Increased maximum power output may improve speech recognition with bone conduction hearing devices

Abstract

Objective

To investigate the influence of maximum power output of bone conduction hearing devices on speech recognition in quiet and in noise in experienced users of bone conduction hearing devices.

Design

Prospective, randomised cross-over investigation comparing speech recognition performance, subjective sound quality, and device preference between two bone conduction hearing devices with different maximum power outputs.

Study sample

Sixteen adult subjects with conductive or mixed hearing loss.

Results

Both speech recognition in quiet and speech recognition in noise improved significantly when using the device with high vs. lower maximum power output. Mean improvement in word recognition score in quiet was 10.5% and the mean speech reception threshold in noise improved by 0.9 dB SNR. Compared to the device with lower maximum power output, the sound quality was rated significantly higher with the device with high maximum power output, which was also the device of preference for 81% of the subjects.

Conclusion

Bone conduction hearing devices with higher maximum power output have the potential to improve speech recognition in both quiet and noisy listening environments.

Keywords:

• Maximum power output
• conductive hearing loss
• mixed hearing loss
• bone conduction
• hearing device
• speech recognition

Introduction

Bone Conduction Hearing Devices (BCHD) is hearing solutions that use the natural process of sound travelling through the bone and soft tissue directly to the cochlea. BCHD consist of an external sound processor that captures and processes sound which is then converted into mechanical vibrations. These vibrations are transmitted to an implant that is anchored to the skull bone, through which they are further conducted to the cochlea. The transmission from the sound processor to the implant happens either percutaneously via an abutment or transcutaneously via a magnetic coupling. With transcutaneous BCHD, the transducer can be an external unit or implanted under the skin and soft tissue, which is referred to as passive vs. active BCHD, respectively. For an overview of the different types of available BCHD, we refer to the recent publication by Håkansson et al. (2019).

BCHD is an effective hearing solution for people with a conductive hearing loss (CHL) or mixed hearing loss (MHL), as they bypass any limitations on sound transmission associated with pathologies or anomalies of the outer or middle ear (den Besten et al. 2019; Dimitriadis et al. 2016; Iseri et al. 2015). These devices can also help patients who suffer from single-sided deafness by transmitting sound received on the deaf side directly to the hearing cochlea (den Besten et al. 2019; Bonne et al. 2019; Flynn et al. 2010). BCHD bypasses the conductive part of the hearing loss and, as such, inherently closes the air-bone gap (ABG). Therefore, less amplification is needed when compared to conventional hearing aids (HA) in patients with a conductive element to their hearing loss. In fact, it has been shown that BCHD provides better hearing outcomes than HA for patients who show an ABG exceeding 30–35 dB HL (de Wolf et al. 2011; Mylanus et al. 1998). When selecting a BCHD for a patient, the gain that the device can deliver is an important factor, as the gain must be sufficient to amplify sounds above the patient’s hearing threshold. Another crucial parameter in the selection process is the maximum power output (MPO), and this is the focus of the current study. The MPO is defined as the highest output level a hearing device can deliver and is an important consideration for prescriptive targets. For instance, the MPO can limit the available gain to the patient. Amplified speech signals should not hit the MPO as this would impact the sound quality and, in turn, speech recognition. For some patients, the MPO might have to be limited to prevent the aided output to be uncomfortably loud. On the other hand, the MPO of a hearing device should be high enough to provide a sufficient dynamic range. Speech is a highly dynamic acoustic signal including slow fluctuations in its overall intensity which are known to be crucial cues for speech recognition (Rosen 1992; Greenberg et al. 2003). To be able to perceive these intensity fluctuations and, in turn, achieve good speech recognition, the MPO of a hearing device should be high enough so that the dynamic range of speech fits within the dynamic range of the impaired ear (Amlani, Punch, and Ching 2002; Zwartenkot et al. 2014). Research has demonstrated that a dynamic range of at least 35 dB is needed to achieve good speech recognition (Zwartenkot et al. 2014; van Barneveld et al. 2018; Rahne and Plontke 2016). A sufficient MPO is even more important for speech recognition in noise than it is for speech in quiet. In noisy listening environments, the overall output level is higher, requiring higher MPO from the hearing device to preserve the dynamics of the speech signal and minimise the risk for distortion. When the MPO is limited, the intensity fluctuations that characterise speech may become less distinct, making it harder to distinguish the target speech from the background noise.

