Hearing Outcome and Predictors after Implanting Bone Conduction or Middle Ear Implants in Ears with Refractory Otitis Media

Bone conduction implants (BCIs) and middle ear implants (MEIs) are promising options for individuals with persistent chronic inflammation of the middle or outer ear. However, the structure of the middle ear is often altered in patients who undergo mastoidectomy or posterior wall removal for refractory otitis media, leaving uncertainty regarding the efficacy of hearing devices. Only a few studies have examined auditory outcomes based on the etiology of hearing impairment. We investigated hearing outcomes, including speech audiometry, in patients who underwent implantation after surgery for refractory otitis media. Our findings indicated that patients who received BCIs or MEIs achieved favorable hearing outcomes. Furthermore, a correlation was observed between the preoperative bone-conduction threshold at 1 kHz in the better ear and the sound-field threshold at 1 kHz with BCIs, whereas no correlation was observed between the preoperative bone-conduction threshold and the sound-field threshold with MEIs. This study highlights the positive impact of BCIs and MEIs in patients who undergo implantation after surgery for refractory otitis media. Additionally, our study identified parameters that predict postoperative efficacy.


Introduction
When the improvement in auditory function remains inadequate following conventional surgical interventions, such as tympanoplasty or tympano-revisions for refractory otitis media, the prevailing course of action is to employ a conventional hearing aid that operates via air conduction. Nevertheless, it is well established that the utilization of air-conduction (AC) hearing aids can pose challenges in cases of persistent chronic inflammations that impact the outer ear canal/radical cavity. In such scenarios, bone conduction

Fitting of VSB
The FMT was coupled to the round window in 12 patients and to the stapes in 1 patient. Fitting was performed by experienced audiologists, and the final gain settings were established by measuring the actual thresholds using a vibrogram.

Sound-Field Thresholds with Baha and VSB
The hearing threshold was measured using these devices. Inspection sounds were presented using audiometers (AA-H1; Rion, Tokyo, Japan) that met the Japanese industrial standards.

Word Recognition Test
To evaluate the word recognition score (WRS), the 67-S (Japanese nonsense monosyllable word-list) was evaluated 1 m from the front speaker unit in a shielded room. Answers were obtained by uttering a sound. All speech perception test grades were calculated as percentages of correct answers. Experienced audiologists performed all the tests.

Masking of the Contralateral Ear
The contralateral ear was covered with earmuffs and ear plugs or masked with narrowband noise during pure-tone and sound-field audiometry and with broadband noise during speech tests. Experienced audiologists determined the necessity of masking.

Statistical Analysis
The results were analyzed using GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA). The results are presented as mean and standard deviation. The level of significance was set at 5%. Two-group comparisons were performed using the two-tailed Wilcoxon signed-rank test. Associations between two variables were evaluated using Pearson's correlation and linear regression.

Baha: Pure-Tone Audiogram
The preoperative pure-tone and aided hearing thresholds of all patients who underwent Baha implantation are shown in Figure 1A. The mean AC and BC thresholds at frequencies of 500, 1000, 2000, and 4000 Hz on the implanted side were 88.70 ± 25.07 dB HL and 44.62 ± 17.53 dB HL, respectively.  Figure 1A,B illustrate the mean hearing thresholds in the sound field with Baha while masking the better-hearing ear. The mean thresholds were found to be 37.34 ± 8.36 dB HL. As it has been previously reported that aided postoperative sound-field thresholds are correlated with the BC threshold of the better ear [7], the BC thresholds in the better-hearing ear are depicted for comparison purposes. The mean BC threshold on the contralateral side was 31.47 ± 14.41 dB HL. The differences between the better BC and aided thresholds at each frequency are shown in Figure 2A. The results showed smaller differences at 1 and 2 kHz, whereas larger differences were observed at lower and higher frequencies. Notably, most Baha users do not use it for thresholds higher than the BC threshold.  Figure 1A,B illustrate the mean hearing thresholds in the sound field with Baha while masking the better-hearing ear. The mean thresholds were found to be 37.34 ± 8.36 dB HL. As it has been previously reported that aided postoperative sound-field thresholds are correlated with the BC threshold of the better ear [7], the BC thresholds in the better-hearing ear are depicted for comparison purposes. The mean BC threshold on the contralateral side was 31.47 ± 14.41 dB HL. The differences between the better BC and aided thresholds at each frequency are shown in Figure 2A. The results showed smaller differences at 1 and 2 kHz, whereas larger differences were observed at lower and higher frequencies. Notably, most Baha users do not use it for thresholds higher than the BC threshold.

