Speech Recognition In Hearing Aids: Used For Assessing CI Candidacy In Tonal Language Population

Objective: CI (cochlear implantation) candidacy is somewhat controversial in severe hearing loss among tonal mandarin-speaking patients. To assess the relationship between pure tone audiometry (PTA) and speech recognition score (SRS), with and without hearing aid amplication, among patients who did not meet the NIH criteria of CI candidacy in tonal language mandarian-speaking countries, especially those with severe hearing loss (70 dB HL < 4FPTA(0.5, 1, 2, 4 KHz) ≤ 90 dB HL) Materials and Methods: A total of 414 patients with sensorineural hearing loss with 774 ears were reviewed retrospectively in a tertiary referral center. The Mandarin Monosyllable Recognition Test (MMRT) was used to evaluate the SRS of these ears. Results: 31% (10/32) of the 32 ears with severe hearing loss, 70-90 dB HL, still showed poor speech recognition (SRS<30%) after hearing aid amplication, while 71% (46/65) of the 65 ears with profound hearing loss, > 90 dB HL, showed poor speech recognition with hearing aid amplication. Conclusions: The speech audiometry with Mandarin Monosyllable Recognition Test (MMRT) helped identify those patients whose 4FPTA< 90 dB HL fell outside the CI candidacy criteria of NIH in tonal language mandarin-speaking countries but showed signicantly poor (SRS< 30%) speech recognition performance.


Introduction:
Hearing impairment can have a profound impact on the quality of life and has been proven to be associated with other comorbidities such as dementia and depression [1]. According to the World Health Organization (WHO), over 5% of the world's population has disabling hearing loss [WHO 2020.03.01] [2]. Hearing aids, one of the treatment options for hearing impairment, have become increasingly sophisticated over the years in the ways they provide ampli cation that ts the needs of individuals with hearing loss. However, the bene ts one can obtain for speech recognition performance via the use of welltted hearing aids have their limitations, especially for those with severe to profound hearing loss [3]. Therefore, it is crucial to use appropriate measures to precisely evaluate the effectiveness of hearing aids.
Pure tone audiometry (PTA) is the most commonly used test during an audiological evaluation as it provides information about an individual's hearing acuity. Nonetheless, PTA alone is insu cient to provide information about the speech recognition ability of the person being assessed. It has been shown that PTA cannot provide a complete picture in patients with sensorineural hearing loss (SNHL) in terms of their damaged cochlear epithelium and, thus, restricted ability in speech recognition, even with appropriate hearing aid ampli cation [4]. Therefore, speech audiometry (speech recognition score, SRS) is administered to patients with SNHL to evaluate the information-carrying capacity of the cochleae and predict hearing aid outcomes. For patients with low post-tting SRS, poor treatment adherence, and reduced quality of life are expected. An alternative treatment option, such as cochlear implantation (CI), may be better suited for these patients.
CI has been developed for more than 40 years. The criteria for CI candidacy have gradually but signi cantly expanded over the years due to surgical innovation, continued advancement in ampli cation technology, and an accumulating number of evidence-based studies [5]. According to Taiwan's National Health Insurance (NHI), adult CI candidates should meet all the following criteria: 1) having post-lingual 0.5K, 1K, 2 K, and 4 KHz > 90 dB HL, 2) obtaining SRS < 30% with a well-t hearing aid ampli cation, and 3) no contraindications to CI[6]. These criteria state that only patients with profound hearing loss (4FPTA(0.5, 1, 2, 4 KHz) > 90 dB HL) are eligible for CI.
This study aimed to assess the relationship between PTA and SRS, with and without hearing aid ampli cation, among tonal mandilion-speaking patients who did not meet the NHI criteria of CI candidacy, especially those with severe hearing loss (70dB HL < 4FPTA ≤ 90dB HL). It is hoped that with the ndings of the present study, NHI criteria for CI candidacy in tonal language patients could be further expanded.
