(BC) auditory steady-state responses

Transcription

1 Original article Downloaded By: [Anne Small, Susan] At: 16:8 24 May 28 Susan Anne Small David Richard Stapells School of Audiology and Speech Sciences, The University of British Columbia, Canada Key Words Auditory steady-state responses Maturation of bone-conduction hearing Infant bone-conduction thresholds Abbreviations AABR: Automated auditory brainstem response ABR: Auditory brainstem response ANOVA: Analysis of variance ANSI: American National Standards Institute ASSR: Auditory steady-state responses BC: Bone-conduction DPOAE: Distortion product otoacoustic emissions db HL: Decibels hearing level db nhl: Decibels normal hearing level db SPL: Decibels sound pressure level db re:1mn: Decibels re: 1 micronewton df: Degrees of freedom DPOAE: Distortion product otoacoustic emissions EEG: Electroencephalogram FFT: Fast Fourier transform F: Fisher s F ratio ms: Milliseconds n: Sample size nv: Nano Volt p: Probability RETFL: Reference equivalent threshold force levels SD: Standard deviation International Journal of Audiology 28, 113, ifirst article Abstract The objective of this study was to compare boneconduction (BC) (ASSR) for infants and adults with normal hearing to investigate the time course of maturation of BC hearing sensitivity. Bone-conduction multiple ASSRs were recorded in 11-month-old (n35), and 1224-monthold infants (n13), and adults (n18). Low-frequency BC ASSR thresholds increased with age, whereas, highfrequency ASSR thresholds were unaffected by age except for a slight improvement at 2 Hz. Compared to adults, BC ASSR amplitudes for young infants were larger for low frequencies, whereas, their amplitudes were smaller or similar for high frequencies. Compared to adults, young infants are much more sensitive to low-frequency BC stimuli, and probably more sensitive to high-frequency BC stimuli; these differences between infants and adults persist until at least two years of age. Different normal levels for infants of different ages must be used and are proposed in this study. Sumario El propósito de este estudio fue comparar las respuestas por vía ósea (BC) de estado estable (ASSR) en niños y adultos con audición normal, para investigar el proceso de maduración temporal de la sensibilidad auditiva de la BC. Se registraron respuestas múltiples ASSR por BC en niños de -11 meses de edad (n35) y de meses (n13) además de adultos (n18). Los umbrales de frecuencias graves ASSR para BC aumentaron con la edad mientras que los umbrales ASSR de frecuencias agudas no se afectaron por la edad excepto por una ligera mejoría en 2 Hz. Comparado con los adultos, las amplitudes BC ASSR en niños pequeños fueron más grandes en las frecuencias graves mientras que fueron menores o similares en las frecuencias agudas. En comparación con los adultos, los niños pequeños son mucho más sensibles a los estímulos por BC en las frecuencias graves y probablemente más sensibles con los estímulos por BC en las frecuencias agudas. Estas diferencias entre niños y adultos persisten por lo menos hasta los dos años de edad. Deben usarse y se proponen en este estudio diferentes niveles normales para niños de edades diferentes. Elevation of hearing thresholds to air-conduction stimuli in infants may result from a sensorineural, conductive, or mixed hearing loss. In this case, it is necessary to obtain boneconduction thresholds, both to distinguish between sensorineural, conductive, and mixed hearing losses, and to determine the magnitude of the air-bone gap. This is routinely done in adults and should also be done when testing infants. Although bone-conduction testing is known to be essential, many clinicians continue to use only air-conduction stimuli when estimating thresholds in infants using the auditory brainstem response ISSN print/issn online DOI: 1.18/ # 28 British Society of Audiology, International Society of Audiology, and Nordic Audiological Society Received: March 22, 27 Accepted: March 1, 28 Susan A. Small School of Audiology & Speech Sciences, The University of British Columbia, 584 Fairview Avenue, Vancouver, BC, V6T 1Z3, Canada. ssmall@audiospeech.ubc.ca

2 (ABR) or auditory steady-state response (ASSR). Despite the importance of bone-conduction testing, there are only a limited number of published studies in infants that have recorded boneconduction thresholds to frequency-specific stimuli using either ABR (Cone-Wesson & Ramirez, 1997; Foxe & Stapells, 1993; Stapells & Ruben, 1989) or ASSR (Small et al, 27; Small & Stapells, 25b, 26) techniques. Collectively, the ABR studies have investigated bone-conduction thresholds for 22 neonates at 5 and 4 Hz (Cone-Wesson & Ramirez, 1997), and 36 sixmonth-old infants at 5 and 2 Hz (Foxe & Stapells, 1993; Stapells & Ruben, 1989). Our earlier ASSR studies estimated bone-conduction thresholds in 29 pre-term infants (Small & Stapells, 26) and, across three studies, a total of 35 infants 11 months of age (Small et al, 27; Small & Stapells, 25b, 26) at 5, 1, 2, and 4 Hz. There are no published frequency-specific bone-conduction threshold data (ABR or ASSR) for infants older than 11 months of age. To establish normal bone-conduction hearing levels for infants and to understand the time course of maturation for bone-conduction hearing, comparisons of bone-conduction hearing thresholds at the same frequencies for large groups of young and older infants and adults are needed. From research to date, we know that infant-adult differences exist for both ABRs and ASSRs to bone-conduction stimuli, and that these differences are age and frequency dependent. Stuart et al (1993) reported that bone-conduction click-abr (i.e. response elicited from a broad range of frequencies along the cochlear partition) thresholds in neonates are 17.5 db (in db nhl) better than those obtained for adults, and they thus suggested that the delivery of a bone-conducted signal is more effective in neonates than in adults (Stuart et al, 199, 1993; Yang et al, 1987). Cone-Wesson and Ramirez (1997) also reported better bone-conduction click-abr thresholds in neonates than adults. In contrast, Cornacchia and Morra (1983) found no difference for bone-conduction click-abr thresholds between older infants (162 months of age) and adults. For brief-tone stimuli, which are frequency specific, Foxe and Stapells (1993) found that bone-conduction ABR thresholds to brief tones for six-month-old infants were similar to adult thresholds at 5 Hz, but 5.5 db poorer compared to adults at 2 Hz. They also found that these infants had bone-conduction ABR thresholds to 5-Hz brief tones that were significantly better (1.5 db) compared to 2 Hz. Cone- Wesson and Ramirez (1997) reported that bone-conduction ABR thresholds in neonates were 2 db better at 5 Hz and 5 db better at 4 Hz compared to adults (n3). They also showed that bone-conduction ABR thresholds for neonates were 14 db better at 5 Hz compared to 4 Hz [it is also noteworthy that their 5-Hz thresholds for