Chapter 3. Basic considerations; the effect of low MPO on amplification and word recognition

What is the effect of a relatively low MPO? Figure 3.1A (top, right) shows the audiogram of a patient with conductive hearing loss. Cochlear (bone-conduction) thresholds are shown. The loudness-discomfort-levels (LDLs; Dillon and Storey, 1998), are also indicated in Figure 3.1B (top, left).

figure 3.1.1    figure 3.1.2

figure 3.1.3

Figure 3.1. Figure A (top left) shows the audiogram; figure B (top, right) present the cochlear thresholds (bone-conduction thresholds) and the LDLs. Figure C (bottom figure) shows the MPO values (labelled with ‘S’) of the applied (passive transcutaneous) ‘BCD’.

By definition, the patient’s dynamic range of hearing is the difference between the cochlear threshold and LDLs. Apparently, Figure 3.1C shows that the MPO values, indicated by symbol S, split the patient’s dynamic range of hearing into two parts; an audible part (45 dB wide, see vertical red arrows) and a second unused part (40 dB wide). The speech area of the patient’s own voice is schematically indicated, with the yellow part being ‘inaudible’.

Intermezzo. The gain of a device is the difference between aided and unaided thresholds. Using the bone-conduction route, the gain, called the effective gain or EG, is the difference between the (BCD) aided thresholds and the bone-conduction thresholds. If the EG=0 dB, this means that with the BCD the air-bone gap is ‘closed’. A negative EG indicates that despite the amplification, a residual air-bone gap exists. See also Appendix 2. End intermezzo.

If the patient is fitted effectively with the BCD (EG=0 dB), then any signal louder that the MPO of the BCD (symbols ‘S’ in Figure 3.1C) cannot be processed properly causing distortions. The patient can prevent distortions by turning down the volume (thus reducing the EG). For instance, in this case, the chosen reduction in EG might be 20 dB. In this way, the louder part of the patient’s own voice (yellow area), which ranges from 50 to 70 dB HL, is perceived undistorted, now ranging from 30 to 50 dB HL. Consequently, the patient’s own speech fits nicely into the BCD-aided dynamic hearing range. However, that 20 dB attenuation EG of – 20 dB negatively affects speech recognition of someone else, who talks at a normal conversational level. The ‘negative EG’ of 20 dB appears in the audiogram as a ‘remaining AB-gap’ of 20 dB. This example highlights that it is better to use a BCD with a high MPO, even in patients with a (sub)normal sensorineural hearing loss component.

The next question is: what is the volume setting (chosen amplification) in daily practise by BCD users with different degrees of cochlear hearing loss? This question was answered using our Baha database by reviewing the data of 89 patients who used Baha devices (Snik et al., 2004). At that time, Baha devices with (mainly) linear amplification were used (Baha Compact, Classic, HC220, Cordelle) and with a volume wheel that allowed the patients to choose their preferred amplification. (Remark. In the next figures the cochlear hearing loss or the sensorineural hearing loss component is referred to as the SNHLc. The SNHLc stands for the mean bone-conduction threshold at 0.5, 1, 2 and 4 kHz averaged, often referred to as the PTA4bc.)

3.2_nieuw

Figure 3.2 The ‘Baha’-aided thresholds (at 0.5, 1 and 2 kHz averaged) as a function of the SNHLc (or PTA4bc). For clarity, individual data of the 89 Baha users were grouped into 12 classes of each 5 dB wide. The diagonal line represents the bone-conduction thresholds

Figure 3.2 shows that patients with (sub)normal cochlear function (low SNHLc) set the volume of their Baha such that the EG is negative: the ‘Baha’-aided thresholds are worse (higher) than the bone conduction thresholds (shown in the figure as data points above the diagonal). This results in a remaining AB-gap of around 10-15 dB, which is seen for SNHLc values up to approx. 25 dB HL.

Patients with more severe SNHLc need ‘positive’ amplification to be able to properly hear other subjects talking, which, according to Figure 3.2, is found for SNHLc > 30 dB HL.

