Each child completed one task with each device configuration before proceeding to the next task. The order of the tasks and the coupling conditions (softband and direct) were counterbalanced across children. For the word recognition task, the children repeated 25 words from one NU-6 word list under each coupling condition. Their verbal responses were captured with a digital recorder and scored at a later time by one of the examiners. This was necessary to prevent any bias during real-time scoring (e.g., unknowingly repairing the utterance to be consistent with the stimulus) and to provide multiple attempts at scoring for children who did not or could not speak clearly.
For the auditory lexical decision task (Pittman & Rash 2016), the children were asked to listen to a word (one to three syllables in length), repeat the word aloud, and then indicate on a touchscreen monitor if the word they heard was real or not real. Their verbal responses were also recorded and scored at a later time by an examiner. Each list of 24 items contained equal numbers of real and nonsense words presented randomly. The experimental software provided visual reinforcement for correct categorizations of each word but not for incorrect responses. Overall performance was calculated only for words correctly categorized and correctly repeated.
In the rapid word learning task, the children were instructed to associate three nonsense words with three novel images using an interactive computer game (Pittman 2008, 2011). Each nonsense word was presented in singular and plural forms by including the phoneme /s/ at the end of the plural form. Singular and plural forms of the novel images were displayed on separate response buttons on a touchscreen monitor for a total of six response buttons. During the task, the children selected an image after hearing each nonsense word. If they selected correctly, a video game (e.g., dot-to-dot, puzzle) advanced one step, whereas nothing happened if they selected incorrectly. Using this trial-by-trial feedback, the children learned to associate the correct word with the correct image. The six words (three singular and three plural) were repeated 15 times each for a total of 90 randomized trials that required approximately 5 min to complete. Performance across the 90 trials was analyzed in blocks of 10 trials each. This yielded nine chronological data points. A line fitted to the data points revealed the number of trials needed to reach 71% correct performance which was considered the threshold of learning [see Pittman (2008, 2011) for a detailed description of this calculation].
For the nonword detection task, children identified nonsense words embedded into four-word sentences. Two lists of 19 sentences were compiled. Each list had five sentences with no nonsense words, eight sentences with one nonsense word (two sentences with nonsense words in each of the four word positions), and six sentences with two nonsense words (two sentences with zero, one, or two words between the two nonsense words). The 20 nonsense words in these sentences were created by replacing a single phoneme within a word(s). The replacement phonemes were selected to maintain the phonotactic probability of the original word (Vitevitch & Luce, 2004). For example, the phoneme /d/ in the word “food” was replaced with an /m/ to create the sentence “Cooks make hot foom.” Likewise, the sentence “Foy traibs move fast” originated from “Toy trains move fast.” The children were instructed to listen to each sentence and identify the position of each nonsense word or words within the sentence. To do this, they selected the appropriate word position displayed as numbered buttons (1, 2, 3, or 4) on a computer monitor. Buttons could be unselected if necessary. When satisfied, they proceeded to the next stimulus by selecting a button labeled “Next” after which reinforcement for correct responses was provided in the form of a video game.
The stimuli for all of the tasks were produced by the same female talker with a standard American English dialect. They were recorded at a sampling rate of 44.1 kHz, 16-bit resolution using a microphone (AKG, C535 EB) with a flat frequency response from 0.1 to 10 kHz (± 2 dB). All testing was conducted in a sound-treated room. The children were seated 1 m from a loudspeaker placed at 0o azimuth. Custom laboratory software was used to randomly select the stimuli, equalize the RMS level, present the stimuli at 50 dB SPL (re: calibrated position), and provide reinforcement.
The Institutional Review Board at Arizona State University approved this study. Before testing, informed assent was obtained from the children, with written consent from the parents. Each test session lasted between 1 and 2 hr. Children were paid $25 per hour for their participation.