the neonates were 25 db better than those reported by Foxe and Stapells (1993)]. We found similar results for bone-conduction ASSRs in our previous studies of pre-term and 8-month-old post-term infants; bone-conduction ASSRs thresholds were significantly better at 5 and 1 Hz compared to 2 Hz (Small et al, 27; Small & Stapells, 26). Foxe and Stapells (1993) found no significant difference in 5- and 2-Hz ABR thresholds (in db nhl) in adults; similarly, bone-conduction ASSR thresholds (in db HL) in adults show little or no difference across frequency (Small & Stapells, 25). Frequency-specific bone-conduction thresholds in infants older than 11 months of age are needed to investigate when bone-conduction thresholds become adult-like. The pattern of amplitudes of bone-conduction ASSRs in young infants as a function of frequency are similar to previously reported bone-conduction ABR amplitudes in infants. Low-frequency bone-conduction ABR amplitudes are larger than those to high frequencies (Cone-Wesson & Ramirez, 1997; Foxe & Stapells, 1993; Stapells & Ruben, 1989); lowfrequency bone-conduction ASSR amplitudes are also larger than those to high-frequency stimuli for pre- and young postterm infants (Small & Stapells, 26; Small et al, 27). Direct comparison of ABR and ASSR amplitude differences across age groups cannot be made because of the additional contribution of the 4-Hz response to the adult ABR (when using a 4/second rate) that is not present in the infant ABR (Foxe & Stapells, 1993). However, comparisons between infants and adults can be made for 8-Hz ASSR amplitudes; adults have larger boneconduction ASSRs than infants at 4 db HL for 2 and 4 Hz but show little difference compared to young infants at 5 and 1 Hz (Small & Stapells, 26). The purpose of the present study was to compare multiple ASSR thresholds to bone-conduction stimuli in infants of different ages to adults*all with normal hearing*to investigate the time course of maturation of bone-conduction thresholds across frequency. It was also the purpose of this study to establish normal levels for bone-conduction ASSRs for young and older infants. To this end, new data as well as previously published bone-conduction ASSR data for three different age groups were combined and compared in this study. Boneconduction ASSRs were investigated for a group of older infants, a group not previously assessed. A larger group of young infants was created for this study by pooling data from the smaller groups of young infants presented in previously published studies (Small et al, 27; Small & Stapells, 25b, 26). Finally, a larger group of adult subjects was formed in order to compare infant and adult bone-conduction ASSRs by pooling new bone-conduction ASSR threshold data collected in this study with previously published adult data (Small & Stapells, 25). Materials and Methods Participants Two groups of infants who passed a hearing screening in both ears and a group of adults with normal hearing (pure-tone behavioural air- and bone-conduction thresholds 525 db HL at 54 Hz) participated. ASSRs to bone-conduction stimuli were recorded in 35 young infants [age range of.544. weeks; mean age of 16. weeks], 13 older infants [age range of 1224 months; mean age of 18.2 months], and 18 adults [age range of 1948 years; mean age of 22.9 years] recruited from the community. Data for the 35 young infants were previously reported as small groups in three separate studies (nine infants from Small et al, 27; 14 infants from Small & Stapells, 26; 12 infants from Small & Stapells, 25b); these small groups were combined to obtain one large group of young infants. Data for 1 of the 18 adults presented in the present study were also previously reported (Small & Stapells, 25) and combined with new data from eight additional adults to form a larger group of adults. Bone-conduction ASSRs in older infants were recorded 2 International Journal of Audiology, Volume Number

3 for the first time in this study. (Note: The ages of the infants from 24 months were not evenly distributed, therefore, the infants were divided into two age groups, 11 and 1224 months of age. The age distribution for the infants tested is shown in Figure 4). Four of the infants recruited from the community were screened using an automatic auditory brainstem response (AABR) screening test at 35 db nhl. The hearing of the other infants was screened using a distortion-product otoacoustic emissions (DPOAE) screening test. The pass criterion for the DPOAE screening was a signal-to-noise ratio5 db at 2, 3, and 4 Hz in both ears. Infants who passed the AABR or DPOAE hearing screening test in both ears were considered to be at low risk for significant hearing loss and thus included in the study. Stimuli The stimuli were sinusoidal bone-conduction tones with the carrier frequencies 5, 1, 2, and 4 Hz that were 1% amplitude and 25% frequency modulated at , , and Hz, respectively. All stimuli were presented simultaneously (i.e. multiple). The stimuli were generated by the Rotman MASTER research system (John and Picton, 2a) using a buffer length of 25 6 points and a digital-to-analog rate of Hz, which is an integer sub-multiple of the 2 MHz clock rate, but not an integer multiple of the carrier frequencies. The stimuli were then attenuated through Tucker- Davis Technologies HB6 and SM3 modules. Before the stimuli were attenuated, they were routed through the Stanford Research Systems Model SR65 to increase the gain of the stimulus by 1 db. The bone-conduction stimuli were presented to a Radioear B-71 bone oscillator which was coupled to the head by an elastic headband for the infants and adults with 445 g of force, except for four of the young infants who had the bone oscillator held in place by hand. The bone oscillator was placed on the temporal bone slightly posterior to the upper part of the pinna for all infants and adults tested (for a detailed description of placement location, see Small et al, 27). Bone-conduction stimuli were presented using 1-dB steps from 5 to -1 db HL [intensities greater than 5 db HL for multiple stimuli result in non-linearities in the oscillator output (Small & Stapells, 24)]. ASSRs for all infants and adults were elicited by a non-inverted stimulus polarity (Small & Stapells, 24). Calibration The bone-conduction stimuli were calibrated in reference equivalent threshold force levels (RETFL) in db re:1mn corresponding to db HL for the mastoid (ANSI S ) using a Brüel and Kjaer Model 2218 sound-level meter and Model 493 artificial mastoid. The oscillator was coupled to the artificial mastoid with 55 g of force. The four sinusoidal boneconduction tones were each calibrated separately in db HL (e.g. a presentation level of 4 db HL for ASSR stimuli presented simultaneously means that each stimulus was calibrated to 4 db HL then combined). ASSR recordings ASSRs were recorded using the Rotman MASTER system. Three electrodes were used to record the electrophysiologic responses for 23 of