It should be noted that with increasing SNHLc, patients used the more powerful Baha sound processors. Per patient, the type of Baha device used was documented and, consequently, its MPO was known. These MPO values were used to obtain the ‘input level at saturation’ or the sound level at which the MPO was reached. Louder levels lead to distortions. This ‘input level’ is simply the MPO value minus the amplification provided by the Baha (output = input + amplification). The amplification itself is easily obtained by subtracting the Baha-aided thresholds from the bone-conduction thresholds. These calculated (individual) input levels at which the Baha devices saturated, were (again) grouped in the 12 classes, presented in Figure 3.2 as the blue line. The figure clearly shows that the input level at saturation restricts the dynamic range of hearing with the Baha significantly. Before discussing the consequences, Figure 3.2 was redrawn, see Figure 3.2bis.

 3.2_bis  3.2_bis_2

Figure 3.2bis Redrawn of Figure 3.2 with more details, see text

The left graph of Figure 3.2bis shows the residual capacity of ear of patients with SNHLc between 0 and 60 dB HL. As mentioned earlier, the diagonal line presents the cochlear thresholds while the dotted line presents the patient’s LDL. The graph on the right shows the aided situation after Baha fitting, taken from Figure 3.2. The difference between aided thresholds and input level at saturation is the patient’s aided dynamic range of hearing or their ‘auditory window’ to the world.

The Baha-aided dynamic range of hearing is limited; for example, with a SNHLc of 45 dB HL, the patient’s unaided dynamic range of hearing is 50 dB HL (95-45 dB HL, left graph, red arrow) and with the Baha, the dynamic range is only 35 dB HL (40 to 75 dB HL, right graph, red arrow).

The subfigure on the right side schematically shows sounds from the real auditory world. Let us consider normal conversational speech. That level fluctuates between 25 dB HL and 55 dB HL (indicated in the subfigure, right side; according to Killion & Christensen, 1998 based on speech with an overall level of 65 dB SPL). By combining the right subfigure and the graph on the right shows that in the current patient group, the 30 dB speech range is only fully audible when the SNHLc is less than 15 dB HL, while at a SNHLc of 60 dB HL, the audible proportion of the speech range has decreased to just 5 dB HL.

In addition, we learn from Figure 3.2 that the patients, regardless of their SNHLc, adjust the volume of their device so that the device saturates at input sound levels of 70-75 dB HL. Indeed, this choice allows for undistorted perception of even the loudest parts of normal conversational speech. Thus, the figure suggests that the patients did not compromise on the input level at saturation, but did make compromises with respect to the softest speech sounds they could hear.

The basic rule for the desired gain in pure sensorineural hearing loss equals approximately half the sensory neural hearing loss (e.g., Dillon, 2012). As the figure shows, that has not been achieved at all. For example, at a SNHLc of 45 dB HL, the EG is only 7 dB (see Figure 3.2). It is concluded that the limited MPO plays an important role in the low gain setting chosen by the patients.

Until now, the analysed data concerned Baha devices. Figure 3.3 shows similar data from all implantable systems available (systematic literature search performed mid-2017), including besides the Baha the percutaneous Ponto device (Oticon Medical, Askim, Sweden), an active tBCD, viz. the Bonebridge device (Med-El, Innsbruck, Austria), two passive tBCDs; the Baha Attract (Cochlear BAS, Gothenburg, Sweden) and Sophono (Medtronic, Boulder, Co, USA) and the Codacs device (Cochlear, Mechelen, Belgium); the latter two devices are no longer on the market (details on the literature search are presented at the end of this section; see Note, just before section 3.2).

Altogether, 48 studies were included in this analysis. Figure 3.3A shows the mean aided free-field threshold at 0.5, 1, 2 and 4 kHz as a function of the mean bone-conduction threshold, (the SNHLc or PTA4bc). This figure again shows that, in general, when the SNHLc is below 25-30 dB HL, a negative EG, thus a remaining AB gap is found (since the data points are above the diagonal). This observation corresponds to Figure 3.2.

