Multiple-ASSR Thresholds in Infants and Young Children with Hearing Loss DOI: /jaaa

Transcription

1 J Am Acad Audiol 21: (2010) Multiple-ASSR s in Infants and Young Children with Hearing Loss DOI: /jaaa Anna Van Maanen* David R. Stapells* Abstract Background: The multiple auditory steady-state response (multiple ASSR) is a promising technique for determining thresholds for infants and children. However, there are few data for infants and young children with hearing loss where multiple-assr thresholds have been compared to frequency-specific gold standard (i.e., behavioral or tone-evoked auditory brainstem response [tone ABR]) measures. Purpose: The study compared multiple-assr and tone-abr thresholds and assessed how well normal ASSR levels differentiate normal from elevated thresholds. Research Design: Multiple-ASSR and tone-abr results (to air-conduction stimuli) were obtained in infants and young children with hearing loss or normal hearing. Study Sample: 98 infants with hearing loss (53 infants provided thresholds) and 34 infants with normal hearing. Data Collection and Analysis: Multiple-ASSR and tone-abr results were typically completed on the same day. Correlations between ASSR and ABR thresholds, linear regressions, and ASSR-minus-ABR threshold difference scores were calculated for each group (normal or hearing loss), and for both groups combined. Results: Multiple-ASSR thresholds (db HL) were strongly correlated (r 5.97) with tone-abr thresholds (db nhl) for 500, 1000, 2000, and 4000 Hz. Mean (61 SD) difference scores (ASSR-minus-ABR) were , , , and db for 500, 1000, 2000, and 4000 Hz, respectively. The previously published normal ASSR levels accurately differentiated normal from elevated thresholds. Out of 523 tests with elevated tone-abr thresholds, the multiple ASSR was normal in only 22 tests. In these 13 infants, other ASSR frequencies were elevated, and thus the infants would not have passed the ASSR. Conclusions: There are few studies of infants and young children comparing ASSR thresholds to frequency-specific gold standard measures, especially using the multiple-assr technique. The present study, comparing multiple-assr to tone-abr thresholds, nearly doubles the multiple-assr sample size in the literature. The results indicate that the multiple-assr and tone-abr thresholds are strongly correlated, and the normal multiple-assr levels of 50, 45, 40, and 40 db HL correctly classified children as having normal or elevated thresholds. However, due to the lack of air- and boneconduction data in infants with different types and degrees of hearing loss, further ASSR research is needed. Key Words: Auditory steady-state response, hearing loss, infants, tone-evoked auditory brainstem response *School of Audiology and Speech Sciences, University of British Columbia, Vancouver; British Columbia s Children s Hospital, Vancouver David R. Stapells, School of Audiology and Speech Sciences, University of British Columbia, 2177 Wesbrook Mall, Vancouver, BC, Canada V6T 1Z3; Phone: ; Fax: ; stapells@audiospeech.ubc.ca Portions of this paper were presented at the XXIth Biennial Symposium of the International Evoked Responses Audiometry Study Group, June 2009, Rio de Janeiro, Brazil. This research was supported by grants from the Canadian Institutes for Health Research (Grant # MOP67103) and the Natural Sciences and Engineering Research Council of Canada (Grant # RGPIN183923). 535

2 Journal of the American Academy of Audiology/Volume 21, Number 8, 2010 Abbreviations: ABR 5 auditory brainstem response; AC 5 air conduction; AEP 5 auditory evoked potential; ASSR 5 auditory steady-state response; BCEHP 5 British Columbia Early Hearing Program; FFT 5 fast Fourier transform; nhl 5 normalized hearing level; ppe 5 peak-to-peak equivalent; SNHL 5 sensorineural hearing loss For infants and children with hearing loss, objective estimation of hearing thresholds (i.e., evoked potential audiometry) is needed when accurate behavioral responses cannot be obtained (American Speech-Language-Hearing Association [ASHA], 2004; Joint Committee on Infant Hearing, 2007). Early identification and habilitation of hearing loss improves access to auditory stimuli, thus improving the prognosis for the development of speech and language (Yoshinaga- Itano et al, 1998; Yoshinaga-Itano and Gravel, 2001; ASHA, 2004; Hyde, 2005; Kennedy et al, 2006; Joint Committee on Infant Hearing, 2007). To facilitate speech and language development in infants and children with hearing loss, the goal of many newborn hearing screening programs, including the British Columbia Early Hearing Program (BCEHP), is identification of permanent hearing loss (of a mild degree or worse) by age 3 mo and amplification by the age of 6mo. An auditory evoked potential (AEP) with high correlation to behavioral threshold is essential for the young infant population and for those older infants and children when behavioral thresholds are unavailable. Two frequency-specific AEPs are currently of interest for infant threshold estimation: the tone-evoked auditory brainstem response (ABR) (reviewed in Sininger and Hyde, 2009; Stapells, forthcoming) and the auditory steady-state response (ASSR) to stimuli modulated from Hz (reviewed in Vander Werff et al, 2008; Cone and Dimitrijevic, 2009; Stapells, forthcoming). The tone-evoked ABR is currently the gold standard for threshold estimation for infants and young children who cannot provide accurate behavioral responses (Joint Committee on Infant Hearing, 2007). Currently of much interest, the ASSR is typically detected using objective frequency domain statistical measures of response presence/absence, in contrast to the ABR where time-domain waveforms are usually visually identified by a clinician. A unique feature of the ASSR is that responses to multiple simultaneous stimuli can be evaluated for at least four frequencies for both ears simultaneously, which may result in faster threshold estimation compared to recordings obtained for one stimulus/one ear at a time (for reviews see Picton et al, 2003; Stapells et al, 2005; Cone and Dimitrijevic, 2009). Many studies in adults and older children with SNHL have shown the ASSR to air-conduction (AC) stimuli provides a good-to-excellent prediction of pure-tone behavioral thresholds in these older subjects (reviewed in Tlumak et al, 2007; Vander Werff et al, 2008; Stapells, forthcoming). Similarly, there are numerous studies of the ASSR in children with normal hearing (reviewed in Van Maanen and Stapells, 2009; Stapells, forthcoming). Recently, in a study of multiple ASSR in infants with normal hearing, we recommended normal AC-ASSR levels of 50, 45, 40, and 40 db HL at 500, 1000, 2000, and 4000 Hz, respectively (Van Maanen and Stapells, 2009). There is a lack of information, however, about the ASSR, especially for the multiple ASSR, in children with hearing loss (Rance, 2008; Vander Werff et al, 2008; Stapells, forthcoming). Although there are many previous studies of ASSR thresholds in children with hearing loss, only a few that have studied infants or young children (i.e.,,7 yr of age) have compared ASSR thresholds to frequency-specific gold standard measures such as behavioral audiometry or tone ABR (Rance et al, 2005; Han et al, 2006; Luts et al, 2006); rather, most have compared ASSR thresholds, which reflect relatively narrow frequency regions (Herdman et al, 2002) to click-abr thresholds, which reflect broad and uncertain frequency regions (Eggermont, 1982; Stapells and Oates, 1997). No study has compared multiple-assr thresholds to tone-abr thresholds. An advantage of using tone-abr thresholds for comparison with ASSR in young infants is that both measures may be obtained in the same test session, and thus hearing is unchanged between tests. In contrast, behavioral thresholds in infants are usually obtained months later. In the current study, we compare multiple-assr thresholds to tone-abr thresholds obtained on the same day in infants and young children with normal hearing or hearing loss. The purpose of the present study was twofold. First, the study aimed to establish threshold data for multiple-assr presence in a relatively large group of infants and children with normal hearing (N5 34) or hearing loss (N5 53) with tone-evoked auditory brainstem response results. Second, the study aimed to assess how well the normal db HL levels we previously recommended (50, 45, 40, and 40 db HL for 500, 1000, 2000, and 4000 Hz, respectively; Van Maanen and Stapells, 2009) differentiate normal from elevated hearing. Participants METHOD All participants were tested at British Columbia Children s Hospital as part of their clinical testing to confirm hearing status. A total of 98 infants and young 536

3 Multiple ASSR in Infants with HL/Van Maanen and Stapells children (median age516.3 mo) with hearing loss provided data for this study, with a subset of 53 of these infants providing ASSR thresholds. Results from an additional 34 infants and young children (median age mo) with normal hearing from our previous study (Van Maanen and Stapells, 2009) were also included. Hearing thresholds for participants with hearing loss (or normal hearing status for the normal group) were confirmed by tone ABR. Of the 98 infants with hearing loss, results were obtained for 141 ears with sensorineural hearing loss (SNHL), 29 ears with conductive hearing loss, and 4 ears with hearing loss of uncertain type. Participants were considered to have hearing loss if they had any ABR threshold above normal levels of 35 db nhl at 500 Hz; 35 db nhl at 1000 Hz; 30 db nhl at 2000; or 25 db nhl at 4000 Hz (BCEHP, 2008; Stapells, forthcoming). The 34 participants in the normal group from the previous study (Van Maanen and Stapells, 2009) showed present ABRs in response to air-conducted brief tones at these normal levels for 500, 2000, and 4000 Hz for both ears. Of the 98 participants with hearing loss, 86 had ABR thresholds for 500, 2000, and 4000 Hz. Six participants were missing thresholds for one frequency (either 500 or 4000 Hz), and the remaining six participants had tone-abr thresholds for only one frequency in at least one ear. Eighteen infants in the hearing loss group had unilateral hearing loss; results for the ear with normal hearing were excluded. Five infants in the normal group had unilateral hearing loss, and data (ABR and ASSR) were included only for the normal ear. Therefore, there was no overlap between the two groups. Stimulus and Recording Parameters All recordings were obtained using the Intelligent Hearing Systems SmartEP ABR and ASSR system. Stimuli were presented using ER-3A insert earphones. Four disk electrodes were placed on the infant s scalp on: (1) on the vertex (noninverting), typically for the older group or, if fontanel was open, midline on the high forehead (typically for the younger group); (2 and 3) inverting electrode on each mastoid; and (4) ground electrode off center on the forehead. Impedance between all electrodes was less than 3 kohms. Tone ABR All tone-abr testing was carried out using the BCEHP parameters and protocols. Full details of these can be found on the World Wide Web (BCEHP, 2008) as well as in Stapells (forthcoming). A brief outline of our tone-abr methods is provided below. Brief tones were Blackman-windowed tones (fivecycle total duration) presented at 39.1/sec to one ear at a time. Calibrations for 0 db nhl were 22 db peak-to-peak equivalent (ppe) SPL at 500 Hz, 25 db ppe SPL at 1000 Hz, 20 db ppe SPL at 2000 Hz, and 26 db ppe SPL at 4000 Hz (BCEHP, 2008; Stapells, forthcoming). Recording parameters included gain of 100,000, band pass filtering from 30 to 1500 Hz (12 db/octave slope), and artifact rejection of 625mV. Recordings were averaged using a 25 msec analysis window (20,000 Hz A-D rate). At least two replications (often three) of typically 2000 trials each were obtained for each condition. The presence or absence of the tone ABR was determined visually by an experienced clinician. In order to conclude no response, the residual noise of the waveforms had to be sufficiently low enough that a small threshold-level response would not be missed. In most cases, we used the Smart-EP online measure of waveform residual noise (RN), requiring RN to be #0.08 mv to conclude no response (BCEHP, 2008; Stapells, forthcoming). Multiple ASSR Multiple-ASSR stimulus and recording parameters were