The maximum output of a BCHD is commonly measured in dB OFL rel. 1uN, denoting the maximum force level that a BCHD can deliver to the cochlea (IEC-60118-9:2019, §7.2). Although maximum force output (MFO) is an accurate term for BCHDs, we will apply the term MPO throughout this manuscript as this is considered a more generic term, allowing for comparisons between different types of hearing devices. The MPO of BHCD can be determined objectively with a skull simulator or through real-ear measurements (Zwartenkot et al. 2014; Mertens et al. 2014; van Barneveld et al. 2018). Research has shown that the MPO of percutaneous BCHD is about 10–15 dB higher, particularly at the high frequencies, when compared to passive and active transcutaneous BCHD (van Barneveld et al. 2018; Van Eeckhoutte et al. 2020; Hol et al. 2013). This inherent difference is merely explained by the higher efficiency of a percutaneous vs. a transcutaneous coupling.

Although the importance of MPO is evident, MPO is an often overlooked feature of hearing devices. To the best of our knowledge, only a few studies have investigated the impact of MPO on speech recognition performance with BCHD. Gawliczek et al. (2020) recruited BCHD users who were diagnosed with a conductive or mixed hearing loss. Speech recognition was measured in quiet and in noise with four different MPO level settings, starting from the highest possible level and decreasing in 6-dB steps. The results showed that patients whose SNHL exceeding 35 dB HL had better speech recognition in noise at higher MPO levels (Gawliczek et al. 2020). Increasing the MPO from the lowest to the highest MPO level resulted in a significant average improvement in speech reception threshold (SRT) of 3.2 dB SNR. No such effect of MPO was found for speech recognition in quiet or if the SNHL was lower (better) than 35 dB HL.

Kurz et al. (2014) recruited experienced BCHD users with a mixed hearing loss. Speech recognition in quiet and in noise was evaluated with maximal MPO and with the MPO reduced by 5 dB. The patients were asked to indicate their preference, following experience with both MPO settings in their home environment. In contrast to the authors’ expectations, no significant difference in speech recognition performance was found between the high and lower MPO. Interestingly, a significant subjective patient preference for the high MPO was found, which suggests that the patients did perceive the difference in MPO. The researchers speculated that the absence of a significant MPO effect on speech recognition may be explained by the rather mild SNHL of their participants, i.e. average BC thresholds of about 30 dB HL.

Bosman et al. (2018) tested speech recognition in noise in patients with a mixed hearing loss aided with either a high or lower MPO BCHD. For frequencies below 1000 Hz, the MPO of both BCHD was similar, but at higher frequencies, the MPO differed by 10 dB. When speech and noise were presented from the front, the SRT improved from −1.3 dB SNR with the lower MPO device to −3.8 dB SNR with the high MPO device. This significant improvement in speech recognition in noise was reflected in the subjective hearing outcomes. The Abbreviated Profile of Hearing Aid Benefit (APHAB) questionnaire revealed significantly better self-perceived hearing abilities for ease of communication and background noise with the high vs. lower MPO device, and the speech and quality subscales of the Speech, Spatial, and Qualities (SSQ) questionnaire showed higher (better) scores for the high compared to the lower MPO device. Furthermore, 90% of the participants indicated that they preferred the high over the lower MPO device.