Figure 3.3B presents the same data as Figure 3.3A. This time, trend-curves are presented, drawn by eye through the data points for each type of device. Given the blue line, representing the trend for VSB, the aided thresholds are at a fairly stable level (25-30 dB HL) up to a SNHLc of approx. 45 dB HL. Above that value, the aided thresholds appear to deteriorate exponentially, suggesting that the device is no longer powerful enough for such degrees of the SNHLc. For the passive tBCDs (Baha Attract / Sophono), similar trends can be seen, however, with lower cut-off SNHLc value, between 15-20 dB HL. Due to their MPO values, evidently, these passive tBCDs and the VSB are not equivalents. No clear limit is found for the pBCDs (Baha/Ponto devices). This is due to the availability of three types of processors (standard, power and superpower) that were used, depending on the patients SNHLc. However, the figure suggests that the cut-off value for pBCDs lies around 50 dB HL. For the Bonebridge, no clear cut-off value was found up to 25 dB HL. Ignoring the trends, Figure 3.3A shows a rather large variability (vertical distribution) between studies. This may be clinic-related, for example due to differences in fitting protocols.

Slide3

Figure 3.3A. Mean aided free-field threshold (0.5, 1, 2, 4 kHz) as a function of the mean SNHLc (mean bone conduction thresholds at the same frequencies) as obtained from 48 studies using the percutaneous Baha/Ponto devices (black diamonds), the VSB (big blue dots), the ‘Bonebridge’ (red squares), the Baha Attract (green triangles), the Sophono (small black dots) or the Codacs (green circles). The Baha Attract and Sophono are passive tBCDs

Slide2

Figure 3.3B. The same figure as Figure 3.3A, however, this time trend lines are indicated for the VSB and the passive tBCDs

It might be argued that updates/upgrades of the devices might have led to better aided thresholds and better EG. However, an overview of more recent data (2018-2025), shows similar outcomes, see next section (3.2).

Note. To find relevant studies, it was decided to first use the studies that had been previously selected by authors who published systematic reviews. This concerned reviews by Colquitt et al., 2011; Verhaert et al., 2013; Nadaraja et al., 2013; Ernst et al., 2016; Sprinzl and Wolf-Mangele, 2016 and Dimitriades et al., 2016. Of all the included studies, only those that presented air- and bone-conduction thresholds and aided free-field thresholds were selected. Duplicates were removed as well as studies with five patients or fewer. Unpublished presentations were also excluded. Due to low numbers, additional searches in the literature were carried out. The search period was 2006-2017. Concerning the percutaneous Baha and Ponto devices, using the reviews, nine publications met the inclusion criteria (Kompis et al., 2007; Priwin et al., 2007; Flynn et al., 2009; Fuchsman et al., 2010; Ricci et al., 2011; Marsella et al., 2011; Bouhabel et al., 2012). Concerning the VSB, using the reviews by Verhaert et al. and Ernst et al., 14 publications were included (Beltram et al. 2009; Streitberger et al. 2009; Colletti et al. 2011; Baumgartner et al., 2010; Frenzel et al., 2010; Beleites et al. 2011; Bernardeschi et al., 2011; 2013). Sprinzl and Wolf-Mangele (2016) reviewed the outcomes of ‘Bonebridge’ application. Only two studies met the criteria (Sprinzl et al., 2013; Ihler et al., 2016). Dimitriades et al. (2016) reviewed the transcutaneous Baha Attract; three studies were included (Iseri et al., 2014; 2015; Briggs et al., 2015). Since the number of included publications regarding the Baha Attract and the ‘Bonebridge’ were low and the publications with the percutaneous Baha/Ponto outcomes were rather dated (published before 2012), we performed an additional literature search, using Pubmed. For Bonebridge and Baha Attract, search terms were the respective device names together with ‘2016’; concerning ‘Bonebridge’, four studies were added (Gerdes et al., 2016; Baumgartner et al., 2016; Smerber et al., 2016; Zernotti et al., 2016), for Baha Attract, no further study with relevant audiometric data was found. Concerning Baha/Ponto, the sound-processor’s names were used; the search period was 2011-2017. In contrast to the other devices, number of included patients had to be ten or more. Seven studies were identified (Pfiffner et al., 2011; Flynn et al., 2012; Bosman et al., 2013; Kurz et al., 2013; Desmet et al., 2013; Kompis et al. 2014; Gerdes et al. 2016). Concerning the Sophono device, a systematic review has not yet been published. Therefore, we searched for studies using only ‘Sophono’ as search term and applying the introduced inclusion criteria. Six studies were included (Siegert and Kanderke, 2013; Sylvester et al., 2013; Hol et al., 2013; Denoyelle et al., 2015; Shin et al., 2016; Magliulo et al., 2015). Finally, data obtained with the Codacs device were included, which is the most powerful implantable device (Zwartenkot et al., 2014). Searching PubMed, three papers were identified comprising audiological data obtained in more than five patients (Busch et al., 2013; Lenarz et al., 2013; 2014).