Before comparing performance across coupling conditions, the data were subjected to regression analysis to determine if any demographic factors accounted for a significant portion of variability in performance. A separate regression analysis was conducted for each task with the difference in scores between the abutment and the softband condition entered as the dependent variable. Age, years amplification, aided threshold difference (dB HL), device output difference (dB µN) and normalized aided threshold and output differences were entered into the regression model in a stepwise fashion. Normalized values were calculated by taking the difference between the softband and abutment values relative to the abutment value. The resulting proportions represented the relative benefit or detriment from abutment use independent of the absolute degree of hearing loss or device output.
The results show that none of the variables entered accounted for the difference in softband and abutment performance for the word recognition (F(7,13) = 1.901, p = 0.226), lexical decision (F(7,13) = 1.468, p = 0.328), nonsense-word detection (F(7,13) = 1.668, p = 0.258), and the word learning (F(7,13) = 0.872, p = 0.575) tasks. On the one hand, these results indicate that the children represented a homogenous group in terms of age and hearing history. On the other hand, the results indicate that differences in aided hearing thresholds and output levels did not vary systematically with performance on any task. Thus, significant differences in performance between the two conditions on the behavioral tasks would suggest that measures of auditory threshold and device output are not sensitive indices of benefit or detriment.
Figure 6 shows word recognition scores for the abutment condition as a function of the softband conduction. Each of the 16 data points represents an individual child. Data for one 12-year-old child was unavailable due to failure to start the digital recorder during testing. The diagonal dashed line represents equal performance for each condition. The filled symbols represent the children with bilateral conductive losses and the open symbols represent the children with unilateral losses. Average performance (±1SD) for each condition is also shown (square symbol).
With the exception of one child whose word recognition improved from 0 to 40% with the abutment, little difference in performance was observed across devices. As in the study by Hol et al. (2013), word recognition improved 7%, on average, with direct stimulation via the abutment compared to indirect stimulation via the softband. Before statistical analyses, the percent-correct values were arcsine transformed to equalize the variance over the large range of scores (Studebaker 1985). The data were then subjected to a repeated-measures analysis of variance (ANOVA) with coupling condition (abutment and softband) as the within-subjects factor. The results revealed significantly higher word recognition for the abutment compared to the softband, F(1,15) = 10.3, p = 0.006, η2p = 0.41.
Lexical Decision Task
Like the word recognition task, the children’s repetition of the real and nonsense words in the lexical decision task were scored off-line by an examiner. A recording for one 15-year-old child was unavailable, again, due to failure to start the digital recorder. Each child’s verbal responses were compared to their categorization of the words (real and nonsense) to represent the performance as the percentage of words that were correctly repeated and categorized. Thus, errors for either part of the response reduced the overall score. Figure 7 shows the performance for the individual children using the same convention as in Figure 6. For this task, larger benefits from direct stimulation via the abutment occurred for many of the children, including two children with unilateral losses. On average, the children’s lexical decisions improved 13% with the abutment compared to the softband. Like word recognition, the percent-correct scores were arcsine transformed and subjected to repeated-measures ANOVA. The improvement in performance was statistically significant, F(1,15) = 11.66, p = 0.004, η2p = 0.44. These results indicate that, on average, the children were better able to differentiate familiar and unfamiliar words with direct stimulation via the implanted abutment and this benefit was greatest for the children with the poorest performance with the softband.
Additional analysis of the real and nonsense words was conducted to determine whether the benefit with the abutment occurred for one type of word (real or nonsense words) or if the benefit was universal. Figure 8 shows average (+1 SE) performance for the real and nonsense words under each coupling condition. Performance improved significantly (13%) for both types of words, real: F(1,15) = 5.474, p = 0.034, η2p = 0.27, nonsense: F(1,15) = 9.986, p = 0.006, η2p = 0.40, indicating a global benefit of direct stimulation rather than benefit for an easier (familiar words) or more difficult (unfamiliar words) aspect of the task.