the same as in our previous paper (Van Maanen and Stapells, 2009) and thus are only outlined here. Air-conducted stimuli (500, 1000, 2000, 4000 Hz) were presented as multiple stimuli to both ears simultaneously (i.e., dichotically), using modulation rates of Hz. ASSR stimuli were individually calibrated in db HL (American National Standards Institute [ANSI], 1996) from peak-to-peak equivalent db SPL measures. Rather than the SmartEP-ASSR default brief-tone stimuli, we created stimuli that more closely resembled continuous ASSR stimuli (i.e., approximately one modulation cycle in duration); details of these stimuli are provided in Table 1. The ASSR stimuli were cosine 3 -windowed sinusoids with acoustic frequency specificity similar to that of continuous sinusoidal AM 2 stimuli for ASSR (John et al, 2002) and Table 1. Details of ASSR Stimuli Frequency Ear Rate (Hz) Duration (msec) Bandwidth (Hz) * 500 Hz L R Hz L R Hz L R Hz L R *Bandwidth at 220 db. 537

4 Journal of the American Academy of Audiology/Volume 21, Number 8, 2010 much better than that of continuous AM/FM stimuli (ASSR) or five-cycle exact-blackman brief-tone stimuli (ABR). Two-channel (ipsilateral and contralateral) EEG (electroencephalogram) recordings were obtained for most participants; however, only the results from the ipsilateral EEG channel were considered when determining ASSR presence and threshold. The EEG was amplified 100,000 times and filtered using a bandpass of 30 to 300 Hz (12 db/octave slope). Similar to our previous paper (Van Maanen and Stapells, 2009), determination of significance was made by online F-statistic from the FFT. To be considered a response, signal-to-noise ratios using the SmartEP- ASSR two noise estimates both had to be significant (p,.05). Recordings were continued until the mean noise in the sidebins was #5 nanovolts or until a significant response ( p,.05) was obtained, whichever occurred first. Responses were considered absent when p..05 and the sidebin noise was #5nanovolts.Amaximum of 800 sweeps (13.65 min) was recorded per intensity. Data were excluded if, after 800 sweeps, results did not meet the sidebin noise criterion of #5 nanovolts. Procedures The study involved one recording session of 1 3 hr for most participants. All infants and young children with normal hearing had tone-abr and multiple-assr testing completed on the same day. Most infants with hearing loss (91%) also had tone ABR and multiple ASSR carried out in the same session. Of the remaining nine infants and young children in this group, all but two had additional testing within three weeks of initial testing, and all nine infants had tympanometry and boneconduction ABR results consistent with SNHL. Participants slept on a bed or in a caregiver s arms while in a double-walled sound-attenuated booth. Infants aged #6 mo were tested under natural sleep whereas children aged.6 mo were tested under conscious sedation (using chloral hydrate). Two infants aged.6 mo were tested in natural sleep due to sedation concerns. For infants and young children with hearing loss, testing typically started with multiple ASSR at the normal intensities (40 50 db HL; Van Maanen and Stapells, 2009) to determine whether thresholds were normal or elevated. When elevated, testing switched to tone-abr testing, as BCEHP protocols require tone ABR be completed prior to other tests such as ASSR (BCEHP, 2008). ASSR testing was continued after tone-abr testing was completed. for tone ABR (hearing loss group) or multiple ASSR (both groups) was found using a 10 db bracketing procedure (except for a db bracket that was used for threshold for 19 of 337 ASSR and 4 of 337 ABR thresholds in the hearing loss group). For the normal group, tone-abr testing did not go below the normal ABR levels (Janssen et al, 2010; BCEHP, 2008). 1 was considered the lowest intensity at which a response was present and a noresponse was obtained at 10 db below that intensity. For both tone-abr and multiple-assr thresholds, intensity was increased or decreased in 10 db steps in the direction that would most likely provide the greatest number of thresholds. The maximum intensity was 100 db nhl for the ABR brief tones and 105 db HL for the multiple-assr tones. DATA AND STATISTICAL ANALYSES s Multiple-ASSR and tone-abr thresholds were found for 53 participants (83 ears) with hearing loss. Ears were considered independent data for infants and children with hearing loss. However, we were not able to obtain both tone-abr and multiple-assr thresholds for both ears and all frequencies for all infants. Tone- ABR thresholds for 1000 Hz were obtained for only three ears; thus, we estimated these thresholds for the other ears by taking the mean of the 500 and 2000 Hz ABR thresholds. 2 s where no response was seen at the maximum intensity were estimated in the following manner: (1) for tone ABR, thresholds were estimated to be 110 db nhl; (2) for multiple ASSR, thresholds were estimated to be 115 db HL. For the normal group from our previous study (Van Maanen and Stapells, 2009), multiple-assr thresholds were determined down to as low as 0 db HL. Tone-ABR thresholds, however, were not formally determined, as testing stopped at the normal levels noted above. As these normal levels are 5 15 db above typical infant ABR thresholds (Stapells, 2000, forthcoming), we therefore estimated ABR threshold for the normal group by subtracting 10 db from the normal level. Due to the similarity of thresholds of normal participants right and left ears, each normal participant contributed only one set of thresholds (ABR and ASSR) per frequency (Van Maanen and Stapells, 2009). Pearson product-moment correlation coefficients and linear regressions were calculated for multiple-assr versus tone-abr threshold for each carrier frequency with all data included, as well as with no-response data excluded. difference scores were also calculated by subtracting tone-abr thresholds (in db nhl) from multiple-assr thresholds (in db HL). Difference scores for all data (i.e., including no-response results), were assessed using a two-way mixed-model analysis of variance (ANOVA), with carrier frequency (500, 538