the 35 young infants, and for all older infants and adults: the non-inverting electrode was placed midline at the high forehead, the inverting electrode was positioned midline at the nape of the neck, just below the hairline, and an electrode placed at the low forehead acted as ground. For 12 of the 35 young infants, four electrodes were used to record the electrophysiologic responses: two inverting electrodes were placed low on the left and right mastoids instead of one inverting electrode positioned at the midline at the nape of the neck 1. All interelectrode impedances were below 3 kohms at 1 Hz. The electroencephalogram (EEG) was filtered using a 325 Hz filter (12 db/oct) and amplified 8 times (8X in Nicolet HGA-2A and Nic51A; 1X in NIDAQ card). The EEG was further filtered using a 3-Hz lowpass anti-aliasing filter [Wavetek Rockland Model 852 (48 db/oct)]. The EEG was then processed using a 125-Hz analog-to-digital conversion rate (Small & Stapells, 24). Each EEG recording sweep was made up of 16 epochs of 124 data points (.819 seconds per epoch) and lasted a total of seconds. Artifact rejection was set to eliminate epochs of electrophysiologic activity that exceeded94 mv in amplitude in order to reduce contributions to the EEG due to muscle artifact. ASSRs were averaged in the time domain then analysed online in the frequency domain using a fast Fourier transform (FFT). Weighted averaging (John et al, 21) was used. The FFT resolution was.76 Hz over a range of to 625 Hz. Amplitudes were measured baseline-to-peak and expressed in nv. An F-ratio was calculated by MASTER to estimate the probability that the amplitude of the ASSR at the modulation frequency for each carrier frequency was significantly different from the average amplitude of the background noise in adjacent frequencies within96 bins of the modulation frequency ( noise ) (John and Picton, 2a). A minimum of seven sweeps were recorded for each test condition. A response was considered to be present if the F-ratio, compared to the critical values for F(2, 24), was significant at a level of pb.5 for at least two consecutive sweeps. A response was considered to be absent if p].5 and the mean amplitude of the noise was less than 11 nv. Alternatively, a response was also considered to be absent when response amplitude wasb1 nv and the p value].3. Procedure Testing was conducted in a double-walled sound-attenuated booth. The ambient noise levels in the sound-attenuated booth for one-octave-wide bands centred at 5, 1, 2, and 4 Hz were 12, 1, 1, and 12 db SPL, respectively. All participants were tested with ears unoccluded. Infant participants were only tested when quiet and asleep; adults were tested when relaxed or asleep while reclined in a comfortable chair. Hearing screening was performed in both ears for the infants at the beginning of the test session to establish that the participants were unlikely to have hearing loss. Behavioural pure-tone air-and bone-conduction thresholds were obtained for adults at the beginning of the test session to ensure that they had normal hearing. Multiple ASSRs were elicited to bone-conduction stimuli beginning at a randomized starting intensity. Threshold for each carrier frequency was determined using a bracketing technique adjusting the presentation level using 1-dB steps. The lowest intensity at which a response was present was considered threshold. In some cases, a response did not reach significance at a level above Small/Stapells 3

4 threshold ; if only one response did not reach significance, the lowest level at which the response was present was considered threshold. If two or more responses were not significant, the response at the lowest level was considered a false positive and the lowest level above the absent responses was deemed threshold. The total recording time was approximately hours, including the time to obtain screening test results. Procedures for this study were approved by the University of British Columbia Behavioural Research Ethics Board. The participants (or parents) signed a consent form before commencing any of the experiments; all participants were paid an honorarium at the end of each session. Data analyses Mean amplitude values were averaged across subjects, including ASSR amplitudes for responses that were not significant. The phase values from MASTER were adjusted by adding 98 to yield the onset phase (John and Picton, 2b). Onset phase values were then converted to phase delay by subtracting the onset phase value from 368. Any phase-delay values that differed ]188 from an adjacent measure were unwrapped by adding 368 to their value (John and Picton, 2b). Phase values for ASSRs that were not significant were excluded from mean phase-delay calculations. Phase-delay values were averaged across subjects. Results were only reported if at least five subjects contributed to the mean. Bone-conduction ASSR thresholds were compared across frequency and age groups. Bone-conduction ASSR amplitudes and phase delays (in ms) at 3 and 4 db HL were also compared across frequency for the two infant groups and adults. Comparisons across age groups were made using a two-way mixed-model analysis of variance (ANOVA). Huynh-Feldt epsilon adjustments for repeated measures were made when appropriate. Newman-Keuls post-hoc comparisons were performed for significant main effects and interactions. To analyse the relationship between bone-conduction ASSR threshold and age, Pearson product-moment correlation coefficients were calculated for each of the carrier frequencies (Note: each of the adults was arbitrarily assigned an age of 12 weeks). Statistical significance for the linear regression analyses was determined using a one-way ANOVA. The criterion for statistical significance was pb.5 for all analyses. Results Representative ASSR results to bone-conduction stimuli are shown for typical infant (young and older) and adult participants in Figure 1. For the 4.5-month-old infant, ASSR thresholds for 5 Hz were 1 db better compared to the older infant and 2 db better compared to the adult. For 1-Hz stimuli, both the young and older infant had thresholds that were 1 db better than the adult. At 2 Hz, this young infant s ASSR threshold was 2 db poorer than both the older infant and the adult. At 4 Hz, the young infant s threshold was 1 db poorer than the older infant and 2 db poorer than the adult. Overall, the 4.5-month-old infant s low-frequency thresholds (5 & 1 Hz) were 25 db better than the high-frequency thresholds (2 & 4 Hz), a pattern which is not seen for the 24-month-old infant or adult whose ASSR thresholds differ by only 51 db between the low and high frequencies. Figure 1. Representative bone-conduction ASSR for an individual young infant (4.5 months), an older infant (24 months), and an adult (22 years). Shown are amplitude spectra resulting from FFT analyses (7515 Hz) of the ASSRs. Filled triangles indicate responses which differ significantly from the background noise (pb.5). Open triangles indicate no response (p].5 and EEG noiseb11 nv). Threshold is defined as the lowest intensity that produced a significant response. 4 International Journal of Audiology, Volume Number