Rapid Word Learning
For the rapid word learning task, learning was calculated as the number of trials needed to achieve 71% performance across the 90 trials. With this approach, faster learning is achieved with fewer trials to criterion performance, whereas slower learning requires more trials. Data for one 12-year-old child was excluded after he reported that he “figured out the trick” to doing the task under the second device condition, indicating that he did not understand the task under the first condition. His very poor performance under the first condition confirmed his observation. Because the first condition could not be repeated (i.e., words cannot be unlearned and learned again), data for both conditions were excluded. Individual performance for the remaining 16 children is shown in Figure 9 using the same convention as in Figures 6 and 7. As with the lexical decision task, the children learned at the same or faster rate with direct stimulation, including the children with unilateral hearing losses. The results of a repeated-measures ANOVA indicated significantly faster learning (fewer trials to criterion) with the abutment compared to the softband, F(1,15) = 7.63, p = 0.015, η2p = 0.34.
To better appreciate the difference in learning rates for the abutment and softband conditions, Figure 10 shows the average (±1 SE) performance for each of the nine trail bins arranged in the order they occurred during testing. The solid and dashed lines are the best fits to the data for each condition. These data reveal similarly shallow learning curves but consistently higher overall performance with the abutment than with the softband. Recall that learning speed was calculated as the number of trials needed to reach 71% performance, which for these conditions was extrapolated from the fitted learning curve. However, determining the number of trials required to reach 50% performance is sufficient to observe the significant differences in learning between these conditions. Specifically, to reach 50% performance the children required 24 trials with the abutment and 60 trials (more than double) with the softband. Because children cannot anticipate the number of repetitions they will receive when learning a new word, learning efficiently with the fewest repetitions is always optimal.
Nonsense-Word Detection Task
An important requirement for learning new words is the ability to detect unfamiliar words in the context of familiar ones. Opportunities to learn new words are lost if unfamiliar words are not, or cannot be, detected. Because the task required the children to detect nonsense words and ignore real words, signal detection theory was used to calculate a sensitivity index (d′) from the standardized hit and false-alarm rates. In this context, a higher d′ value (a dimensionless statistic) indicates that the child was better able to identify nonsense words surrounded by real words than a child with a lower d′ value. Figure 11 shows the children’s sensitivity to the nonsense words with the abutment as a function of their sensitivity with the softband. The results show that while many of the children benefited from direct stimulation via the abutment, many performed similarly with both devices and one child performed best with the softband. Repeated-measures ANOVA revealed no difference in sensitivity to the nonsense words between the two conditions, F(1,16) = 2.24, p = 0.154, η2p = 0.12, indicating that, on average, the children did not receive additional benefit from direct stimulation of the temporal bone via the abutment for this task.
Figure 12 shows the average (+1 SE) error rate for the nonsense words that were missed (filled portion of the bars) and for the real words that were incorrectly identified as nonsense (open portion of the bars). These error rates indicate that children with hearing loss are far more likely to ignore nonsense words than to misperceive real words as nonsense. These results are consisted with those of Pittman & Rash (2016) who showed that children with hearing loss unknowingly repair nonsense words into real words. Although not statistically significant, the error rate for missed nonsense words decreased 9% with the abutment, F(1,16) = 2.7, p = 0.151, η2p = 0.12, whereas little change (1%) was observed for the misperception of real words, F(1,16) = 0.49, p = 0.50, η2p= 0.03.
Last, it was of interest to determine if each child’s benefit or detriment from direct stimulation via the abutment occurred across tasks. Difference scores between the abutment and softband conditions were calculated and the number of difference scores that indicated benefit (positive scores regardless of magnitude) were counted for each child. Figure 13 shows the number of tasks in which the children’s score indicated that they benefited from the abutment compared to the softband. The data in the figure are arranged as a function of age. The results show that all but one child received benefit from the abutment for at least one task and that the younger children benefited more than the older children. The relationship between age and benefit was significant, r = 0.51, p = 0.02. Because the magnitude of the benefit is not considered in this analysis, the results indicate that the youngest children experienced benefit from direct stimulation for a wider range of auditory tasks than the older children who demonstrated benefit for just one or two tasks.