5 Multiple ASSR in Infants with HL/Van Maanen and Stapells 1000, 2000, and 4000 Hz) as the repeated-measures factor and group (normal and hearing loss) as a betweensubjects factor. Huynh-Feldt epsilon-adjustments for repeated measures were made where appropriate. Newman-Keuls post hoc comparisons were carried out for significant main effects and interactions. For difference scores with no-response results excluded, an ANOVA was not carried out as too few subjects with complete results (i.e., thresholds for all four frequencies) remained. Thus, in order to test for possible group differences, t-tests were carried out at each frequency. The criterion for significance for all statistical tests was p,.05. Normal versus Elevated In our previous study in normal infants (Van Maanen and Stapells, 2009), we proposed normal ASSR levels of 50, 45, 40, and 40 db HL for 500, 1000, 2000, and 4000 Hz, respectively. In the present study, we assessed how well the multiple ASSR differentiates normal from elevated thresholds using the proposed normal levels in infants. ASSR and tone- ABR results (response present or absent) at the normal levels were available for all 98 participants in the hearing loss group and 34 participants with normal hearing. We calculated how often the results were in the correct category for each frequency and across all frequencies combined. RESULTS Figure 1 presents scatterplots comparing multiple- ASSR thresholds to tone-abr thresholds for all participants with thresholds at 500, 1000, 2000, and/or 4000 Hz. When all data are considered, including any noresponse results, correlations between multiple-assr and tone-abr thresholds were.97 for 500, 1000, 2000, and 4000 Hz. These results indicate excellent estimation of tone-abr thresholds by the multiple ASSR. Linear regressions calculated on these data, shown in Table 2, indicate that multiple-assr thresholds (in db HL) are higher than tone-abr thresholds (in db nhl) by 7 (4000 Hz) to 18 (500 Hz) db, with standard errors of the estimate of 8 10 db. The near-unity slopes indicate the relationship between multiple-assr and tone-abr thresholds does not change with degree of hearing loss. Figure 2 shows tone-abr and multiple-assr thresholds for six cases with thresholds that ranged from normal to severe-profound and had differing configurations across frequency. Most ABR and ASSR thresholds were within 15 db of each other, typical of the group results. Figure 1. Multiple-ASSR (in db HL) and tone-abr (in db nhl) thresholds for 500, 1000, 2000, and 4000 Hz. The number of ears for each frequency is shown in parentheses. Points with multiple thresholds have symbols offset (61 db per threshold) to show clearly the overlapping data. s plotted on or beyond the dashed lines ( ) indicate no-response results (110 db nhl for ABR and 115 db HL for ASSR). Diagonals (solid lines) represent perfect ASSR/ABR threshold correspondence and are not regression lines. The exception to this is case 21L (bottom left), which showed larger discrepancies. Table 3 shows the difference scores, calculated for each group (normal and hearing loss) as well as both groups combined. ASSR threshold (in db HL) is typically 6 to 11 db higher than the ABR (in db nhl). Pooled across frequency, there was a small (5 db) but significant difference between the groups [F(1, 63) , p5.017], with the hearing loss group showing smaller difference scores. Pooled across groups, mean difference scores decreased as frequency increased; the small (4 db) difference across frequencies was significant [F(3, 189)57.65, e 50.93, p,.0001]. Post hoc testing indicated the difference score at 4000 Hz was significantly smaller than all other carrier frequencies (p5.019 to.0001). There was no difference between 500, 1000, and 2000 Hz (p 5.14 to.68); there was no frequency by group interaction [F(3, 189)51.78, e 50.93, p5.152]. When all threshold data are considered (i.e., both groups including no-response data), 65% of multiple- ASSR thresholds across all carrier frequencies were within 10 db of tone-abr thresholds (50, 62, 69, and 76% for 500, 1000, 2000, and 4000 Hz, respectively). Similarly, across all carrier frequencies, 90% of multiple- ASSR thresholds were within 20 db of tone-abr 539

6 Journal of the American Academy of Audiology/Volume 21, Number 8, 2010 Table 2. Linear Regression Analyses for All Ears (normal and hearing loss) Frequency (Hz) All Data Y intercept, db nhl Slope SE of estimate (db) Correlation coefficient (r) Number of ears Excluding ASSR or ABR No Response Y intercept, db nhl Slope SE of estimate (db) Correlation coefficient (r) Number of ears X 5ASSR threshold (db HL); Y 5ABR threshold (db nhl) thresholds (88, 83, 94, and 94% for 500, 1000, 2000, and 4000 Hz, respectively). Inclusion of thresholds estimated from no-response results could alter the findings and inflate the correlations. We therefore recalculated the regression analyses and difference scores excluding any thresholds estimated from tone-abr or multiple-assr noresponse results. Correlations decreased but remained strong, at.88,.77,.85, and.89 for 500, 1000, 2000, and 4000 Hz, respectively. Table 1 shows the linear regression results with the no-response results excluded. Multiple-ASSR thresholds are higher than tone ABR and, with the exception of 1000 Hz, slopes remain close to 1 (i.e., 0.9 to 1.1). The poorer correlation and smaller slope for 1000 Hz when the no-response results are excluded is likely due to a restriction-ofrange problem, as there are no thresholds in the severe range. Table 3 presents the difference scores calculated for each group, excluding the no-response data. In general, the same trend for smaller difference scores as frequency increases is evident, although an ANOVA was not carried out as many of the remaining hearing loss subjects did not have thresholds for all four frequencies. To assess difference between groups, we carried out t-tests for each frequency. There was a significantly Figure 2. Multiple-ASSR (in db HL) and tone-abr (in db nhl) thresholds for six individuals with differing degrees and configurations of hearing loss (normal hearing, low-frequency hearing loss, mild-moderate hearing loss, moderate hearing loss, moderate-to-severe sloping hearing loss, and severe-profound hearing loss). s for some normal (response at normal intensity) or severe-profound (no response at maximum intensity) results were estimated and are shown as triangles (see Methods for details). 540