5 Detectability Figure 2 shows the percent occurrence of ASSRs for the groups of young and older infants and adults for each of the carrier frequencies. There are many more ASSRs present at 5 and 1 Hz for the infants, particularly the younger infants. At 2 Hz, the adults have more responses present than the infants, whereas at 4 Hz there is little difference in detectability for the different groups of infants or adults, except for a few more responses present in infants at 1 and 2 db HL. Threshold The effects of age group on ASSR threshold for each carrier frequency are shown in Figure 3. For low frequencies, ASSR thresholds increase (worsen) substantially with age, whereas those for high frequencies are unaffected by age except for a slight improvement for 2 Hz. For the young infants, mean ASSR thresholds at 5 and 1 Hz were 8 db better compared to the older infants and 1719 db better compared to the adults. In contrast, for young infants, mean ASSR thresholds at 2 Hz were not significantly different compared to the older infants but were 6 db poorer compared to adults. For older infants, ASSR mean thresholds were 811 db better at 5 and 1 Hz compared to adults, whereas, at 2 Hz, thresholds tended to be slightly poorer compared to adults. Mean ASSR thresholds at 4 Hz differed by less than 3 db across the three age groups. Results of an ANOVA comparing mean ASSR thresholds at 5, 1, 2, and 4 Hz across age (young infants, older infants, and adults) revealed a significant effect of age [F(2,63) 1.57, p.2], frequency [F(3,189) 15.21, pb.1], and interaction between age and frequency [F(6,189) 8.389, pb.1]. Post hoc comparisons revealed that the significant effect of age was explained by overall poorer mean ASSR thresholds in adults compared to young infants (p.7). Post hoc comparisons indicated that the significant interaction between age and frequency was explained, in part, by better ASSR thresholds for young infants compared to adults at 5 and 1 Hz, and better ASSR thresholds for older infants compared to adults at 1 Hz. The differences in ASSR thresholds at 5 Hz between the young and older infants approached significance (p.57); there were no significant differences in ASSR thresholds at 2 and 4 Hz for any of the three age groups. Figure 3 also shows that there are differences in ASSR thresholds between frequencies within each age group. Post hoc comparisons for young infants revealed that ASSR thresholds at 1 Hz were significantly better compared to other frequencies and ASSR thresholds at 2 Hz were poorer compared to all other frequencies; ASSR thresholds at 5 versus 4 Hz were not significantly different (p.931). For older infants, ASSR thresholds at 1 Hz were significantly better compared to ASSR thresholds at 2 Hz, and the difference in ASSR thresholds between 5 and 1 Hz approached significance (p.57); there were no significant differences for 5 versus 2 Hz (p.647), 5 versus 4 Hz (p.75), and 1 versus 4 Hz (p1.). For adults, ASSR thresholds were significantly worse at 5 Hz compared to 2 and 4 Hz; there were no differences in ASSR threshold for 5 versus 1 Hz (p.179), 1 versus 2 Hz (p.465), or 1 Hz versus 4 Hz (p.85). Figure 4 shows the relationship between age and threshold of individual subjects at each of the carrier frequencies. ASSR thresholds decrease (i.e. get poorer) significantly with age at 5 Hz (r.46, pb.1) and 1 Hz (r.6, pb.1). There was a slight but significant improvement in ASSR threshold with age at 2 Hz (r.28, p.25). ASSR threshold did not change with age at 4 Hz (r.7, p.584). Amplitude ASSR mean amplitudes for each age group and carrier frequency are shown in Figure 5. At 5 Hz, there was no 1 5 Hz Hz % occurrence Hz -11months 12-24months adult 4 Hz Intensity (db HL) Figure 2. Cumulative percent occurrence of subjects with significant responses for young infants (n35), older infants (n13), and adults (n18) across frequency Small/Stapells 5

6 Threshold (db HL) Hz 1 Hz 2 Hz X 4 Hz X -11 months months Age Group adult Figure 3. Mean bone-conduction ASSR thresholds (91 SD) at each carrier frequency for 35 young infants, 13 older infants, and 18 adults with normal hearing difference in ASSR mean amplitudes for the young and older infants and adults. At 1 and 2 Hz, there was no difference in mean amplitudes between the two infant groups; however, at 1 Hz, both infant groups had larger response amplitudes than adults, whereas, at 2 Hz, they had smaller response amplitudes than adults. At 4 Hz, young infants tended to have smaller ASSR amplitudes compared to both older infants and adults, whereas, there was no difference in response amplitude between older infants and adults. As shown in the top of Table 1, results of an ANOVA comparing ASSR mean amplitudes for each age group and carrier frequency at 3 and 4 db HL indicated significantly X X larger amplitudes at 4 db HL compared to 3 db HL, a significant main effect of frequency, and significant age frequency and ageintensity interactions. The age effect, intensityfrequency interaction, and the intensity frequencyage interaction were not significant. Post hoc comparisons indicated that the significant effect of frequency, pooled for age, was due to significantly larger amplitudes at 1 Hz compared to 5 Hz (p.1), 2 Hz (pb.1), and 4 Hz (p.1), and significantly smaller amplitudes at 2 compared to 4 Hz (p.2); response amplitudes also tended to be larger at 5 compared to 2 Hz but the difference did not quite reach significance (p.53). Post hoc comparisons revealed that the significant agefrequency interaction was explained by larger amplitudes at 1 Hz for younger (p.51) and older infants (p.36) compared to adults, smaller amplitudes at 2 Hz for young (p.6) and older infants (p.6) compared to adults, and smaller amplitudes at 4 Hz for younger infants compared to adults (p.45). Response amplitudes at 4 Hz for young infants also tended to be smaller compared to older infants but the difference did not reach significance (p.66). The significant interaction between age and intensity was explained by significantly larger ASSR amplitudes at 3 db HL for adults compared to young (p.4) and older (p.45) infants, and larger amplitudes for older infants compared to young infants (p.3). Phase delay As shown in Figure 6, there are age- and frequency-dependent differences in ASSR mean phase delays for young and older infants and adults. At 5 Hz, ASSR mean phase delays tended to be longer for both groups of infants compared to adults. At 2 Hz, ASSR mean phase delays were similar across age Threshold (db HL) Hz y =.15x r=.46* 2Hz y= -.6x r=.28* Hz y=.16x+2.65 r=.6* 4 Hz y=.2x r= Adult Adult Age (weeks) Figure 4. Graphical representation of linear regression analysis comparing age in weeks to ASSR thresholds at each of the carrier frequencies (age was arbitrarily set at 12 weeks for adults). Regression equations (y= mx + b; where y is ASSR threshold (in db HL), x is age (in weeks), m is the slope of the regression and b is the y-intercept) and correlation coefficients (r) are shown in the upper right corner of each graph. Significant correlations are marked with an asterisk. 6 International Journal of Audiology, Volume Number