Recall that the purpose of the present study was to determine if the benefits that children receive from the implanted abutment versus the softband are limited to small improvements in speech perception or if similar or greater improvements occur for other auditory tasks important to learning and communication. Performance for four auditory tasks was examined to test the hypothesis that if direct stimulation of the temporal bone is superior to indirect stimulation, then the performance for these tasks would be better when using a device coupled to an abutment than to a softband. While small improvements in word recognition have been reported with an implanted abutment, it was reasoned that the risks associated with surgery and the commitment to care of the abutment site could be justified if the benefits of an implanted abutment extend to more than just the repetition of familiar words. The results replicated the benefits of direct stimulation for word recognition and showed additional benefits for differentiating familiar words from unfamiliar ones (auditory lexical decision task) and for learning new words rapidly.
These results are consistent with recent studies comparing standard audiologic measures obtained via direct and indirect stimulation of the temporal bone (Verstraeten et al. 2009; Finbow et al. 2015; Kara et al. 2016). An important contribution of this work is that the detection and learning tasks represent processes children use to learn new information on a daily basis. To learn new words, children must first determine if a word is known or unknown to them before they can attempt to assign a semantic (meaningful) representation to the word. If the child cannot identify an unknown word in isolation or within a sentence, he loses that opportunity to learn the new word or learn more about a word he already knows. These missed opportunities may be responsible, in part, for the smaller vocabularies of children with hearing loss compared to their counterparts with normal hearing (Pittman et al. 2005). Also, the results of the present study are consistent with previous studies showing that new-word detection and learning in children with sensorineural hearing loss is directly related to the quality of the acoustic signal they receive (Pittman 2008, 2011).
The results are also in agreement with work by Lunner and colleagues (2016) who reported benefits of direct stimulation for a cognitive task in which the listeners held information in memory for a short time. Specifically, 16 adults (26 to 78 years of age) were tested with the same configuration of bone conduction devices as in the present study. Each listener repeated the final word of 7 sentences (word recognition) and then repeated the seven words at the end of each list (recall). The results showed that although recognition was similarly high for the abutment and softband conditions (96%), significantly more words were recalled with the abutment (52%) than with the softband (46%). They also reported a difference in the high-frequency BC in-situ hearing thresholds with the abutment and softband similar to the differences reported in the present study. The authors concluded that working memory was enhanced by the more efficient energy transduction of the amplified signal via the abutment. Specifically, although the output of the device was higher when coupled to a softband, layers of hair and skin attenuated the signal transmission. Also, the softband coupling can cause the higher output of the device to saturate at lower input levels and cause additional distortion of the amplified signal.
However, the results for the nonsense-word detection task (detecting nonsense words within sentences) were inconsistent with those of previous studies. For example, Pittman et al (2017) reported that the detection of nonsense words in context improves significantly with small improvements in the acoustic signal in children with sensorineural hearing loss. It is possible that stimulation from just one bone conduction device was not sufficient to overcome ambiguities in the amplified signal with either coupling condition causing children to revert to their strong repair strategies (Pittman & Rash 2016). Although bilateral bone conduction stimulation has not been shown to improve perception of familiar words compared to unilateral stimulation (Dutt et al. 2002b), bilateral implantation may provide sufficient energy transduction to optimize children’s ability to identify new words in conversation and improve overall satisfaction with the devices (Dutt et al. 2002a).