7 Multiple ASSR in Infants with HL/Van Maanen and Stapells Table 3. Difference Scores (db): Multiple-ASSR (in db HL) Minus Tone-ABR (in db nhl) larger difference score for the normal hearing group at 500 Hz (t , df 5 56, p ); however, no other results reached significance (1000 Hz: p 5.240; 2000 Hz: p5.196; 4000 Hz: p5.321). Normal versus Elevated s This study also assessed whether the normal ASSR levels (50, 45, 40, and 40 db HL for 500, 1000, 2000, and 4000 Hz) we previously proposed (Van Maanen and Stapells, 2009) accurately categorized normal versus elevated thresholds. All participants (34 normal and 98 hearing loss) provided data for this analysis. As shown in Table 4, the multiple ASSR categorizes hearing similarlytotoneabrforalmostallparticipants. Acrossallfrequencies, the multiple ASSR erred in only 6% of the 681 comparisons. Of these, only 3% were cases of the multiple ASSR missing a hearing loss; in all these cases, ASSR results at other frequencies were elevated. Table 3 also shows the breakdown of these results for each frequency, with no clear differences seen between frequencies. s Frequency (Hz) All Data Normal hearing ears Mean SD N Hearing loss ears Mean SD N All ears Mean SD N Excluding NR Results Hearing loss ears Mean SD N All ears Mean SD N DISCUSSION We compared tone-abr thresholds, the current gold standard threshold measure for young infants, to multiple- ASSR thresholds in a relatively large group (N 598; 174 ears) of infants and young children, 53 (83 ears) of whom had hearing loss. Correlations between these two measures were very strong (r 5.97) when all data were included, and remained strong (r ) even when the no-response data were excluded. difference scores (ASSR in db HL minus ABR in db nhl) ranged from 6 11 db across all subjects, with standard deviations of 9 10 db. Mean difference scores were larger for the normal-hearing group, although differences between groups were not significant (except at 500 Hz) when no-response results were excluded. Most studies of ASSR thresholds in children with hearing loss that compared ASSR thresholds to frequency-specific gold standard measures (such as behavioral audiometry or tone ABR) have included few, if any, infants or young children (i.e.,,7 yr of age) in their study samples. Many studies have compared frequency-specific ASSR to click-abr thresholds in infants and young children, but click-abr thresholds are not frequency specific. Two previous studies have compared ASSR threshold to tone-abr thresholds in young children (Vander Werff et al, 2002; Stueve and O Rourke, 2003); unfortunately these studies assessed the tone ABR only to low-frequency stimuli (250 and 500 Hz), and both had technical issues (for review, see Stapells et al, 2005). There are three previous studies that have compared ASSR and behavioral thresholds in infants and young children. The research by Rance and colleagues (Rance et al, 2005) used the single-stimulus ASSR and has provided the largest sample size (285 with normal hearing and 271 with SNHL). The other two studies (Han et al, 2006; Luts et al, 2006) used the multiple-assr technique and have a total sample size of about 70 ears. The present study, comparing multiple-assr to tone-abr thresholds, nearly doubles the multiple-assr sample size in the literature. A unique aspect of the present study is that it provides tone-abr thresholds that were obtained on the same day as the multiple-assr thresholds for over 93% of participants. Other studies in infants and young children, which compared ASSR thresholds to behavioral thresholds, typically obtained the behavioral thresholds much later at a follow-up visit. During the intervening time, there will be changes in ear-canal acoustics, behavioral maturation, and/or possible progression in hearing loss, any of which will alter the ASSR versus behavioral threshold relationship (Stapells et al, 1995; Luts et al, 2006). These are not concerns for the current study, where ASSR and ABR were obtained in the same session. The very strong correlations (.97) between multiple- ASSR and tone-abr thresholds are similar to those presented by Rance and colleagues (2005), who reported 541

8 Journal of the American Academy of Audiology/Volume 21, Number 8, 2010 Table 4. Accuracy of Normal Multiple-ASSR Levels for Classification of Normal versus Elevated ALL 500 Hz 1000 Hz 2000 Hz 4000 Hz Tone-ABR Tone-ABR Tone-ABR Tone-ABR Tone-ABR Normal Elevated Normal Elevated Normal Elevated Normal Elevated Normal Elevated ASSR Normal Elevated 136 (20%) 22 (3%) 22 (3%) 501 (74%) ASSR Normal Elevated 34 (21%) 2 (1%) 4 (3%) 122 (75%) ASSR Normal Elevated 32 (21%) 6 (4%) 7 (4%) 112 (71%) ASSR Normal Elevated 37 (19%) 10 (5%) 1 (1%) 142 (75%) ASSR Normal Elevated 33 (19%) 4 (2%) 10 (5%) 125 (74%) Notes: Percent of total in parentheses. ASSR normal levels: 50, 45, 40, and 40 db HL for 500, 1000, 2000, and 4000 Hz (Van Maanen and Stapells, 2009). Tone-ABR normal levels: 35, 35, 30, and 25 db nhl for 500, 1000, 2000, and 4000 Hz (BCEHP, 2008) correlations of r between single-stimulus ASSR and behavioral thresholds. The studies by Han et al (2006) and Luts et al (2006) also report strong correlations, albeit slightly lower (r 5.8.9) than the present study. Rance and colleagues regression formulae intercepts and slopes differ from the current study. Their intercepts are larger, likely due to their use of much shorter averaging times (Picton et al, 2005) but also possibly because their intercepts represent a different prediction (the ASSR threshold for 0 db HL behavioral threshold) than the current study (the tone-abr threshold for 0 db HL ASSR threshold). The more-elevated thresholds in normal-hearing infants in the studies by Rance and colleagues also affect the slopes of their regression formulae, which are not near unity, in contrast to those of the current study (as well as the studies by Han et al [2006] and by Luts et al [2006]). As a result of their nonunity