Hodgetts et al. (2011) investigated the efficacy of a traditional and more advanced fitting approach for BCHD. One of the major differences was the higher MPO, particularly at high frequencies (>2000 Hz), with the advanced vs. the traditional fitting. Speech recognition performance was compared between both fitting approaches in BCHD users with a mixed hearing loss. The results showed significantly improved speech recognition in quiet and in noise for the advanced vs. the traditional fitting. The authors discussed that most of this benefit must be a consequence of the increased high-frequency MPO. Anecdotally, participants reported that sounds were much clearer with the advanced fitting (higher MPO).

Hol et al. (2013) tested speech recognition in quiet in 12 children with a permanent conductive hearing loss. Six of them were implanted with a transcutaneous BCHD and six children received a percutaneous BCHD, with both groups being matched for hearing loss. For both devices, similar settings were applied. The main difference was the MPO, which was 10 dB higher for the percutaneous BCHD. The results indicated better speech reception thresholds and word recognition scores with the percutaneous compared to the transcutaneous BCHD. The authors attributed this benefit to the potential for higher gain and MPO with the percutaneous device.

Altogether, studies on the impact of MPO on the speech recognition performance with BCHD are scarce, yet they typically suggest that increasing the MPO could improve speech recognition. The present study aimed to investigate and expand on this idea. The main goal was to compare the speech recognition performance between a BCHD with high vs. lower MPO in a group of experienced users. In addition, we compared the subjective sound quality between the high and lower MPO devices and evaluated patient preference based on testing in their home/normal hearing environment. We hypothesised that patients would show better speech recognition when using the BCHD with high vs. lower MPO and that this improvement would be most pronounced for speech recognition in noise, given the increased importance of MPO in noisy listening environments. Also, we expected that the objective hearing benefit would be reflected in the patients’ subjective sound quality and device preference, i.e., in favour of the BCHD with the high vs. lower MPO.

Materials and methods

Study population

Sixteen Swedish-speaking participants were included in the study, comprising of 11 women and 5 men between the age of 22–69 years old (mean age: 56 years). All subjects had at least three years of experience from using a percutaneous BCHD (mean experience: 22 years, range 3–26 years). The aetiology of the hearing loss was chronic otitis media in five subjects, malformations in six subjects, otosclerosis in two subjects, cholesteatoma in two subjects, and hearing loss caused by trauma in one subject. Nine participants had a pure CHL and seven subjects had a mild to moderate MHL, and five out of 16 participants used bilateral BCHD. All participants were frequent users of BCHD and had Swedish as their first language. The hearing thresholds were measured using standard audiological procedures. Table 1 provides the mean and standard deviation (SD) of the bone conduction (BC) hearing thresholds of the test ear for the study sample. Table 2 shows the individual frequencies and mean hearing thresholds for the lower MPO device, high MPO device, and the unaided condition.

Table 1. Mean BC hearing thresholds for the test ear, with standard deviations (SD), measured for the study sample between 250 and 6000 Hz.

	Frequency (Hz)
	250	500	750	1000	1500	2000	3000	4000	6000
Mean BC hearing threshold (dB HL)	10.0	15.0	17.3	20.6	26.3	36.6	32.8	29.1	21.5
SD	11.3	12.6	12.0	16.8	18.4	19.6	17.6	20.4	21.0

Table 2. Mean hearing thresholds in the sound field for subjects when using the high MPO device, the lower MPO device, or the unaided condition.

Hearing thresholds in the sound field
Frequency (Hz)	High MPO device	Lower MPO device	Unaided condition
250	29.7 (SD 7.2)	34.1 (SD 8.2)	46.25 (SD 13.6)
500	20.9 (SD 9.4)	24.1 (SD 10.8)	49.69 (SD 12.0)
1000	20.3 (SD 8.8)	23.1 (SD 10.5)	54.38 (SD 12.6)
2000	23.4 (SD 11.4)	25.3 (SD 12.7)	50.63 (SD 16.1)
3000	29.1 (SD 15.4)	33.4 (SD 15.7)	55.31 (SD 16.4)
4000	35.0 (SD 14.1)	37.5 (SD 16.2)	64.06 (SD 17.8)
6000	36.9 (SD 15.2)	39.7 (SD 17.4)	65.00 (SD 18.4)
8000	35.3 (SD 20.0)	56.9 (SD 22.0)	65.63 (SD 17.3)