Baha: Sound-Field Thresholds
As it has been previously reported that aided postoperative sound-field thresholds are correlated with the BC threshold of the better ear [7], the BC thresholds in the better-hearing ear are depicted for comparison purposes. The mean BC threshold on the contralateral side was 31.47 ± 14.41 dB HL. The differences between the better BC and aided thresholds at each frequency are shown in Figure 2A. The results showed smaller differences at 1 and 2 kHz, whereas larger differences were observed at lower and higher frequencies. Notably, most Baha users do not use it for thresholds higher than the BC threshold.

Baha: Functional Gain
FG was calculated as the difference between the AC threshold on the implanted side and aided sound-field threshold.  Figure 2B). The average FG value at the four frequencies was 51.36 ± 25.77 dB HL.

Baha: Functional Gain
FG was calculated as the difference between the AC threshold on the implanted side and aided sound-field threshold.  Figure 2B). The average FG value at the four frequencies was 51.36 ± 25.77 dB HL.

Baha: Correlation between Preoperative BC Threshold and Aided Threshold
Linear regression analysis was conducted to examine the relationship between the preoperative BC threshold and aided threshold to investigate the potential of the preoperative BC threshold as a predictor of the efficacy of Baha. Specifically, the mean BC thresholds at 500, 1000, 2000, and 4000 Hz for the better BC threshold were treated as independent variables, whereas the mean aided threshold was treated as the dependent variable. However, the results revealed no significant correlation between these variables, as evidenced by an R 2 value of 0.003 and a p-value of 0.798 ( Figure 3A). Conversely, when the preoperative BC threshold at 1000 Hz for the better-hearing ear was used as the independent variable and the aided threshold at 1000 Hz was used as the dependent variable, a positive correlation was observed, with an R 2 value of 0.335 and a p-value of 0.004 ( Figure 3B).

Baha: Correlation between Preoperative BC Threshold and Aided Threshold
Linear regression analysis was conducted to examine the relationship between the preoperative BC threshold and aided threshold to investigate the potential of the preoperative BC threshold as a predictor of the efficacy of Baha. Specifically, the mean BC thresholds at 500, 1000, 2000, and 4000 Hz for the better BC threshold were treated as independent variables, whereas the mean aided threshold was treated as the dependent variable. However, the results revealed no significant correlation between these variables, as evidenced by an R 2 value of 0.003 and a p-value of 0.798 ( Figure 3A). Conversely, when the preoperative BC threshold at 1000 Hz for the better-hearing ear was used as the independent variable and the aided threshold at 1000 Hz was used as the dependent variable, a positive correlation was observed, with an R 2 value of 0.335 and a p-value of 0.004 ( Figure  3B). Hz for the better BC threshold were treated as independent variables, while the mean aided threshold was treated as the dependent variable. (B) Preoperative BC threshold at 1000 Hz for the better-hearing ear was used as the independent variable and the aided threshold at 1000 Hz was used as the dependent variable.

Baha: Word Recognition Score
In each patient, the WRS was measured in the aided or unaided condition with varying sound pressures. For each patient, the sound pressure was increased until the WRS reached its peak. However, the peak WRS with Baha did not improve compared to that of Hz for the better BC threshold were treated as independent variables, while the mean aided threshold was treated as the dependent variable. (B) Preoperative BC threshold at 1000 Hz for the better-hearing ear was used as the independent variable and the aided threshold at 1000 Hz was used as the dependent variable.