slopes, Rance and colleagues recommended the use of regressions to estimate behavioral thresholds, rather than simple correction factors. In the current study, regression slopes are close to unity (Table 1), thus ASSR-ABR correction factors (based on difference scores, Table 2) are essentially the same for any hearing level estimated by tone ABR. Difference scores in the current study ranged from 10.7 db (500 Hz) to 6.3 db (4000 Hz) for all data, and 12.9 db (500 Hz) to 7.3 db (4000 Hz) when no-response results were excluded. Standard deviations of the difference scores were 9 10 db for all data, similar to those of other ASSR studies in infants (e.g., Rance and Briggs, 2002; Han et al, 2006; Luts et al, 2006), as well as tone- ABR studies (e.g., Stapells, 2000; Vander Werff et al, 2009). For almost all cases (95%), the ASSR threshold (in db HL) was higher than the tone-abr threshold (in db nhl). The finding of larger difference scores for 500 Hz is similar to previous ASSR research in children (Lins et al, 1996; Cone-Wesson et al, 2002; Swanepoel and Steyn, 2005; Han et al, 2006) and adults (for review: Tlumak et al, 2007; Vander Werff et al, 2008), as well as tone-abr studies in children (Stapells et al, 1995; Stapells, 2000, Lee et al, 2008; Vander Werff et al, 2009). John et al (2004) hypothesized that the poorer 500 Hz ASSR thresholds in young infants may be due to (1) poor neural synchronization and/or to (2) less-effective low-frequency transmission due to the presence of vernix, middle-ear pressure variations, or to maturational changes of the outer and middle ear. In the present study, infants were beyond the age where vernix would be a factor. Ear-canal changes or middle-ear variations are also ruled out as both tests (tone ABR andmultipleassr)werecarriedoutonthesameday for over 93% of infants. It is important to note, however, that our difference scores reflect multiple-assr versus tone-abr threshold differences. Thus, the larger difference scores at 500 Hz suggest an issue specific to the ASSR; otherwise, we would expect both ABR and ASSR thresholds to be equally elevated. One possibility is that youngauditorysystemshavereducedabilitiestoprocess the very rapid modulation rates ( Hz) used for the brainstem ASSR whereas the relatively slower rates (30 50/sec) for the tone ABR pose no problem (Burkard et al, 1990; Lasky, 1991; Rance, 2008). Based on our results, this problem seems greater for lower carrier frequencies, although Rance and colleagues, studying much younger infants, showed this for both 500 and 4000 Hz (Rance et al, 2006). However, larger difference scores at 500 Hz are also seen in adult ASSRs (for reviews, see Tlumak et al, 2007; Vander Werff et al, 2008), which mayindicateageneralissuewithrapidratesat500hz and not reflect maturational factors. Our difference scores for all data were significantly better (smaller) for 4000 Hz than all other carrier frequencies. Previous tone-abr and ASSR studies also indicate better thresholds and smaller difference scores for 4000 Hz compared to lower frequencies (for reviews, see Stapells, 2000; Tlumak et al, 2007; Stapells, forthcoming). Although ear-canal factors might play a role (e.g., Sininger et al, 1997; Bagatto et al, 2006; Rance and Tomlin, 2006), neither they nor transducer variations can fully account for the better results at 4000 Hz. In the current study, multiple-assr and tone-abr thresholds were obtained using the same transducers 542

9 Multiple ASSR in Infants with HL/Van Maanen and Stapells and on the same day. It is possible that a trade-off between ear-canal SPL and neural synchrony occurs such that, in young infants, better 4000 Hz thresholds due to higher ear-canal SPL are partly offset by poorer neural synchrony, whereas in the older infants the ear-canal SPL decrease is offset by better neural synchrony. Further research is required. Difference scores in the hearing loss group were significantly smaller than the normal-hearing group, albeit only by 4 db. This is partially explained by the no-response results decreasing the difference scores in the hearing loss group. When no-response results were excluded, the larger difference scores were only seen at 500 Hz. It is possible that the difference scores would have been even larger in the normal-hearing group, especially for 500 Hz, had we obtained true tone-abr threshold rather than stopping at the normal levels of 35, 30, and 25 db nhl at 500, 2000, and 4000 Hz. As discussed above, this may be related to neural synchrony issues with ASSR, especially at 500 Hz. Previous ASSR studies in infants compared ASSR thresholds to behavioral thresholds; thus, their difference scores are not directly comparable to those of the present study, which reflect ASSR versus tone-abr threshold differences. In order to compare our results to previous studies, the tone-abr thresholds must be corrected to estimated behavioral thresholds. Many studies have investigated tone-abr and behavioral thresholds in infants and young children with hearing loss (for review, see Stapells, forthcoming); meta-analysis of most of these studies shows mean ABR-behavioral difference scores for infants with SNHL of 16, 15, 11, and 28 db for 500, 1000, 2000, and 4000 Hz, respectively (Stapells, 2000). Based on the ABR meta-analysis difference scores, the ASSR-minus-behavioral difference scores for the participants with hearing loss in the present study would be estimated to be 114, 113, 19, and 23dB for 500, 1000, 2000, and 4000 Hz, respectively. These corrected difference scores are similar to those of Han et al (2006) but are larger than those of Luts et al (2006) and Rance and Briggs (2002). Possible reasons for differences between the present study and others may lie in the fact that the current compared ASSR thresholds to tone-abr thresholds, and these thresholds were obtained on the same day as the ASSR. In contrast, other studies compared ASSR thresholds to behavioral thresholds obtained on a later date (usually several months later). Thus, in addition to differences related to ABR versus behavioral methods, changes in thresholds may have occurred for the other studies due to such factors as ear-canal maturation or progression of hearing loss. Although ASSR thresholds in db HL are higher than tone-abr thresholds (in db nhl), when converted to db ppe SPL, ASSR thresholds are lower than tone- ABR thresholds, on average, by 7, 16, 8, and 14 db at 500, 1000, 2000, and 4000 Hz, respectively. 