Sound processors and fitting of hearing devices

Tests were performed with Baha^® 5 (Cochlear Ltd, Sydney, Australia) and Baha 6 Max (Cochlear Ltd, Sydney, Australia) sound processors on abutments, which have fitting ranges of up to 45 dB HL and up to 55 dB HL, respectively. The difference between the sound processors is that a new actuator and new signal processing have been implemented in Baha 6 Max, resulting in greater MPO and gain at higher frequencies. Figure 1 shows the MPO levels and aided speech spectra measured for 65 dB SPL speech-shaped noise based on the mean hearing level of the study sample (see Table 1) for both sound processors. The average MPO (500, 1000, 2000, and 4000 Hz) available to the patient was 108 dB for the Baha 5 sound processor and 114 dB for the Baha 6 Max sound processor, i.e. a 6 dB difference. Both sound processors were individually fitted according to the manufacturer’s recommendations, using Cochlear Baha Fitting Software (Version 6.0). For fitting the devices, in-situ measurements of the BC thresholds (BC-direct thresholds) were obtained. This method makes use of the hearing devices acting as an audiometer and the resulting configuration of the BC-direct thresholds is used to calculate the appropriate gain and compression characteristics. A feedback test was also performed to make sure the devices would not whistle and to estimate maximum available gain. In each fitting session, the participants acclimatised about 10 min through normal use to the new fitting and were then asked whether they wanted to have a gain adjustment. To emulate a real-life clinical situation, an everyday setting program was utilised in both devices for all measurements, ensuring that all features were the same between both devices and within-subjects throughout the trial. In the everyday setting program, directional microphones and gain were set to automatic and adaptive, respectively, and position compensation was switched off due to the automatic setting. Furthermore, mild noise reduction and a moderate level of feedback management were activated in both devices.

Figure 1. MPO (black lines) and aided speech spectra (grey lines) for the lower MPO device (solid lines) and the high MPO device (dashed lines) measured on skull simulator TU-1000 across the frequency range of 250–8000 Hz.

Measurements

All tests were performed in a sound-insulated audiometric booth using calibrated equipment with the non-test ear double blocked using an earmuff and earplug in case of asymmetry between ears and when the licenced audiologist deemed it was needed to evaluate the devices correctly. To emulate clinical standards, both speech and noise were presented 1 m in front of the participants at 0° azimuth.

Two speech recognition tests at 65 dB SPL were used to compare speech recognition performance between devices. The Swedish phonemically balanced word lists were used as speech material to evaluate speech recognition in quiet (Liden and Fant 1954). Each list is composed of 50 monosyllabic words. The participants were instructed to repeat the word that followed a carrier phrase, and they were encouraged to guess. Performance was recorded as the percentage of correctly repeated words. The Swedish version of the Matrix test was employed to determine the speech recognition threshold in noise (Kollmeier et al. 2015). All sentence stimuli were presented in the presence of speech-weighted noise, i.e. with a long-term average spectrum identical to that of the corpus. The noise, starting at 65 dB SPL, was presented continuously between sentences and was adapted stepwise using predefined dB steps to establish the final SNR at which 50% of the words are correctly repeated.

Preference of device was obtained at the last visit using a forced-choice method. The participants were instructed to choose either of the devices they had been trying out at home based on personal preference. Thus, options, such as “neutral” or “no opinion” were not included.