Baha: Word Recognition Score
In each patient, the WRS was measured in the aided or unaided condition with varying sound pressures. For each patient, the sound pressure was increased until the WRS reached its peak. However, the peak WRS with Baha did not improve compared to that of the better ear preoperatively ( Figure 4A). Interestingly, the sound pressure (SP) required to reach the peak WRS was significantly reduced by Baha (p < 0.001) ( Figure 4B). To further investigate whether this improvement could have been predicted preoperatively, we used the preoperative SP at peak WRS as the independent variable and the improvement in SP at peak WRS as the dependent variable. Our analysis revealed a positive correlation between these variables, with an R 2 value of 0.659 and a p-value < 0.001.

VSB: Pure-Tone Audiogram
The mean preoperative pure-tone audiogram and aided hearing threshold of al tients who underwent VSB implantation are shown in Figure 5. The average AC an thresholds at 500, 1000, 2000, and 4000 Hz on the implanted side were 72.21 ± 12.57 dB and 36.92 ± 11.45 dB HL, respectively.

VSB: Pure-Tone Audiogram
The mean preoperative pure-tone audiogram and aided hearing threshold of all patients who underwent VSB implantation are shown in Figure 5. The average AC and BC thresholds at 500, 1000, 2000, and 4000 Hz on the implanted side were 72.21 ± 12.57 dB HL and 36.92 ± 11.45 dB HL, respectively. Figure 5 illustrates the average aided sound-field threshold with VSB while masking the better-hearing ear, with a mean threshold of 39.62 ± 9.77 dB HL. Figure 6A shows the disparity between the BC and aided threshold at each frequency. The results for VSB were similar to those for Baha, demonstrating smaller values at 1 kHz and 2 kHz and larger values at lower and higher frequencies. However, one difference between the two was that VSB users often utilized it above the BC threshold at 1 kHz and 2 kHz. Figure 4. Word recognition score (WRS) compared between unaided and aided condition w (A) Peak WRS in unaided condition for both ears and when aided with Baha. (B) Sound (SP) required to reach the peak WRS in unaided and aided condition. **** p < 0.0001. (C) Co between preoperative SP at peak and the improvement of SP at peak.

VSB: Pure-Tone Audiogram
The mean preoperative pure-tone audiogram and aided hearing threshold o tients who underwent VSB implantation are shown in Figure 5. The average AC thresholds at 500, 1000, 2000, and 4000 Hz on the implanted side were 72.21 ± 12.5 and 36.92 ± 11.45 dB HL, respectively.   Figure 5 illustrates the average aided sound-field threshold with VSB while masking the better-hearing ear, with a mean threshold of 39.62 ± 9.77 dB HL. Figure 6A shows the disparity between the BC and aided threshold at each frequency. The results for VSB were similar to those for Baha, demonstrating smaller values at 1 kHz and 2 kHz and larger values at lower and higher frequencies. However, one difference between the two was that VSB users often utilized it above the BC threshold at 1 kHz and 2 kHz.

VSB: Correlation between Preoperative BC Threshold and Aided Threshold
Linear regression analysis of preoperative BC and aided thresholds was performed to test whether the preoperative BC threshold could predict the effectiveness of VSB. When the mean BC thresholds of 500, 1000, 2000, and 4000 Hz were set as independent variables and the mean aided threshold was set as the dependent variable, no significant correlation was found, with an R 2 value of 0.113 and a p-value of 0.261 ( Figure 7A). Unlike the results from Baha, when the preoperative BC threshold of the better-hearing ear at 1000 Hz was set as the independent variable and the aided threshold at 1000 Hz as the dependent variable, no positive correlation was observed, with an R 2 value of 0.010 and a p-value of 0.743 ( Figure 7B).