Also unique to this project is that the children with unilateral and bilateral losses received similar benefit from the implanted abutment. Although the sample of children with unilateral losses was quite small (n = 3), they were expected to perform at the highest levels on each task because they had one normal- or near-normal-hearing ear. Instead, their performance was within the range observed for the children with bilateral losses, especially for the most difficult tasks. These results are consistent with reports stating that children with unilateral hearing losses experience significant academic, social, memory, and attention deficits (Tharpe 2008) due to functional reorganization of the brain as a result of unilateral stimulation [see Vila and Lieu (2015) for a review of the consequences of unilateral hearing loss]. Put simply, the brains of children with unilateral profound hearing loss receive auditory information from one cochlea, while the brains of children with bilateral losses (conductive or sensorineural) receive information from two cochleae. Implantation of a bone conduction device on the affected side cannot overcome the uneven stimulation to the brain but may provide a more comprehensive signal (i.e., in amplitude and bandwidth) than can be achieved with normal hearing in one ear (Christensen et al. 2010a). Direct examination of the outcomes of children with unilateral hearing losses (conductive or sensorineural) using a bone conduction device is a worthwhile area of further research.
In summary, the results indicate that direct stimulation of the temporal bone via an implanted abutment provides improved signal quality compared to indirect stimulation via a softband. In addition to small benefits for word recognition (repetition of familiar words), children can expect improved identification, repetition, and acquisition of unfamiliar words; critical processes for vocabulary and language development. The results also indicate that, with the exception of the nonword detection task, the children with the poorest performance with the softband tended to benefit the most when using the device coupled to the abutment than to the softband. Last, the younger children showed more global benefits (improved performance with the abutment for more tasks) than the older children.
The author would like to acknowledge the staff and students working in the Pediatric Amplification Laboratory at Arizona State University including Ashley Wright, Jacelyn Olson, Elizabeth Rainy, Lauren Meadows, and Beatriz de Diego-Lazaro as well as the contributions of Tove Rosenbom, Ravi Sockalingam, Liz Presson, and Jessica Ågren from Oticon Medical, Denmark, for their assistance with technical issues and participant recruitment. Last, on behalf of the research community, the author would like to thank the children and their parents who gave generously of their time to participate in this research.
Amonoo-Kuofi K., Kelly A., Neeff M., et al. Experience of bone-anchored hearing aid implantation in children
younger than 5 years of age. Int J Pediatr Otorhinolaryngol, (2015). 79, 474–480.
Briggs R., Van H. A., Luntz M., et al. Clinical performance of a new magnetic bone conduction hearing implant system: Results from a prospective, multicenter, clinical investigation. Otol. Neurotol, (2015). 36, 834–841.
Busch S., Giere T., Lenarz T., et al. Comparison of audiologic results and patient satisfaction for two osseointegrated bone conduction devices: Results of a prospective study. Otol Neurotol, (2015). 36, 842–848.
Christensen L., Richter G. T., Dornhoffer J. L. Update on bone-anchored hearing aids in pediatric patients with profound unilateral sensorineural hearing loss. Arch Otolaryngol Head Neck Surg, 2010a136, 175–177.
Christensen L., Smith-Olinde L., Kimberlain J., et al. Comparison of traditional bone-conduction hearing AIDS with the Baha system. J Am Acad Audiol, 2010b21, 267–273.
Dun C. A., Faber H. T., de Wolf M. J., et al. Assessment of more than 1,000 implanted percutaneous bone conduction devices: Skin reactions and implant survival. Otol Neurotol, (2012). 33, 192–198.
Dutt S. N., Mcdermott A. L., Burrell S. P., et al. Patient satisfaction with bilateral bone-anchored hearing aids: The Birmingham experience J Laryngol Otol Suppl, 2002a28, 37–46.
Dutt S. N., Mcdermott A. L., Burrell S. P., et al. Speech intelligibility with bilateral bone-anchored hearing aids: The Birmingham experience J Laryngol Otol Suppl, 2002b28, 47–51.
Finbow J., Bance M., Aiken S., et al. A comparison between wireless cros and bone-anchored hearing devices for single-sided deafness: A pilot study. Otol Neurotol, (2015). 36, 819–825.
Hol M. K., Cremers C. W., Coppens-Schellekens W., et al. The BAHA Softband. A new treatment for young children
with bilateral congenital aural atresia. Int J Pediatr Otorhinolaryngol, (2005). 69, 973–980.