For subjective measurements, two sound clips were evaluated between devices at the last session in a single-blinded randomised manner using a 5-point Likert scale (Likert 1932). The Likert scale was used for the assessment of sound quality evaluation as it is a common and relatively sensitive measure, which was also confirmed by a previous pilot study conducted on both sound processors. The first sound clip presented a female voice reading from a book at 65 dB SPL with four-talker babble as background noise at 60 dB SPL (+5 dB SNR) for 1 min. The second sound clip presented a 1-min music clip from the song “Don’t stop me now” by Queen at 70 dB SPL. The participants were instructed to listen to both sounds clips and evaluate the general perception of sound quality. After listening to each sound clip, with each device, the participants were instructed to rate the sound quality on a scale ranging from 1 to 5 where 1 indicates “very bad” sound quality, and 5 “very good” sound quality.

Procedure

The study employed a randomised cross-over design. To complete data collection, each participant had to spend a minimum of three separate sessions, each lasting 1–2.5 h, spaced over three weeks. At the first appointment (week 1), the unaided audiogram was obtained. After the test, a two-week cross-over trial was conducted in which the 16 participants switched devices after 1 week of use. The study sample was randomly divided into two groups. The first group was fitted with the device with lower MPO first and the second group fitted with the device with high MPO first. After the fitting, speech recognition tests were carried out in quiet and in noise, and each group went home with the device they were assigned.

At the second session (week 2), the first group was fitted with the device with high MPO, and the second group fitted with the lower MPO device. Speech recognition tests in quiet and in noise were carried out again and both groups went home with the devices for another week.

At the third session (week 3), the participants returned and were instructed to choose between the lower and high MPO devices in terms of subjective preference. After stating their personal choice of device, a subjective sound quality evaluation of the two sound clips was collected.

Statistical analysis

All statistical analyses were paired, and non-parametric analyses were employed in IBM^® SPSS^® Statistics Software Package (IBM Corporation, Armonk, NY, USA). For continuous variables and ordered categorical variables, the Wilcoxon signed-rank test was applied and for binary variables, the binomial test was applied. Adjustment for covariates was not required based on study design as each subject acted as their own comparator. As the a priori predictions were directional, one‐tailed tests are reported, with an alpha level of 0.05. Where applicable, effect sizes (r) are calculated by dividing the Z statistic by the square root of the sample size.

Ethical consideration

The investigation was approved by the Swedish Ethical Review Authority (Dnr 2020-01496) and performed by the ethical principles that have their origin in the Declaration of Helsinki and Good Clinical Practice ISO 14155:2020.

Results

Speech recognition in quiet

Speech recognition in quiet scores were 77.9% (SD 25.1, range 22–98) for the lower MPO device and 88.4% (SD 14.4, range 56–100) for the high MPO device (Figure 2). The mean difference in speech recognition between devices was 10.5% and the Wilcoxon signed-rank test revealed a significant difference (Z = −2.83, p = 0.002, r = 0.71). Thus, the subjects had a better speech recognition score in quiet when using the high MPO device.

Figure 2. Mean word recognition scores were measured in quiet using the lower MPO device (light grey) and the high MPO device (dark grey). Error bars represent the standard error of the mean.

Speech recognition in noise

For speech recognition in noise, the mean performance with the high MPO device was −3.7 dB SNR (SD 2.8, range −7.2 to 2.0) and with the lower MPO device, it was −2.8 dB SNR (SD 2.4, range −5.8 to 1.8) (Figure 3). The mean difference between devices was 0.9 dB SNR, which was statistically significant (Z = −1.684, p = 0.048, r = 0.41) as determined by the Wilcoxon signed-rank test.

Figure 3. Mean speech reception thresholds were measured in noise using the lower MPO device (light grey) and the high MPO device (dark grey). Error bars represent the standard error of the mean.

Subjective preference

When instructed to choose between the lower and high MPO device at the last visit, 13 subjects (81%) preferred the high MPO device and three subjects (19%) preferred the device with the lower MPO. The binomial test revealed that preference for the high MPO device was statistically significant (p < 0.001). Subjective factors reported as influencing preference for the high MPO device included better sound quality in terms of richness and clarity, better speech recognition, and more high-frequency sounds perceived. Eleven of 13 subjects (85%) stated that they preferred the high MPO device due to comfortable sound perception and nine subjects (69%) preferred the high MPO device due to better speech understanding.