3 Possible explanations for this may involve response characteristics or methodology, or combinations of both. For example, the ASSR may be inherently more sensitive than ABR wave V. Alternatively, methodological issues such as (1) stimulus differences (longer stimuli for ASSR) and/ or (2) response detection method (visual observation for ABR versus FFT analyses for ASSR) may favor the ASSR. It also possible that differences in recording methodology and/or time resulted in lower residual EEG noise and thus better thresholds for the ASSR than the tone ABR. Differences across frequencies are more complicated. Future research would need to control noise issues and stimulus differences across methods in order to better assess and understand the ASSR-ABR differences. Although differences in SPL are interesting theoretically, it is the ASSR/tone-ABR differences in HL/nHL that are of greatest interest clinically. Normal versus Elevated s We previously reported preliminary results for 23 children that indicated that the normal levels for multiple ASSR would not pass a child with hearing loss (Van Maanen and Stapells, 2009). The present study confirms this for a larger sample size; in almost all cases, the normal multiple-assr levels of 50, 45, 40, and 40 db HL correctly classified children as normal or elevated at each frequency. The multiple ASSR missed hearing loss in only 3% of all comparisons (22/681), or 4% (22/523) of those with elevated tone-abr thresholds. These came from 13 infants, none of whom would have passed the multiple ASSR at every frequency; thus these multiple-assr normal levels miss very few hearing losses. Although the multiple ASSR also accurately classified most of the comparisons with normal tone-abr thresholds, it did indicate as being elevated about 14% of these comparisons. This contrasts with our earlier study where at least 95% of infants with normal hearing showed multiple ASSRs at the normal levels (Van Maanen and Stapells, 2009). However, in the present study, most of the infants (12/18) where multiple ASSR erroneously suggested an elevated threshold actually had hearing loss at other frequencies; thus, these infants hearing was not truly normal. It is possible, therefore, that the multiple ASSR, with its multiple simultaneous stimuli, may be more sensitive (than the ABR) to dysfunction even when thresholds are normal. CONCLUSIONS As noted above, many studies of threshold estimation using the ASSR exist, but only a few studied infants and young children and compared ASSR thresholds to frequency-specific gold standard 543

10 Journal of the American Academy of Audiology/Volume 21, Number 8, 2010 measures. The results of four studies (Rance et al, 2005; Han et al, 2006; Luts et al, 2006; present study) that satisfy this criterion all show that the ASSR provides a reasonably accurate estimate of hearing threshold. Each of these four studies used different stimulus and analysis techniques as well as different gold standard measures of hearing for comparison. Thus, there are relatively few infant subjects with hearing loss with each technique, and it remains unclear which methodology is best. Further research is required. ASSR can be recommended as a stand-alone measure only if it can be used confidently to accurately estimate hearing thresholds in very young children with a variety of hearing losses. One also must be able to differentiate conductive from sensorineural or mixed hearing loss, for which bone-conduction data are needed. Although Small and Stapells (2006, 2008) have published normative boneconduction ASSR data infants, there are few boneconduction ASSR data in young children with hearing loss (Swanepoel et al, 2008) and none with frequencyspecific confirmation of their hearing thresholds. Thus, further air- and bone-conduction ASSR studies are needed in infants and young children with hearing loss. Based on the above, it is premature to rely on the ASSR as the primary electrophysiological measure of threshold in infants, except possibly as the first step in a diagnostic protocol to quickly differentiate normal from elevated thresholds. As the present study s results show, the normal multiple-assr levels of 50, 45, 40, and 40 db HL correctly classified children as having normal or elevated thresholds. Due to the lack of air- and bone-conduction data in infants with hearing loss, we recommend switching to tone ABR when thresholds are elevated. At this point, only the tone ABR has the sufficient research, clinical database, and clinical history to be the primary method to diagnose type, degree, and configuration of hearing loss in infants. ASSR thresholds provide an excellent cross-check of the air-conduction tone-abr thresholds. A cross-check of elevated tone-abr thresholds could be carried out by comparing these to ASSR thresholds, after applying a correction factor such as the ASSRminus-ABR difference scores (Table 2). For example, if the mean difference scores of 11, 9, 9, and 6 db are subtracted from the current study s ASSR thresholds (in HL), across frequencies 84% of these corrected ASSR thresholds are within 10 db of the actual tone-abr thresholds. Large ASSR/tone-ABR differences in an individual patient might alert a clinician that a problem exists with the test results. However, it is possible that somewhat larger corrections would be preferable because (1) the above difference scores will overestimate thresholds in many individuals with normal hearing, and (2) there may be concerns about overamplification in infants. To estimate behavioral threshold, an additional correction must be applied. Future research should determine optimal corrections on a new group of infants and young children having various types and degrees of hearing loss. Acknowledgments. We thank Laurie Usher and Renée Janssen for their assistance in and support of this research. NOTES 1. Eight infants in the normal group were tested at lower intensities, showing present ABRs to brief tones at 30 db nhl at 500 Hz, 20 db nhl at 2000 Hz, and 20 db nhl at 4000 Hz for both ears. 2. BCEHP protocols only require thresholds for 1000 Hz when a steeply sloping hearing is present; for more information see BCEHP protocols (BCEHP, 2008). 