poor to good in accordance with the Downs and Black (1998) checklist, with 10 out of 14 studies rated as poor or fair. Most studies were downgraded due to insufficient “reporting”, “external validity”, and “internal validity”. On this basis, it is apparent that further, high-quality evidence (i.e. randomised clinical controlled trials) is required in this area that addresses these limitations.

Taken together, our findings provide evidence to suggest that premium PSAPs, but not basic PSAPs or smartphone amplification apps, could be a suitable alternative to hearing aids, given that they improve hearing-related outcomes in adults with MMHL. We have shown that the key factors when comparing non-medical devices to conventional hearing aids are the features available in both device groups, as well as the fitting procedures employed. It may well be that the features and fitting, rather than a device simply being identified and named as a PSAP or hearing aid, need to be considered when drawing comparisons. Nevertheless, the current evidence-base is lacking in high-quality research that assesses non-medical amplification devices in the real-world, where self-fitting procedures are the expected service delivery model. As such, although non-medical amplification devices may provide an accessible and more affordable option to conventional hearing aids, the extent to which these devices are effective without audiological input remains to be determined.

Disclosure statement

No potential conflict of interest was reported by the author(s).