Subjective evaluations of sound quality

Statistically significant differences between devices were observed for general sound quality in favour of the high MPO device when tested using two sound clips. For female voice in four-talker babble, the high MPO device demonstrated a mean score of 4.25 (SD 0.86, range 2–5), indicating good to very good sound quality. For this sound clip, the lower MPO device demonstrated a mean score of 3.31 (SD 1.20, range 1–5), indicating acceptable to good sound quality, which was significantly lower (worse) (Z = −2.60, p = 0.005, r = 0.65). The same result pattern was observed for the music clip, where the high MPO device scored an average of 4.31 (SD 0.70, range 3–5) points, indicating good to very good sound quality, and the lower MPO device averaged 3.44 (SD 1.09, range 2–5) points, indicating acceptable to good sound quality. The Wilcoxon signed-rank test revealed that this difference was also statistically significant (Z = −2.10, p = 0.018, r = 0.52) in favour of the high MPO device.

Discussion

The current study aimed to compare speech recognition performance and subjective sound quality between a BCHD with a high vs. lower MPO in a group of experienced BCHD users. Also, the participants’ preference for either the high or lower MPO device was evaluated. This study demonstrates that the high MPO device leads to better speech recognition, both in quiet and in noise. Our participants indicated that the sound quality was better with the high MPO device and that they preferred the high MPO over the lower MPO device.

For speech recognition in quiet, the present findings demonstrate a statistically significant improvement with the high MPO device when compared to the lower MPO device (i.e., word recognition score of 88 vs. 78%). This outcome confirms the beneficial effect of a higher MPO on speech recognition in quiet as reported by Hodgetts et al. (2011) and Hol et al. (2013). Yet, Gawliczek et al. (2020) and Kurz et al. (2014) did not find speech recognition in quiet to improve with increasing MPO levels. Gawliczek et al. (2020) measured speech recognition in quiet with four different MPO level settings, starting from the highest possible level of the BCHD and decreasing in 6-dB steps. The absence of a significant MPO effect in the study of Gawliczek et al. (2020) may be attributed to a ceiling effect since most of their participants had a word recognition score exceeding 90% in each of the four MPO conditions. In the present study, aiding with the lower MPO device did not result in a ceiling effect, which left room for further speech recognition improvement with the high MPO device. In the study of Kurz et al. (2014), a 5 dB-reduction in MPO did not affect word recognition in quiet, in contrast to the authors’ expectation. Given the high degree of similarity in methods with our study (i.e., similar MPO difference, number of participants, the average degree of hearing loss, and speech test), it remains unclear why no effect of MPO was found by Kurz and colleagues.