7 Amplitude (nv) Hz -11 months months adult 2 Hz groups up to 3 db HL; however, at 4 db HL, ASSR mean phase delays were longer for both infant groups compared to adults. At 1 and 4 Hz, ASSR mean phase delay was similar across age groups. With the exception of shorter phase delays at 5 Hz compared to 1 Hz, the trend was for phase delay to become shorter as frequency increased from 1 to 4 Hz for each of the age groups, a pattern that has been reported in our previous studies (Adults: Small & Stapells, 25; Infants: Small & Stapells, 26). 1 Hz 4 Hz Intensity (db HL) Figure 5. Mean bone-conduction ASSR amplitudes (91 SD) at each carrier frequency for 35 young infants, 13 older infants, and 18 adults with normal hearing The bottom of Table 1 shows the results of an ANOVA comparing ASSR phase delays for each age group and each carrier frequency presented at 3 and 4 db HL. The statistical results revealed significantly longer ASSR phase delays at 3 compared to 4 db HL (pb.1), and a significant effect of frequency; the main effect of age was not significant. Post hoc comparisons revealed that the significant effect of frequency was due to shorter phase delays for 4 Hz compared to lower frequencies (p.1 for each comparison); and shorter phase Table 1. Bone-conduction ASSR amplitude and phase delay: Three-way mixed ANOVAs showing comparisons between intensities (3 and 4 db HL), across age groups (35 young infants, 13 older infants, and 18 adults) and carrier frequencies (5, 1, 2, and 4 Hz). Source df F o a p b Amplitude (nv) Phase delay (ms) Age 2, Frequency 3, * Intensity 1, B.1* AgeIntensity 2, * IntensityFrequency 3, AgeFrequency 6, B.1* AgeFrequency x Intensity 6, Age 2, Frequency 3, B.1* Intensity 1, B.1* AgeIntensity 2, * IntensityFrequency 3, * AgeFrequency 6, * AgeFrequency x Intensity 6, * a Huynh-Feldt epsilon (o) correction factor for degrees of freedom b Probability reflects corrected degrees of freedom *significant (pb.5) Small/Stapells 7

8 Phase delay (ms) Hz delays for 5 compared to 1 Hz (p.1), and 2 compared to 1 Hz (p.1); no significant differences in phase delay were found between 5 and 2 Hz (p.285). All of the interactions between factors were significant. The significant ageintensity interaction was explained by shorter phase delays at 4 db HL compared to 3 db HL for older infants (p.4) and adults (p.1), longer phase delays for 4 db HL for young (p.2) and older (p.2) infants compared to adults, and longer phase delays for 3 db HL for older infants compared to adults (p.7). Post hoc comparisons for the significant agefrequency interaction revealed significantly longer ASSR phase delays at 5 Hz for older infants compared to adults (p.19); phase delays for younger infants compared to adults were nearly significantly longer (p.6); whereas, no significant differences in ASSR phase delay were found at 5 Hz between infant groups (p.531). For 2 Hz, ASSR phase delays were significantly longer for younger (p.7) and older (p.2) infants compared to adults; no significant differences in ASSR phase delay were found between infant groups (p.596). For 1 and 4 Hz, no significant differences in ASSR phase delay were found for any of the age groups. Post hoc comparisons for the significant intensityfrequency interaction revealed that phase delays at 4 db HL were significantly shorter for 5 Hz compared to 1 Hz (p.1) and 2 Hz (p.3); whereas, phase delays were significantly longer for 5 compared to 4 Hz (p.5), 1 Hz compared to 2 (p.2) and 4 Hz (p.1), and 2 compared to 4 Hz (p.1). The same frequencydependent differences in ASSR phase delay were seen at 3 db HL except that the phase delays for 5 and 2 Hz were not significantly different (p.895). Post hoc comparisons revealed that the significant age intensityfrequency interaction was explained, in part, by longer ASSR phase delays at 3 compared to 4 db HL for Hz Intensity (db HL) older infants at 5 Hz (p.1) and adults at 1 Hz (p.1). Comparisons of ASSR phase delays at 4 db HL also revealed longer phase delays for young infants compared to adults at 5 (p.21) and 1 Hz (p.47), and longer phase delays for young (p.1) and older infants (p.1) compared to adults at 2 Hz. At 3 db HL, the phase delays tended to be longer for young (p.64) and older infants (p.71) compared to adults at 2 Hz, but these did not reach significance. Discussion 1 Hz 4 Hz -11 months months adult Figure 6. Mean bone-conduction ASSR phase delays (91 SD) at each carrier frequency for young infants, older infants, and adults with normal hearing This is the first study to compare bone-conduction ASSRs in young and older infants to bone-conduction ASSRs in adults. The results of this study clearly show that there are frequencydependent maturational changes in bone-conduction ASSRs and that infant-adult differences remain until at least two years of age. With maturation, bone-conduction ASSR thresholds increase (i.e. get worse) significantly at 5 and 1 Hz, improve slightly at 2 Hz, but do not change significantly at 4 Hz. Also, there are frequency-dependent differences in bone-conduction ASSR thresholds that differ with maturation. Young infants tend to have better bone-conduction ASSR thresholds in the low frequencies compared to the high frequencies (with the exception of 5 versus 4 Hz). Older infants also have significantly better thresholds for 1 compared to 2 Hz, but do not have significant differences in bone-conduction ASSR thresholds at 5, 1, and 4 Hz. For adults, there was no difference in bone-conduction ASSR thresholds at 1, 2, and 4 Hz; the only differences seen in adult bone-conduction ASSR thresholds were poorer thresholds at 5 Hz compared to 2 and 4 Hz, which is opposite to the trend observed for low frequencies in infants [but similar to air-conduction multiple-assr results (for review see Picton et al, 23)]. 8 International Journal of Audiology, Volume Number