This study was based on a retrospective analysis of data retrieved from the database of the Mackay Memorial Hospital, a tertiary medical center in Taiwan. The study was approved by the Mackay Memorial Hospital institutional review board (No. 21MMHIS071e Mackay Memorial Hospital IRB board chairman name is Yi-Shing Leu). All study methods were performed in accordance with the Declaration of Helsinki regulations. Because this study was retrospectively done by medical record review, informed consent was not routinely obtained from all our subjects. The requirement to obtain informed consent was waived by Mackay Memorial Hospital institutional review board. The results of the PTA and speech audiometry tests of 427 tonal language patients were reviewed over four years (2016-2020). Finally, data from 414 patients with 774 ears were analyzed. Ears with conductive, mixed, or retrocochlear hearing loss were excluded. Among these 414 patients, only 80 had hearing aids (46 wore hearing aids bilaterally and 34 unilaterally).
Pure tone air conduction thresholds were measured from 250 to 8000 Hz. 4FPTA (0.5, 1, 2, 4 KHz) was calculated to evaluate the severity of the hearing loss. Patients ears were classi ed into four groups according to their 4FPTA: 1) normal-hearing (4FPTA ≤ 25dBHL), 2) mild-to-moderate hearing loss (25dB HL < 4FPTA ≤ 70dB HL), 3) severe hearing loss (70dB HL < 4FPTA ≤ 90dB HL), and 4) profound hearing loss (4FPTA > 90dB HL) [7,8]. The Mandarin Monosyllable Recognition Test (MMRT) [9] was used to evaluate the speech word recognition abilities of these tonal language patients. The MMRT stimuli were presented via headphones in a quiet environment for all patients at 65 dB HL, and the maximum achievable word recognition score (unaided PBmax) was recorded in each ear. If a patient scored less than 90% SRS at 65 dB HL, the unaided PBmax of that patient would be retested with a 10dB-steps increase in presentation level or with the loudest possible presentation level for that patient. As for patients tted with hearing aids, further sound eld speech word recognition tests were also performed using the MMRT stimuli with 65 dB HL (ranging from 50-70 dB HL) to obtain the aided PBmax. The contralateral ear was properly occluded with earplugs in cases of the unilateral tting, and speech-shaped noise masking was applied when appropriate.
Logistic regression analysis was used to analyze PBmax in relation to 4FPTA. A box plot was presented to show the PBmax obtained as a function of 4FPTA. Data from patients in the severe hearing loss group were further analyzed. An aided PBmax of less than 30% was considered a poor speech word recognition performance after hearing aids [10]. Statistical analyses were performed using SPSS V25.

Results:
Demographic information of the patients in this study is shown in Table 1. Among the 774 ears, there were 151 ears with normal hearing, 419 with mild to moderate hearing loss, 83 with severe hearing loss, and 121 with profound hearing loss. The 4FPTA hearing threshold average was 15.7 ± 5.8, 50.5 ± 11.9, 80.5 ± 6.3, and 105.7 ± 9.0 dB HL in each group respectively. The age of the patients ranged from 5 to 94-years-old and the overall mean age was 58.6±23.2 years. Female patients were slightly predominant.  Figure 1A presents the unaided PBmax values of the 774 ears as a function of their corresponding unaided 4FPTA. The 4FPTA displayed on the x-axis was divided into 12 regular intervals, which started at 5 dB HL and had a unit of 10 dB in each interval. Figure 1B  however, PBmax deteriorated accordingly as the value of 4FPTA increased. Figure 1B shows that the interval of 4FPTA ranging from 81 to 90 dB HL had the greatest standard deviation. Figure 2 presents the relationship between the aided PBmax and unaided 4FPTA of 114 ears that had been tted with hearing aids. The 4FPTA displayed on the x-axis was also divided into regular intervals with a unit of 10 dB in each interval.