Unexpectedly, Kurz et al. did not find a significant effect of MPO on speech recognition in noise, as we did find a significant improvement in SRT when increasing the MPO by 6 dB in our study. Gawliczek et al. (2020), Bosman et al. (2018), and Hodgetts et al. (2011) also concluded that increasing the MPO resulted in significantly improved speech recognition in noise. Gawliczek et al. (2020) recruited 12 BCHD users with a CHL or MHL (BC thresholds: 4–45 dB HL) and investigated speech recognition in noise with a matrix test at four MPO level settings, i.e. the highest possible MPO level vs. 6, 12, and 18 dB below. Similar to the current study, Gawliczek et al. (2020) found that higher MPO levels resulted in better speech recognition in noise. It should be noted, however, that the MPO effect reported by these authors was somewhat restricted when compared to the present study. Gawliczek et al. (2020) found a significant improvement in speech recognition of 3.2 dB SNR, only when the MPO levels differed by 18 dB. When the MPO differed by 12 or 6 dB, SNR improvements ranging between 0.3 and 1.8 dB were observed but they were not statistically significant. In the present study, the high MPO device differed by 6 dB from the lower MPO device and this resulted in a statistically significant, average improvement in the SRT of 0.9 dB SNR. It could be argued that the mixed-effect linear model analysis applied by Gawliczek et al. (2020) could have obscured the significance of the observed SNR improvements. Such models generally require large sample sizes to be powerful, yet only 12 participants were included. Bosman et al. (2018) reported an improvement in SRT of 2.5 dB SNR when their participants used a BCHD with a 10-dB higher MPO. When compared to the SRT improvement of 0.9 dB SNR in the present study, this is a relatively large improvement that is not unexpected. Firstly, because the MPO difference under investigation in their study is larger, i.e. 10 vs. 6 dB. Secondly, Bosman et al. (2018) included a group with moderate SNHL (mean pure-tone average: 42.8 dB HL), whereas our participants only showed a mild to moderate SNHL (mean pure-tone average: 26.7 dB HL). A more beneficial impact of increased MPO with increasing hearing loss has been suggested by Gawliczek et al. (2020), where the authors discuss how aiding hearing losses >35 dB HL with higher MPO devices would benefit speech recognition in noise, with diminished effects in individuals with hearing thresholds better than 35 dB HL. This rationale could also explain why Hodgetts et al. (2011) found a significant improvement in SRT as high as 2.68 dB SNR when they increased the high-frequency MPO of the BCHD. In fact, the average SNHL in their participant group equalled 39.1 dB HL (>35 dB HL).

Regarding patients’ preference, 81% of our participants (n = 13) preferred the high over the lower MPO device when instructed to choose between devices at the end of the cross-over trial. Kurz et al. (2014) also demonstrated a significant subjective patient preference for the high vs. lower MPO offered in their blinded home-based experiment. Likewise, Bosman et al. (2018) reported that nine of the 10 participants preferred the high over the lower MPO device after having tested the devices in their home environment. The significant patient preference was reflected in the results of the APHAB and SSQ questionnaires. In both questionnaires, better scores were allocated to the high vs. lower MPO device. More specifically, the high MPO device was associated with better sound quality and better subjective speech perception abilities. Similar patient reports were retrieved from the current study. Most of our participants stated that the reason for preferring the high MPO device was because of the better sound quality and improved ability to understand speech compared to the lower MPO device, which is, indeed, supported by the speech recognition results. Seven out of 13 also reported that the sound with the high MPO device was more pleasant and natural, and described it as “not as compressed” as with the lower MPO device. When asked about how this hearing experience translated into their daily life, many of the participants pointed out that the positive impact in real life was most pronounced for outdoor environments. For example, when walking in the woods in the presence of considerable wind noise, they perceived the speech of their companion more clearly when using the high MPO device. Another instance where the high MPO device appeared to facilitate conversations was when going downtown to shop for groceries with a lot of traffic noise in the background. One of the participants said he could hear the speech of the vendor much more clearly at the local market. Lastly, one participant also described the canteen situation at work as easier to perceive speech and stated that the sounds were more natural compared to the lower MPO device. While all participants described that they still perceived interfering background noise in their real-life environments, most of them reported that it was not as disturbing when compared to the lower MPO device. These findings regarding real-life, noisy listening environments are interesting as our participants were not able to switch programs (e.g. outdoor programs). Moreover, with both devices, a similar gain was prescribed (i.e. based on the patient’s in-situ threshold measurements) and all automatic features were on. Hence, it seems likely that the main difference to which the positive real-life experiences can be attributed, is the high vs. lower MPO level. More research is, however, desirable to confirm this assumption.