VSB: Correlation between Preoperative BC Threshold and Aided Threshold
Linear regression analysis of preoperative BC and aided thresholds was performed to test whether the preoperative BC threshold could predict the effectiveness of VSB. When the mean BC thresholds of 500, 1000, 2000, and 4000 Hz were set as independent variables and the mean aided threshold was set as the dependent variable, no significant correlation was found, with an R 2 value of 0.113 and a p-value of 0.261 ( Figure 7A). Unlike the results from Baha, when the preoperative BC threshold of the better-hearing ear at 1000 Hz was set as the independent variable and the aided threshold at 1000 Hz as the dependent variable, no positive correlation was observed, with an R 2 value of 0.010 and a p-value of 0.743 ( Figure 7B). Hz for the better BC threshold were treated as independent variables, while the mean aided threshold was treated as the dependent variable. (B) Preoperative BC threshold at 1000 Hz for the better-hearing ear was used as the independent variable and the aided threshold at 1000 Hz was used as the dependent variable.

VSB: Word Recognition Score
The peak WRS with VSB did not show improvement compared to that of the better ear preoperatively ( Figure 8A). However, similar to the results of Baha, the SP required to reach the peak WRS significantly decreased with VSB (p < 0.001) ( Figure 8B). To investigate whether this improvement could be predicted preoperatively, we utilized preoperative SP at peak as the independent variable and the improvement in SP at peak as the dependent variable. The resulting R 2 value was 0.896, and the p-value was <0.0001, indicating a positive correlation.  Hz for the better BC threshold were treated as independent variables, while the mean aided threshold was treated as the dependent variable. (B) Preoperative BC threshold at 1000 Hz for the better-hearing ear was used as the independent variable and the aided threshold at 1000 Hz was used as the dependent variable.

VSB: Word Recognition Score
The peak WRS with VSB did not show improvement compared to that of the better ear preoperatively ( Figure 8A). However, similar to the results of Baha, the SP required to reach the peak WRS significantly decreased with VSB (p < 0.001) ( Figure 8B). To investigate whether this improvement could be predicted preoperatively, we utilized preoperative SP at peak as the independent variable and the improvement in SP at peak as the dependent variable. The resulting R 2 value was 0.896, and the p-value was <0.0001, indicating a positive correlation.

VSB: Word Recognition Score
The peak WRS with VSB did not show improvement compared to that of the better ear preoperatively ( Figure 8A). However, similar to the results of Baha, the SP required to reach the peak WRS significantly decreased with VSB (p < 0.001) ( Figure 8B). To investigate whether this improvement could be predicted preoperatively, we utilized preoperative SP at peak as the independent variable and the improvement in SP at peak as the dependent variable. The resulting R 2 value was 0.896, and the p-value was <0.0001, indicating a positive correlation.