However, the scale starts from 41 to 50 dB and ends at 111 to 120 dB HL, as shown in Figure 2. Based on the NHI criteria for CI candidacy, only patients with aided PBmax less than 30% were considered to demonstrate poor speech recognition performance. Therefore, the percentage of patients who scored less than 30% in each 4FPTA interval was calculated as poor

Discussion:
The purpose of the present study was to examine the relationship between pure tone thresholds and speech recognition performance in tonal mandarian-speaking patients. The results of the logistic regression analysis showed that 4FPTA is highly correlated with unaided PB max in patients with normal hearing and profound hearing loss in cases where patients with 4FPTA in the range of normal hearing consistently scored higher values of unaided PB max and patients with 4FPTA in the range of profound hearing loss scored the poorest value of unaided PB max . However, the variability of speech recognition performance has increased greatly in patients with moderate and severe hearing loss, with the greatest variability occurring when the 4FPTA falls within the range of 60 to 80 dB HL. This trend in tonal mandarian-speaking patients is consistent with those reported by Hoppe et al. and Maeda et al. [11,12]. As a result, 4FPTA cannot accurately re ect an individual's speech recognition ability. In other words, 4FPTA is a poor predictor of unaided PB max , which is highly associated with different extents of cochlear hair cell damage [4,10], especially for those with 70-90 dB HL severe hearing loss. Therefore, one needs to be aware that it is possible to underestimate the severity of hearing disability among patients with severe hearing loss based solely on the information provided by PTA alone.
Although unaided PB max is a poor predictor of aided word recognition performance [13], in practice, unaided PB max is still regarded as an index of cochlear function and is used to predict the outcome bene t of hearing aid ampli cation. Another study [10] also suggested that unaided PB max should be regarded as an individual's potential upper limit of word recognition performance with hearing aid ampli cation. Therefore, the relationship between the 4FPTA and unaided and aided PB max was examined in our study to determine how 4FPTA correlated with speech recognition performance with hearing aids. As shown in Figure 1B, patients who underwent 4FPTA fell within the range of severe hearing loss (70-90 dB HL). They showed the greatest variability in their unaided PB max . Therefore, the aided PB max of the severe hearing loss patients was the main interest for this study to see how much bene t they could obtain in terms of speech recognition performance with hearing aid ampli cation, and whether their 4FPTA was in any way correlated with the aided PB max . Figure 2. results showed that about 30% of the patients with 70-90 dB HL severe hearing loss did not su ciently bene t from hearing aids ampli cation and scored an aided PB max < 30%. Our 30% data in tonal language mandarian-speaking patients is relatively similar but slightly lower than other language studies' ndings, which were mostly greater than 50% [10,14]. This difference is speculated to be caused by the different speech materials used, different test environments, and different ways of conducting speech audiometry.
Evaluating post-lingual CI candidacy has become increasingly challenging in clinical practice because of the changing and expanding CI indications [15]. Different countries and studies have adopted different sets of CI candidacy criteria. In America, CI indications included adults with bilateral moderate to profound sensorineural hearing loss and PB max ≤ 50% in the intended implant ear and < 60% in the contralateral ear [16]. In the UK, the CI indications suggested by National Institute for Health and Care Excellence included bilateral severe to profound hearing loss ( > 80 dB HL at low or more frequencies from 500 to 2000 Hz) and an aided phonemic PBmax less than 50% [17]. Although different CI candidacy criteria were adopted, the worldwide consensus is to loosen up the candidacy criteria so that individuals, especially those with severe hearing loss, who have not gained su cient bene t from hearing aids, could be considered as candidates and start the CI evaluation. Our study in tonal mandarian-speaking patients rea rms the importance of speech recognition testing in the CI evaluation process. The speech recognition audiometry using Mandarin Monosyllable Recognition Test (MMRT) could help identify those whose 4FPTA< 90 dB HL fell outside the CI candidacy criteria of NHI but showed signi cantly poorer speech recognition performance. In conclusion, our study suggest that the NHI in tonal language mandarin-speaking countries could start expanding the CI candidacy criteria to include patients who have 4FPTA< 90 dB HL but could not bene t from hearing aid ampli cation with poor SRS <30%.