Hol M. K., Nelissen R. C., Agterberg M. J., et al. Comparison between a new implantable transcutaneous bone conductor and percutaneous bone-conduction hearing implant. Otol Neurotol, (2013). 34, 1071–1075.
Holt B., Tripathi A., Morgan J. Viscoelastic response of human skin to low magnitude physiologically relevant shear. J Biomech, (2008). 41, 2689–2695.
Johansson M. L.. The percutaneous implant. The effects of design, host site and surgery on the tissue response
2018). Gothenburg, Sweden.Department of Biomaterials, Institute of Clinical Sciences, University of Gothenburg.
Kara A., Iseri M., Durgut M., et al. Comparing audiological test results obtained from a sound processor attached to a Softband with direct and magnetic passive bone conduction hearing implant systems. Eur Arch Otorhinolaryngol, (2016). 273, 4193–4198.
Lunner T., Rudner M., Rosenbom T., et al. Using speech recall in hearing aid fitting and outcome evaluation under ecological test conditions Ear Hear, (2016). 37(Suppl 1)145S–154S.
Nicholson N., Christensen L., Dornhoffer J., et al. Verification of speech spectrum audibility for pediatric Baha Softband users with craniofacial anomalies. Cleft Palate Craniofac J, (2011). 48, 56–65.
Pittman A. Age-related benefits of digital noise reduction for short-term word learning in children
with hearing loss. J Speech Lang Hear Res, (2011). 54, 1448–1463.
Pittman A. L.. Short-term word-learning rate in children
with normal hearing and children
with hearing loss in limited and extended high-frequency bandwidths. J Speech Lang Hear Res, (2008). 51, 785–797.
Pittman A. L., Lewis D. E., Hoover B. M., et al. Rapid word-learning in normal-hearing and hearing-impaired children
: Effects of age, receptive vocabulary, and high-frequency amplification Ear Hear, (2005). 26, 619–629.
Pittman A. L., Rash M. A.. Auditory lexical decision and repetition in children
: Effects of acoustic and lexical constraints. Ear Hear, (2016). 37, e119–e128.
Pittman A. L., Stewart E. C., Willman A. P., et al. Word recognition and learning: Effects of hearing loss and amplification feature. Trends Hear, (2017). 21, 2331216517709597.
Reinfeldt S., Håkansson B., Taghavi H., et al. New developments in bone-conduction hearing implants: A review. Med Devices (Auckl), (2015). 8, 79–93.
Rigato C., Reinfeldt S., Håkansson B., et al. Audiometric comparison between the first patients with the transcutaneous bone conduction implant and matched percutaneous bone anchored hearing device users. Otol Neurotol, (2016). 37, 1381–1387.
Snik A. F., Mylanus E. A., Proops D. W., et al. Consensus statements on the BAHA system: Where do we stand at present? Ann Otol Rhinol Laryngol Suppl, (2005). 195, 2–12.
Studebaker G. A.. A “rationalized” arcsine transform. J Speech Hear Res, (1985). 28, 455–462.
Tharpe A. M.. Unilateral and mild bilateral hearing loss in children
: Past and current perspectives. Trends Amplif, (2008). 12, 7–15.
Verstraeten N., Zarowski A. J., Somers T., et al. Comparison of the audiologic results obtained with the bone-anchored hearing aid attached to the headband, the testband, and to the “snap” abutment. Otol Neurotol, (2009). 30, 70–75.
Vila P. M., Lieu J. E.. Asymmetric and unilateral hearing loss in children
. Cell Tissue Res, (2015). 361, 271–278.
Vitevitch M. S., Luce P. A.. A web-based interface to calculate phonotactic probability for words and nonwords in English. Behav Res Methods Instrum Comput, (2004). 36, 481–487.
Keywords:Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
Bone conduction amplification; Children; Conductive hearing loss