9 In our previous study that compared bone-conduction ASSR thresholds across carrier frequency for small groups of young infants and adults (Small & Stapells, 26), we reported that young infants had better mean ASSR thresholds in the low frequencies, and slightly poorer thresholds in the high frequencies compared to adults (we also found this pattern for pre-term infants); the results reported in the present study for a larger group of young infants are consistent with these earlier findings, with the exception of a few differences in statistical results at individual carrier frequencies. For the large group of young infants in this study (n35), ASSR thresholds were significantly poorer at 5 Hz compared to 1 Hz, whereas, in our earlier study (Small & Stapells, 26), for a smaller group of infants (n14), the difference in ASSR thresholds for 5 versus 1 Hz was not statistically significant. Similarly, bone-conduction ASSR thresholds in young infants were significantly poorer at 2 compared to 4 Hz in the present study, but were not significantly different for the smaller group of young infants (Small & Stapells, 26). We also found differences for the larger group of adults in this study (n18) compared to the smaller group of adults (n1) in Small & Stapells (25); in the earlier study, there were no differences in adult bone-conduction ASSR threshold across frequency, yet we found significantly poorer thresholds at 5 Hz compared to 2 and 4 Hz in the present study. The results for the larger groups of young infants and adults probably more accurately represent bone-conduction ASSR thresholds for the population that these groups represent than the results reported previously for the smaller groups. There are also frequency-dependent differences in boneconduction ASSR amplitudes within and among age groups. Compared to adults, infants had larger response amplitudes at 1 Hz, smaller response amplitudes at 2 Hz, and no significant difference in amplitude at 5 Hz. At 4 Hz, only the young infants had significantly smaller response amplitudes compared to adults. There were no significant differences in ASSR mean amplitudes between infant groups across frequency (with the exception of larger amplitudes at 4 Hz for older infants which nearly reached significance). Within the group of young infants, ASSR amplitudes were significantly larger at 1 Hz compared to the other frequencies. For older infants, ASSR amplitudes were also larger at 1 Hz compared to 5 and 2 Hz, but not compared to 4 Hz. Bone-conduction ASSR phase delays also showed frequencydependent infant-adult differences. There were no significant differences in phase delay among age groups at 1 and 4 Hz; however, there were differences in phase delay at 5 and 2 Hz for both infant groups. Both infant groups had longer phase delays at both 5 and 2 Hz (4 db HL only). In contrast, bone-conduction ABR wave V latencies show different trends at 5 and 2 Hz when comparing infants to adults; bone-conduction ABR latencies are longer at 2 Hz in infants (Foxe & Stapells, 1993, Nousak & Stapells, 1992) [same trend as ASSR phase delay], and shorter at 5 Hz for infants (Foxe & Stapells, 1993, Nousak & Stapells, 1992) [opposite trend to ASSR phase delay]. Frequency-dependent differences were also found within age groups. For all age groups, ASSR phase delays became shorter as frequency increased from 1 to 4 Hz [expected trend based on the travelling wave theory of cochlear mechanics (von Békésy, 196)]; however, phase delays for 5 Hz at 4 db HL were shorter compared to 1 Hz and 2 Hz (4 db HL only) [opposite to the expected trend]. As discussed in Small & Stapells (26), it is not clear why infants have longer ASSR phase delays compared to adults at 5 Hz, when boneconduction ABR results show the opposite trend, or why phase delays for 5 Hz are shorter compared to 1 and 2 Hz for both infants and adults when it is expected that responses to 5 Hz would occur later in time than 1 and 2 Hz. John & Picton (2b) suggest that adding an extra cycle to the phase delay values for 5 Hz results in a phase value that is approximately equivalent to group delay or apparent latency. It may be the case that an additional cycle (equivalent to ms) must be added to adult bone-conduction ASSR phase delays at 5 Hz. The adjusted adult bone-conduction ASSR phase delays at 5 Hz would result in longer phase delays at 5 Hz compared to higher frequencies, which is more sensible, and longer bone-conduction ASSR phase delays at 5 Hz in adults compared to infants, which is consistent with bone-conduction ABR latency findings. Should the same adjustment also be made to the infant bone-conduction ASSR data? If one cycle is added to the infant bone-conduction ASSR phase delays at 5 Hz, phase delays at 5 Hz become longer at 5 Hz compared to higher frequencies, which is sensible; however, this adjustment results in shorter phase delays at 5 Hz for adults compared to infants, which are the opposite to previous ABR findings as discussed above. It is possible to justify that addition of an extra cycle is needed only for adults at 5 Hz because a 5-Hz stimulus is much more intense in infants compared to adults and the ASSR phase delay does not require adjustment so that the adult ASSR phase delays at 5 Hz remain longer compared to infants, consistent with ABR results; however, without adding a cycle to the infant bone-conduction ASSR phase delays at 5 Hz we cannot explain why 5-Hz phase delays are shorter compared to higher frequencies. These differences between ASSR phase delay and ABR latency trends at 5 Hz may be due to difficulty with the circular nature of phase measures (John & Picton, 2b) and remain a puzzle. Our ASSR results are consistent with the idea that the sensitivity of infants to bone-conducted signals differs from adults up until at least two years of age. Theories to explain the differences in bone-conduction hearing sensitivity between infants and adults were originally based on bone-conducted click-abr data. Yang et al (1987) reported longer ABR wave V latencies for bone-conducted versus air-conducted clicks for oneyear-old infants and adults, and were consistent with Weber s suggestion that the longer latencies were due to the boneconducted clicks lower frequency spectral content compared to an air-conducted click (Weber, 1983). Weber hypothesized that lower frequencies would result in a longer travelling time to reach the apical regions of the cochlea, thus explaining why ABR wave V latencies are longer for bone-conducted clicks compared to air-conducted clicks. However, for neonates, Yang et al (1987) found that ABR wave V latencies for bone-conducted clicks were shorter than for air-conducted clicks. They proposed that boneconducted clicks were more effective on the much smaller, more isolated temporal bone of the neonate to explain why their neonate results did not fit the pattern for older infants and adults. The bone-conducted click-abr latency results are complex and difficult to interpret for the following reasons: (1) latency measures are not directly related to threshold, (2) frequency-dependent changes in latency are obscured due to Small/Stapells 9