Discussion
This report assessed the efficacy of BCIs and MEIs as hearing support devices for patients who underwent surgery for refractory otitis media. The average aided thresholds for BCIs and MEIs were 37.34 dB HL and 39.62 dB HL, respectively, which were close to the BC threshold of the better-hearing ear side for BCIs and the BC threshold of the implanted side for MEIs. The FG for BCIs was 51.36 dB on average across the four frequencies, whereas for it was 32.60 dB for MEIs. This study is the first to assess the effect of these devices only in postoperative patients with refractory otitis media. The FG of each device was comparable to that of previous reports assessing the effect of mixed or conductive hearing loss.
Several studies have compared the unaided hearing ability to the aided threshold with Baha. Cooper et al. analyzed patients in four subgroups according to previously used hearing aids (AC or BC hearing aid) and etiology (chronic suppurative otitis media or congenital hearing loss). In all subgroups, the mean sound-field warble-tone thresholds showed significant improvements from unaided to aided with Baha [10]. Béjar-Solar et al. performed Baha implantation in 11 patients with bilateral atresia and reported that the threshold improved from 64 dB in the unaided condition to 19 dB in the Baha-aided condition [14]. Burrell et al. reported the effect of Baha in nine patients with otosclerosis and found that the threshold improved from 49.4 dB in the unaided condition to 30.6 dB in the Baha-aided condition [15]. Compared with these results, the effect of a Baha in postoperative patients with refractory otitis media in this study was considered favorable. The following two factors were considered to be responsible for the large FG obtained in our report compared to other reports: (1) the air-bone gap of the postoperative patients with refractory otitis media was large (44.08 ± 12.17 dB), and (2) the BC threshold on the implanted side of the Baha was better than that on the contralateral side. It is possible that this could be attributed to damage within the inner ear caused by chronic inflammation, and there was a significant difference between the mean BC threshold of 44.62 dB on the implanted side and 31.47 dB on the contralateral side (p = 0.013). Since the Baha provides a BC threshold close to that of the better-hearing ear, in cases with a preserved BC threshold of the better-hearing ear, it is possible that a large FG could be obtained.
Several studies have compared auditory outcomes in the unaided condition with those in the aided condition using a VSB. However, only a handful of studies have analyzed the impact of a VSB on hearing with a focus on its etiology. Bernardeschi et al. conducted a study in patients with aplasia (n = 3), chronic otitis media (n = 18), otosclerosis (n = 6), and external auditory meatus stenosis (n = 2) after VSB implantation. They reported a noteworthy improvement in the hearing of all patients after VSB activation, without any discernible differences between each disease [16]. The postoperative average pure-tone threshold in the unaided condition was 63 dB (n = 24) and that in the aided condition with VSB in the free field was 24 dB (n = 22). Henseler et al. also performed long-term observations with VSB (FMT within the round window niche) implantation in 10 patients after subtotal petrosectomy and reported an average FG of 35 dB (range 25-44 dB) [11]. The etiology of hearing loss has not yet been investigated. However, a review conducted by the Australian Medical Services Advisory Committee examined the FG of patients with mixed hearing loss by categorizing them according to the severity of their condition. The results showed that the mild-moderate group had an FG range of 26.17-32.0 dB, while the severe group had an FG range of 34.5-49.3 dB [17]. The mean AC and BC thresholds of patients in this article were 72.21 dB HL and 36.92 dB HL, respectively. Because this group of patients was classified as having severe mixed hearing loss, an FG of 32.6 dB was considered low.
In this study, we examined the correlation between the preoperative and aided BC thresholds. For the Baha, as shown in Figure 3, the preoperative mean BC threshold in the better-hearing ear did not correlate with the mean-aided threshold, while the preoperative 1000 Hz better-hearing side threshold correlated with the 1000 Hz aided threshold. This may be because the aided thresholds of the Baha system were close to the better-hearing BC threshold at 1000 Hz, whereas there were discrepancies at 500 and 4000 Hz. However, the results for VSB users showed no correlation, even at 1000 Hz, as shown in Figure 7. These findings suggest that VSBs can potentially achieve a consistent aided threshold irrespective of preoperative hearing ability. The VSB was specifically designed to address the shortcomings of AC hearing aids such as distortion, occlusion, and acoustic feedback. Similar to AC hearing aids, the VSB has proven to be effective in patients with sensorineural hearing loss [18]. Therefore, it is feasible to attain FG without restrictions on the preoperative threshold. As depicted in Figure 6, VSB users in this study exhibited an aided threshold that exceeded the BC threshold of the patient.
In this study, the WRS was analyzed under both unaided and aided conditions at different SPs. Neither device improved the peak WRS. However, the SP required to reach the peak WRS was reduced to approximately 60 dB HL. This effect may be useful in speech comprehension, particularly at low SPs. Linear regression analysis was performed to determine whether the improvement in SP at peak by the devices correlated with the unaided preoperative SP at peak. Our findings indicate a strong association between higher preoperative SP at peak and the effectiveness of the devices in improving SP at peak. Conversely, it was difficult to achieve a similar effect when the initial peak SP was close to 60 dB HL. This information may be valuable in predicting the indications for surgical implants and their postoperative outcomes.