Three participants (19%) preferred the lower MPO device at the end of the cross-over trial. These participants had in common that they were fond of their own devices, which were relatively old BCHD (i.e. previous generations). The main reason for not upgrading to a more modern BCHD was that they were very pleased with their own device in terms of sound quality. For them, using a BCHD with higher MPO did not equate to better sound quality. A longer acclimatisation (>1 week) might have been required for these participants to experience the benefits of the high MPO device. Also, one subject had Treacher Collins syndrome and was very cautious about changes in their everyday life in general. These cases emphasise that, even though speech recognition data demonstrate the benefits of a high MPO device, it is important to consider aspects like subjective acclimatisation and psychosocial factors when selecting a new hearing device in clinical practice. Note that audiologists tend to anticipate that hearing loss worsens over time when selecting a hearing device and, therefore, often select a device with a higher MPO, taking a certain amount of dB in reserve. Together with previous research (Hol et al. 2013; Gawliczek et al. 2020; Bosman et al. 2018; Hodgetts et al. 2011; Kuk et al. 2011), this study demonstrates that speech recognition performance may be another rationale for selecting a high MPO device, in addition to accounting for worse hearing. As mentioned before, the MPO is an important consideration for prescriptive targets as, for instance, it should approximate yet not exceed the loudness discomfort levels of the patient. We propose a holistic, case-by-case assessment when selecting devices with high MPO, thereby considering the patients’ hearing profile, acclimatisation, psychosocial factors, as well as speech recognition performance. Also, cosmetics and high-frequency MPO can influence the choice of device. Regarding high-frequency MPO, research shows that extended device frequency bandwidth may have a positive impact on speech recognition (Van Eeckhoutte et al. 2020; Levy et al. 2015). Notably, as illustrated in Figure 1, the high MPO device produced more audibility for speech-level output, delivering a broader aided bandwidth for speech input at 2000 Hz and above. Therefore, inherent speech peaks may have been saturated with the lower MPO device (e.g. at 500 Hz), while there would have been sufficient headroom with the high MPO device. It is not unlikely that the improvement in speech recognition with the high MPO device can partly be attributed to these factors.

A few limitations of the present study should be addressed. Firstly, the results should only be generalised to patients with mild to moderate hearing loss using BCHD. Based on our results, no conclusions can be made regarding the patterns that apply to patients with severe to profound hearing loss and/or hearing solutions other than BCHD (i.e. conventional hearing aids, middle ear implants, cochlear implants). Secondly, different tests were used to evaluate speech recognition in quiet, on the one hand, and speech recognition in noise, on the other hand, which made it difficult to compare the impact of MPO between quiet and noisy listening situations. However, the purpose of the present study was to emulate clinical/real-life situations (i.e. standard clinical tests and everyday settings for the BCHDs with all automatic features activated). Since the monosyllabic word test and the Matrix test are the standard tests in clinical practice in Sweden, it was decided to use them in the present study. Thirdly, there was a one-week trial period with each device. Even though all our participants were experienced BCHD users, three participants did not prefer the high MPO device at the end of the trial. One could speculate that a longer acclimatisation period with each device might have yielded different outcomes. Lastly, for subjective evaluations, the preference of the device was not blinded in the home trial due to visible/physical differences in the sound processors (i.e. 2 mm shorter snap coupling for the high MPO device). Blinding was, however, conducted for the subjective measurements of sound quality.

Summary

In summary, this study shows that increased MPO can improve speech recognition with BCHD, both in quiet and in noise and that experienced BCHD users generally prefer higher MPO devices because of better sound quality and improved ability to understand speech. For clinical practice, we suggest a case-by-case assessment for selecting BCHD with high MPO, thereby considering acclimatisation and psychosocial factors, in addition to hearing profile, and speech recognition performance. While the current study provides evidence in agreement with the limited number of previous studies, more research is needed to unravel the impact of MPO on speech recognition for the wider BCHD population.

Acknowledgements

The authors would like to thank Theres Björk, Therese Agat, Jenny Andersson, and Jona Hoffman who carried out the tests and technical measurements. Thanks are also due to Clas Johansson and Christiane D’hondt for providing valuable comments on the manuscript.

Disclosure statement

Håkan Hua, Tine Goossens, and Aaran Lewis are current employees of Cochlear Bone Anchored Solutions AB, a manufacturer, and supplier of bone conduction hearing devices.

Additional information

Funding

This work was supported by Cochlear Bone Anchored Solutions AB.