10 the broad-band nature of a click stimulus, and (3) wave V latency decreases with maturation (Ponton et al, 1992, 1993, 1994 & 1996) which may add to or cancel out changes in boneconduction hearing sensitivity for infants of different ages. It has been shown subsequently, using high-pass noise derivedband responses, that ABRs elicited by air- and bone-conducted clicks receive the same contributions from different regions across the cochlear partition (i.e. 48 Hz) (Durrant & Hyre, 1993; Kramer, 1992) proving Weber s hypothesis to be incorrect 2. These issues with click stimuli and latency measures highlight the need for estimating threshold directly using frequency-specific stimuli. Air- and bone-conducted brief-tones are frequency specific which has been shown definitively using masking techniques similar to those described above (e.g. Kramer, 1992; Nousak & Stapells, 1992), and are better suited to investigating the maturation of hearing, which is frequency dependent, than click stimuli. [Note: The misconception that the ABRs to low-frequency brief tones are not frequency specific is still held by some despite the evidence to the contrary (for review of this issue, see Stapells & Oates, 1997)]. Other theories have been proposed to explain infant-adult differences in bone-conduction hearing. Cone-Wesson and Ramirez (21) reported that bone-conduction stimuli transmit 521 db SPL more acoustic energy for 5-Hz stimuli to the external ear canal (i.e. the osseotympanic pathway for a boneconducted signal) when presented to an infant s head than when presented to an adult s head. They hypothesized that boneconduction hearing is enhanced via this pathway by 521 db at this frequency and that better bone-conduction thresholds can be explained by the overall level difference of the stimulus presented to the infant ear canal. This hypothesis has not been investigated further; however, it is possible to test this theory using the bone-conduction ABR threshold data for infants with conductive hearing losses. If an infant has a conductive loss, any low-frequency bone-conducted energy that is transmitted via air conduction will be blocked by the abnormal middle ear, and cannot enhance the effectiveness of low-frequency bone-conducted sound at the cochlea. Stapells and Ruben (1989) reported better bone-conduction thresholds at 5 Hz in infants compared to adults both for infants with normal hearing and for those with conductive losses; therefore, Cone-Wesson and Ramirez s hypothesis must be incorrect. Additionally, our recent research showed that there is no difference in infant boneconduction ASSR thresholds whether the ear canal is unoccluded or occluded (Small et al, 27). The underlying mechanism for the occlusion effect in adults is the osseotympanic pathway of bone-conducted sound (Tondorff, 1966), consequently, the absence of an occlusion effect for young infants further refutes Cone-Wesson and Ramirez s hypothesis that greater low-frequency acoustic energy in the infant ear canal explains a more effective bone-conducted stimulus at the cochlea compared to adults. In force levels (db re: 1mN), we know that adults are more sensitive to high- than to low-frequency stimuli transmitted by a bone oscillator, which is reflected in the RETFLs in db re: 1mN that correspond to db HL at the mastoid. In order to make the spectra of a bone oscillator equal across frequency (i.e. converted to db HL), force levels need to be 38., 22.5, 11., and 15.5 re: 1mN at 5, 1, 2, and 4 Hz, respectively (ANSI S ). Similarly, air-conduction stimuli calibrated in db HL take into account any differences in sensitivity that adults have to tonal stimuli across frequency; the presentation of stimuli calibrated in db HL allows us to directly compare air- and bone-conduction thresholds in adults to determine whether an air-bone gap or conductive component is present. Our results show that infants do not require the same boost in the low frequencies as adults to make bone-conducted ASSR stimuli detectable, but require approximately the same force levels as adults to detect high frequencies. It is important to understand that stimulus- and frequency-dependent sensitivities specific to infants of different ages are not taken into consideration when stimuli are calibrated in db HL. As a consequence, for infants, it is not possible to directly compare air-and bone-conduction thresholds in db HL to identify and quantify an air-bone gap. Further research is needed to be able to predict behavioural thresholds from infant bone-conduction ASSR thresholds calibrated in adult db HL. We need correction factors to account for infant-adult differences in bone-conduction hearing sensitivity across frequency at different ages. In addition to this calibration correction, we need to make adjustments for ASSRbehavioural threshold differences, and these differences need to be determined for both normal- and hearing-impaired infants. Application of these correction factors to infant bone-conduction ASSR thresholds in adult db HL would yield predicted bone-conduction behavioural thresholds. Predicted air- and bone-conduction behavioural thresholds could then be compared to identify and quantify the presence of an air-bone gap. What are the physiologic mechanisms that contribute to the infant-adult differences in bone-conduction ASSRs? It is likely that the infant-adult differences in bone-conduction ASSR (and ABR) thresholds and response characteristics relate, in part, to the size and structural differences between the infant and adult skull. The infant skull is much smaller than that of an adult. Eby and Nadol (1986) measured the dimensions of the mastoid bone in human infants and found that the width, length, and depth increase rapidly in the first two years of life. Also, flexible sutures connect the temporal bone to the other bones of the cranium in the infant skull until bony sutures develop at approximately one year of age (Anson & Donaldson, 1981), in contrast to the adult skull which is a rigid structure with fused bones. Based on better click-abr thresholds to bone- versus air-conducted stimuli in neonates, Stuart et al (199) suggested that the flexible sutures in the infant skull result in less energy dissipating to the rest of the skull, causing the temporal bone to oscillate more in isolation, thus resulting in a more effective bone-conducted stimulus across frequencies in infants. Foxe and Stapells (1993) suggested that the smaller mass of the temporal bone in infants results in a more intense signal activating the cochlea; they estimated that the bone-conducted stimulus was 917 db more effective at 5 Hz for infants compared to adults, and 12.8 db better at 2 Hz (after latency adjustment based on expected infant-adult ABR latency differences for air-conduction stimuli). The idea of a more isolated temporal bone (Stuart et al, 199) is supported by Sohmer and colleagues who found that the acceleration of vibratory energy across the infant fontanelle was 14 db less than across bone adjacent to the fontanelle for a click stimulus (Sohmer et al, 2). The presence of the fontanelle appears to contribute to significant interaural attenuation of a boneconducted signal and, as a result, to reduced dissipation of the bone-conducted signal compared to an adult skull (Stuart et al, 1 International Journal of Audiology, Volume Number