A limitation of this report is that the small number of cases did not allow us to examine postoperative middle ear conditions separately. The following postoperative conditions of refractory otitis media were encompassed in Tables 1 and 2: radical cavity, post-canal wall-up surgery, and blind sac closure. A subjective evaluation using a questionnaire should also be considered; however, this was not described in this report. This is a topic for future research.
In postoperative cases of refractory otitis media, the condition of the middle ear differed from normal and included mastoidectomy and/or removal of the posterior canal wall. However, our findings indicated that both BCIs and MEIs yielded favorable hearing outcomes. In terms of auditory characteristics, MEIs boast unilateral stimulation of the aided ear, which is anticipated to facilitate binaural effects such as summation, head shadow, and squelch. It also confers an advantage for sound localization. Conversely, BCIs stimulate both the implanted and contralateral sides, rendering it arduous to ascertain the direction. However, contralateral stimulation allows the sound originating from the afflicted ear to be perceived via the inner ear of the healthy ear, making it a suitable choice for single-sided deafness [19]. Each of these devices exhibits unique advantages, grounded not only in their auditory characteristics but also in their physical features. Representative intraoperative findings and postoperative CT images from our surgeries with each of these devices are demonstrated in Figure 9. For instance, Baha implantation requires a relatively simple procedure, which entails making a small skin incision and embedding an implant in the skull bone. When using the Baha Connect system, an abutment becomes exposed on the skin surface. On the other hand, VSB consists of a coil and an FMT emanating from an implant affixed to the temporal bone with screws. Ensuring the secure placement of the FMT on the round window niche or the ossicles necessitates a higher degree of surgical precision. Contrary to Baha, no part of the implant is exposed on the skin surface in VSB. Given these distinct characteristics of each device, selecting the appropriate one is paramount, and the presence or absence of infection in the middle ear cavity could be a significant factor. In cases where surgery is performed on the infected ear, a BCI could be suitable. A viable strategy when choosing an MEI involves undertaking preparation surgeries, such as blind sac closure of the external auditory canal or posterior wall reconstruction, to control inflammation before the implantation. Indeed, we performed preparation surgery in three cases and proceeded with the MEI implantation in the second operation (Table 2). Caution must be exercised when considering a VSB implantation for patients with an open cavity, regardless of the presence or absence of chronic inflammation, where the posterior wall of the external auditory canal has been removed. Barbara [20].
While the report suggested that corrective surgery using large cartilage grafts was effective, preparation surgery can be anticipated to be effective for open cavities. It also should be noted that assessing the brain or lateral skull with magnetic resonance imaging can be more challenging when using a VSB, as the generated artifact can interfere with the accuracy of the results. These characteristics of each device, in conjunction with those obtained in the present investigation, could help in selecting an appropriate hearing support method for patients who have undergone middle ear surgery and in predicting the effectiveness of the selected device.
removed. Barbara et al. reported postoperative complications in approximately half of the 21 adult cases where VSB was implanted in an open cavity, with lead exposure (23.8%) being the most common complication [20]. While the report suggested that corrective surgery using large cartilage grafts was effective, preparation surgery can be anticipated to be effective for open cavities. It also should be noted that assessing the brain or lateral skull with magnetic resonance imaging can be more challenging when using a VSB, as the generated artifact can interfere with the accuracy of the results. These characteristics of each device, in conjunction with those obtained in the present investigation, could help in selecting an appropriate hearing support method for patients who have undergone middle ear surgery and in predicting the effectiveness of the selected device.

Conclusions
This study investigated the hearing outcomes in patients who underwent MEI or BCI implantation after surgery for refractory otitis media. Furthermore, we examined the preoperative parameters that may serve as predictors of the effectiveness of hearing outcomes after implantation. The findings of this study have significant implications for selecting suitable hearing devices for postoperative patients with refractory otitis media and predicting the effect of such devices.

Conclusions
This study investigated the hearing outcomes in patients who underwent MEI or BCI implantation after surgery for refractory otitis media. Furthermore, we examined the preoperative parameters that may serve as predictors of the effectiveness of hearing outcomes after implantation. The findings of this study have significant implications for selecting suitable hearing devices for postoperative patients with refractory otitis media and predicting the effect of such devices.
Funding: This work was supported by JSPS KAKENHI Grant 22K16914 awarded to T.N.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Review Board of the University of Miyazaki (protocol code O-0702).