This is a retrospective and concurrent study conducted in an otologic referral center (Otolaryngology Department, Hospital Universitario, Fundación Santafé, (FSFB), Bogotá, Colombia). 37 patients were evaluated with the APHAB and GBI questionnaires, and 32 patients with audiometric testing, which includes 25 patients fitted with unilateral BAHA and 7 fitted with bilateral BAHA (39 ears), between 2003 and 2011. All patients with osseointegrated BAHA were invited to participate in this study. The study was approved by the ethics committee of Fundación Santafé de Bogotá.

Audiometric measurement included sound-field pure-tone audiometry and sound-field speech audiometry. They were taken at clinical consultation with and without the semi-implantable bone-conduction sound processor fitted in place. Air-conduction thresholds in sound-field pure-tone audiometry at frequencies 0.25, 0.5, 1, 2, 3 and 4 kHz were recorded using calibrated warble tones [26] . Hearing improvement was calculated by subtracting the aided thresholds at different frequencies from air-conduction thresholds without the device.

Speech audiometry was performed using a list of bisyllabic spondees Spanish words to calculate the speech reception threshold and maximum speech discrimination (% and dB). Sound stimuli were delivered from 2 loudspeakers placed at horizontal azimuth of 0 degrees [26] [27] . All the measurements were conducted in sou- nd-treated double-walled booths by 1 experienced audiologist [19] .

Two questionnaires were used in this study: the Glasgow Benefit Inventory (GBI) and Abbreviated Profile of Hearing Aid Benefit (APHAB) and the subjects completed both questionnaires in a follow up consult and were supervised by an audiologist and a medical student who explained the different questions when needed. Children always responded to the questionnaires in company of their parents.

The GBI is a retrospective generic quality-of-life questionnaire developed by Robinson et al. [28] to measure outcomes after otorhinolaryngologic procedures. It is sensitive to changes in health status that result from an intervention, and it enables comparisons between different interventions. 18 items cover three domains, 12 related to general improvement, 3 to social improvement, and 3 to physical improvement. Responses can be given on a 5-point Likert scale. Scores range from −100 (maximum lack of benefit), to 0 (no benefit), to + 100 (maximum benefit) [29] . We used the Spanish translation given by the Institute of Hearing research web site [29] .

The APHAB is a hearing disability specific questionnaire that assesses auditory functioning with 24 items scored in four 6-item subscales [30] . It produces scores for unaided and aided conditions, and benefit is calculated by comparing the patient’s reported difficulty in the unaided condition with their difficulty with amplification. Three of these subscales address speech understanding in various everyday environments: ease of communication (EC, under relatively favorable conditions), listening under reverberant conditions (RV, communication in reverberant rooms), and listening in background noise (BN, in settings with high background noise levels). The aversiveness (AV) of sounds subscale measures the negative reactions to environmental sounds. The APH- AB has a scoring scale from 1 to 99; the higher the score, the greater the hearing disability. An overall difference in the scores of more than 10 points for a given subscale (EC, RV, BN, and AV) was considered statistically significant [19] [30] . We used the Spanish version of the APHAB provided by the University of Memphis website [31] .

The audiometries, GBI and APHAB questionnaires variables were explored using medians and interquartile ranges. For each audiological test, d_i is defined as the difference between the results of the test between the i ear with and without BAHA. We intended to prove the null hypothesis (H0): No difference exists between the results with and without the bone-anchored hearing device, as opposed to the alternative hypothesis (HA): The bone-anchored hearing aid results in audiological and audiometric benefits. In order to prove/reject these hypotheses, we used the Wilcoxon signed rank test for paired samples due to the lack of normal distribution of data. The same methodology was used to evaluate the differences in the APHAB for the different subscales and for the global scores. Results for the GBI questionnaire between mixed and conductive hearing loss were assessed using the U Mann-Whitney test. This test was used due to the presence of independent samples and ordinal variables. In addition, the U Mann-Whitney test was also used to determine if the difference between unaided and aided results were determined by the type of hearing loss in the implanted ear or by the type of the processor. Moreover, we used a regression model that adjusted baseline hearing levels in order to determine basal differences between groups, where gains were analyzed as the dependent variable, patient groups as the factor variable and the basal level scores as covariates. A level of significance of α = 0.05 was chosen, and a p value lower than α rejected the null hypothesis. The statistical analysis was performed using the statistical software, SPSS.

Regarding sound-field pure tone audiometry results, we found a statistically significant gain in all frequencies using the bone-anchored device (p < 0.001) (Table 2, Figure 1). Moreover, sound-field speech audiometry results showed a significant increase in maximum speech discrimination at 60 dB (p < 0.001) (Figure 1).