11 199). Recently, we estimated at least 13, 13, 15, and 14 db of interaural attenuation (range of at least 3 db) for a boneconducted stimulus in young infants at 5, 1, 2, and 4 Hz, respectively (Small & Stapells, 28), which is similar to the difference in acceleration of vibratory energy across the temporal bone versus the fontanelle (Sohmer et al, 2), and to interaural attenuation estimated by Yang et al (1987). Our estimation of interaural attenuation of bone-conducted signals presented at the mastoid also suggests that the amount of boneconducted signal that is transferred across the infant skull does not vary with frequency. The infant-adult differences in skull size and structure, and the frequency-independent interaural attenuation of bone-conducted stimuli in infants may support the idea of a smaller, more isolated temporal bone that is more effective at transmitting bone-conducted stimuli, but these findings fail to explain why low-frequency bone-conducted stimuli are detectable at much lower intensities compared to high frequencies in infants. The frequency-dependent pattern observed for bone-conduction ASSR and ABR thresholds differs from that observed for air-conduction stimuli. Air-conduction ASSR thresholds at 5 Hz are poorer than those at high frequencies (reviewed in: Cone- Wesson et al, 22; Picton et al, 23; Stapells et al, 25), which is opposite to the trend we report for infant bone-conduction ASSRs. ASSR threshold and amplitude changes with maturation also differ substantially for air- and bone-conduction stimuli. With maturation, air-conduction ASSR thresholds improve and amplitudes become larger at all frequencies, with larger changes seen for the higher frequencies (Rance & Tomlin, 26; Savio et al, 21). Sininger et al (1997) found that brieftone air-conduction ABR thresholds (in db SPL in the ear canal) were 3 and 24 db poorer in infants compared to adults at 5 and 4 Hz, respectively (i.e. infants required levels that were slightly more intense at 5 Hz and much more intense at 4 Hz, compared to adults). Rance and Tomlin (26) also found differences when they compared air-conduction ASSR thresholds (in db SPL in the ear canal) at 5 and 4 Hz in neonates and adults; ASSR thresholds in neonates were 28 db and 38 db poorer compared to adults at 5 and 4 Hz, respectively. Rance and Tomlin (26) and Sininger et al (1997) concluded that their infant-adult threshold differences in the neonatal period are the result of neural (auditory brainstem) development. This explanation may be feasible given the ABR work done by Ponton and colleagues that showed that synaptic transmission time continues to shorten until approximately three years of age (Moore et al, 1996; Ponton et al, 1992, 1993, 1994, 1996). It is not necessarily the case that longer synaptic transmission times result in higher ABR thresholds; however, it is possible that less efficient neural transmission may require a higher stimulus intensity to elicit a response, and result in higher ABR thresholds. If the conclusions drawn from the air-conduction ABR and ASSR data are correct (i.e. hearing thresholds improve across frequency with maturation, and perhaps more so at high frequencies, due to neural maturation rather than differences in acoustic energy that reach the cochlea), to explain our ASSR results, a bone-conducted stimulus must be more intense on a infant skull compared to an adult skull both at low and high frequencies. If immaturities in neural processes do account for 3- and 24-dB of the infant-adult ABR threshold differences at 5 and 4 Hz, respectively (Sininger et al, 1997), then the transmission of a bone-conducted stimulus must be 2 db (3 17 db) better at 5 Hz, and 22 db (242 db) better at 4 Hz. If, as suggested by Rance and colleagues, maturation of neural pathways account for at least 28 and 38 db of the infantadult difference in air-conduction ASSR thresholds, then the transmission of bone-conducted stimuli could be more effective by as much as 45 db (28 17 db) at 5 Hz and 36 db (382 db) at 4 Hz. Our results, using this explanation, would thus indicate that the transmission of bone-conducted energy by the infant skull is boosted in intensity both at low and high frequencies, but especially at low frequencies. The infant-adult differences in neural maturation hypothesized from the air-conduction brief-tone ABR and ASSR data do not take into account possible middle-ear changes with maturation that may affect the signal that actually reaches the cochlea (Keefe et al, 1993, 1994). Keefe et al (1993) compared input conductance for an air-conducted stimulus for infants 124 months of age in order to measure total middle-ear conductance which is an estimate of the power absorbed by the middle ear, i.e. power available to stimulate the cochlea. Their findings showed that, for very young infants (less than one month of age), the middle ear absorbs -1, 5, 4, and 11 db less power at 5, 1, 2, and 4 Hz, respectively, compared to adults. For infants 624 months of age, the amount of power absorbed by the infant middle ear was only 24 db less than adults for 54 Hz (Keefe et al, 1993). These middle-ear findings suggest that the contribution of neural maturation to neonate-adult differences in air-conduction ABR and ASSR thresholds may be overestimated by 11 db at 4 Hz; however, in infants older than six months, maturation of the middle ear likely plays a minor role in explaining why infants require greater air-conduction sound pressure levels at the tympanic membrane compared to adults, particularly at high frequencies. Although the effects that maturation of the neural pathways have on hearing thresholds should be the same for air- and boneconducted stimuli, it is not clear how developmental changes in the middle ear, and perhaps the outer ear, might impact boneconduction hearing thresholds. Recent studies of bone-conduction mechanisms (reviewed in Stenfelt and Goode, 25) suggest that potential contributors to bone-conduction hearing for adults include: sound radiated in the ear canal as a boneconduction stimulus (including the possible role of the mandible), middle-ear cavity radiation and ossicle inertia, inertia of the cochlear fluids, compression of the cochlear walls, and pressure transmission from the cerebrospinal fluid. Stenfelt and Goode (25) suggested that most of these possible mechanisms actually contribute very little to bone-conduction hearing in unoccluded ears, and hypothesized that inertial forces arising from skull vibrations, which then affect the movement of the cochlear fluids, are responsible for most of the transmission of vibratory energy that contributes to bone-conduction hearing. Other researchers have suggested that pressure transmission from the cerebrospinal fluid may be a major pathway for boneconduction stimuli to reach the cochlea (Freeman et al, 2; Sohmer et al, 2). When adult ears are occluded, the relative importance of different bone-conduction mechanisms change; sound pressure in the ear canal, generated from the vibratory energy of a bone-conducted signal, is thought to be the main contributor to bone-conduction hearing with occluded ear and Small/Stapells 11