3.1 Update 2026, The effect of low MPO on aided thresholds and word recognition, a further elaboration of the data presented in section 2.3.

Based on the MPO values in Table 2.2, the audiological capacity of the BCDs and MEIs can be ranked. While MPO offers a theoretical limit, real-world performance depends on aided thresholds and, relatedly, speech recognition. To verify that the differences in MPO translate into functional benefit, a targeted literature review was conducted. Contrary to the design used in the previous section, it was decided to just select studies on patients with conductive hearing loss, to create homogeneous groups of BCD and MEI users.

The aim of the review was to collect aided hearing thresholds in patients using a pBCD or active tBCD, unilaterally. The search period was 2018-Februari 2025. Studies should present bone-conduction thresholds and aided free-field thresholds.  For more details on the selection process, see Snik et al. (2026). Recently, more data have been gathered, following the same criteria, on passive tBCDs and the VSB.

Table 3.1 shows an overview of the number of enrolled studies and patients by device type, as well as the weighted average PTA4bc and ‘aided PTA4’ (averaged over studies, taking into account the number of patients). The calculated mean EG4 (0.5 to 4 kHz) is also displayed. A similar result is found when ranking the mean aided PTA4 and comparing it with that of the MPO (Table 2.2).

Table 3.1. The included number of studies and patients as well as the (weighted) mean aided PTA4 (adjusted for the number of patients per study) as well as the mean EG

Device N

Study

N

(pats)

Mean PTA4bc

dB HL

Mean aided PTA4

± s.d.*; dB HL

Mean EG4

dB

Passive tBCD 12 301 11 29±3 -18
Bonebridge 10 296 9 28±4 -19
Osia 4 239 14 25±2 -11
VSB 4 81 10 27±2 -17
pBCD 8 224 12 20±2 -8

*spread in outcomes; best estimates

In accordance with the previous section (3.1), negative EG values are found again in these patients with subnormal cochlear function. Clear differences can be seen between devices, with the aided PTA4 varying between 20 dB HL and 30 dB HL. This has consequences for speech recognition. The model presented by Killion and Mueller (2011; based on the speech intelligibility index) makes it possible to assess the word recognition score (WRS) based on the aided thresholds. For speech at a general level of 60 dB SPL (normal conversational level) with the aided thresholds at 20 dB HL, the WRS is 95% and for 30 dB HL, 85%. For slightly softer speech, although still within the range of normal conversational levels (50 dB SPL), the WRS is 90% and 70%, respectively.

In summary, there are clear differences in the audiological capacity between the different types of BCDs and VSB, not only in technical terms (MPO) but also in terms of speech recognition. Although not further explained here, device noise also pays a role. In contrast to the active tBCDs and the VSB, device noise from pBCDs is not audible (e.g., Carlsson & Hakansson, 1997).

Apparently, the pBCD has a unique audiological capacity. However, there may be practical, medical, surgical, financial and/or esthetical reasons for choosing one of the other amplification options. See next chapter.