The global APHAB score significantly decreased 28 points (from 55% to 27%, p < 0.001). Mean scores for all the subscales (EC, RV, BN and AV) were all above the difference of 10-point level. Scores in all subscales significantly decreased, except for the aversiveness subscale, which significantly increased 49 points from 6.83 to 56 (p < 0.001) (Table 2).

We found statistically significant gain on the GBI overall scale, and general and physical benefit subscales (Table 3) on mixed hearing loss patients compared to conductive hearing loss patients; except for the social support

Median values and interquartile ranges and Wilcoxon p values.

Median and interquartile ranges and wilcoxon p values.

scale, where no statistically significant results were found. The mean scores for all items in both groups were all above 3, suggesting quality of life improvement in conductive and mixed hearing loss patients [24] . Sensorineural hearing loss was not analyzed due to the small number of patients with this type of hearing loss.

Table 4 summarizes the audiological and APHAB results without the device and comparisons between audiological and APHAB gains depending on the type of hearing loss in the implanted ear. Without the device, statistically significant audiometric differences were found in the two groups of patients: conductive hearing loss in comparison with mixed hearing for air-conduction thresholds of 500 Hz, 3 KHz and 4 KHz (p < 0.001).

With the device, audiometric results show a gain in the group of patients with conductive hearing loss for sound-field pure tone audiometry at the 1 KHz frequency. Moreover, adjusted results for the unaided condition (without) only revealed statistically significant results for the air conduction threshold PTA and for frequencies at 1 KHz and 2 KHz.

In terms of APHAB results for the unaided condition (without the device) and for the aided condition (with the device), no statistically significant differences were found between the two groups of patients with conductive and mixed hearing loss (Table 4).

No statistically significant differences between BP100 and Divino users were found for audiological or APHAB results (Table 5). Adjusted results for unaided condition (without) didn’t show statistically significant differences both audiometrically and for the APHAB results. However, the BP100 did show greater overall gains in all subscales.

First, our results correlate with those published by other authors, [11] [19] [32] - [34] who found statistically significant improvements in nearly all audiometric results (sound-field pure-tone audiometry and sound-field speech audiometry) and subjective satisfaction measures. Our study validates their findings in our cohort, as we also found an increase gain in all frequencies in pure-tone audiometry using the device. Unadjusted results for the unaided condition (without) revealed no statistically significant differences for sound-field pure tone audiometry between conductive hearing loss and mixed hearing loss patients. However, adjusted results for the unaided condition (without) did reveal statistically significant gains for conductive hearing loss patients compared with mixed hearing loss in air-conduction thresholds 1 kHz, 2 kHz and PTA, which correlates with results found in other surveys [19] .

^aMedian values and interquartile ranges; ^bU Mann-Whitney test, p values; ^cRegression model p values for gain adjusted for hearing.

^aMedian values and interquartile ranges; ^bU Mann-Whitney test, p values; ^cRegression model p values for gain adjusted for hearing.

Again, our results correlate with published reports indicating global subjective audiological benefits, but still lack of benefit of the aversiveness subscale [19] [34] [35] . Our report further highlights the need for improvement in auditory aversive situations. Similarly to the results published by Boleas-Aguirre [19] , we didn’t find statistically significant differences between conductive and mixed hearing loss without and with the processor.

Regarding the Glasgow Benefit Inventory results, most of the subscales revealed statistically significant differences between conductive hearing loss and mixed hearing loss patients. Studies have compared GBI results between bilateral and unilateral patients [11] [24] revealing more subjective and objective benefit for bilateral patients; however, statistically significant differences between conductive and mixed hearing loss patients for GBI results have yet to be proved. We argue that due to the small number of patients in studies comparing conductive and mixed hearing loss patients, it is still difficult to thoroughly compare quality of life between these two groups. The precision of the different results are very low due to the small number of the patients.

Our findings didn’t show any statistically significant audiometric differences in gain between Divino and BP100 users. However, BP100 users showed more clinically significant gains in the APHAB global score and Ease of communication subscale. Likewise, one study by Boleas-Aguirre et al. compared audiometric and APHAB results between the Compact, Divino and Intenso processors and didn’t find any difference between these three processors [19] . Another study by Wazen et al. stated that the Intenso processor resulted in better subjective hearing satisfaction than the Divino [36] . Specifically, a recent study by Pfiffner et al. showed that speech understanding in noise is significantly better with the Baha BP100 than with the Baha Divino [37] . Moreover, in patients with single-sided deafness, significantly better results for the Background Noise and Reverberant Conditions subscales, and better average results for the Ease of Communication scale, have also been found when comparing the BP100 processor with the other model [38] . Therefore, our study further highlights the audiological subjective benefits provided by the BP100 in comparison with the Divino model in our specific studied group.