3. This conversion to SPL was carried out only for participants with hearing loss; no-response and normal results were excluded as these were not actual thresholds. REFERENCES American National Standards Institute (ANSI). (1996) American National Standard Specifications for Audiometers. Vol. S New York: American National Standards Institute. American Speech-Language-Hearing Association (ASHA). (2004) Guidelines for the audiologic assessment of children from birth to 5 years of age. Bagatto MP, Seewald RC, Scollie SD, Tharpe AM. (2006) Evaluation of a probe-tube insertion technique for measuring the realear-to-coupler difference (RECD) in young infants. J Am Acad Audiol 17: British Columbia Early Hearing Program (BCEHP). (2008) Diagnostic Audiology Protocol. Burkard R, Shi Y, Hecox KE. (1990) Brain-stem auditory-evoked responses elicited by maximum length sequences: effect on simultaneous masking noise. J Acoust Soc Am 87: Cone B, Dimitrijevic A. (2009) In: The auditory steady-state response.katz J, Medwetsky L, Burkard R, Hood L, eds. Handbook of Clinical Audiology. 6th ed. Baltimore: Lippincott, Williams and Wilkins, Cone-Wesson B, Dowell RC, Tomlin D, Rance G, Ming WJ. (2002) The auditory steady-state response: comparisons with the auditory brainstem response. J Am Acad Audiol 13: Eggermont JJ. (1982) The inadequacy of click-evoked auditory brainstem responses in audiological applications. Ann N Y Acad Sci 388: Han D, Mo L, Lui H, Chen J, Huang L. (2006) estimation in children using auditory steady-state responses to multiple simultaneous stimuli. ORL J Otorhinolaryngol Relat Spec 68: Herdman AT, Picton TW, Stapells DR. (2002) Place specificity of multiple auditory steady-state responses. J Acoust Soc Am 112: Hyde M. (2005) In: Evidence-based practice, ethics and EDHI program quality.seewald RC, Bamford J, eds. A Sound Foundation through Early Amplification Staefa: Phonak AG, Janssen RM, Usher L, Stapells DR. (2010) The British Columbia s Children s Hospital tone-evoked auditory brainstem response 544

11 Multiple ASSR in Infants with HL/Van Maanen and Stapells protocol: how long do infants sleep, and how much information can be obtained in one appointment? Ear Hear 31: John MS, Brown DK, Muir PJ, Picton TW. (2004) Recording auditory steady-state responses in young infants. Ear Hear 25: John MS, Dimitrijevic A, Picton TW. (2002) Auditory steady-state responses to exponential modulation envelopes. Ear Hear 23: Joint Committee on Infant Hearing. (2007) Year 2007 position statement: principles and guidelines for early hearing detection and intervention programs. Pediatrics 120: Kennedy CR, McCann DC, Campbell MJ, et al. (2006) Language ability after early detection of permanent childhood hearing impairment. N Engl J Med 354: Lasky RE. (1991) The effects of rate and forward masking on human adult and newborn auditory evoked brainstem response thresholds. Dev Psychobiol 24: Lee HS, Ahn JH, Chung JW, Yoon TH, Lee KS. (2008) Clinical comparison of the auditory steady-state response with the click auditory brainstem response in infants. Clin Exp Otorhinolaryngol 1: Lins OG, Picton TW, Boucher BL, et al. (1996) Frequency-specific audiometry using steady-state responses. Ear Hear 17: Luts H, Desloovere C, Wouters J. (2006) Clinical application of dichotic multiple-stimulus auditory steady-state responses in high risk newborns and young children. Audiol Neurootol 11: Picton TW, Dimitrijevic A, Perez-Abalo MC, Van Roon P. (2005) Estimating audiometric thresholds using auditory steady-state responses. J Am Acad Audiol 16: Picton TW, John MS, Dimitrijevic A, Purcell D. (2003) Human auditory steady-state responses. Int J Audiol 42: Rance G. (2008) In: Auditory steady-state responses in neonates and infants.rance G, ed. The Auditory Steady-State Response. Generation, Recording, and Clinical Application. San Diego: Plural Publishing, Rance G, Briggs RJ. (2002) Assessment of hearing in infants with moderate to profound impairment: the Melbourne experience with auditory steady-state evoked potential testing. Ann Otol Rhinol Laryngol Suppl 189: Rance G, Roper R, Symons L, et al. (2005) Hearing threshold estimation in infants using auditory steady-state responses. JAm Acad Audiol 16: Rance G, Tomlin D. (2006) Maturation of auditory steady-state responses in normal babies. Ear Hear 27: Rance G, Tomlin D, Rickards FW. (2006) Comparison of auditory steady-state responses and tone-burst auditory brainstem responses in normal babies. Ear Hear 27: Sininger YS, Abdala C, Cone-Wesson B. (1997) Auditory threshold sensitivity of the human neonate as measured by the auditory brainstem response. Hear Res 104: Sininger YS, Hyde ML. (2009) In: Auditory brainstem response in audiometric threshold prediction.katz J, Medwetsky L, Burkard R, Hood L, eds. Handbook of Clinical Audiology. 6th edition. Baltimore: Lippincott, Williams and Wilkins, Small SA, Stapells DR. (2006) Multiple auditory steady-state response thresholds to bone-conduction stimuli in young infants with normal hearing. Ear Hear 27: Small SA, Stapells DR. (2008) Normal ipsilateral/contralateral asymmetries in infant multiple auditory steady-state responses to air- and bone-conduction stimuli. Ear Hear 29: Stapells DR. (2000) estimation by the tone-evoked auditory brainstem response: a literature meta-analysis. J Speech Lang Pathol Audiol 224: Stapells DR. (Forthcoming) Frequency-specific threshold assessment in young infants using the transient ABR and the brainstem ASSR. In: Seewald RC, Tharpe AM, eds. Comprehensive Handbook of Pediatric Audiology. San Diego: Plural Publishing. Stapells DR, Gravel JS, Martin BA. (1995) s for auditory brain stem responses to tones in notched noise from infants and young children with normal hearing or sensorineural hearing loss. Ear Hear 16: Stapells DR, Herdman A, Small SA, Dimitrijevic A, Hatton J. (2005) In: Current status of the auditory steady-state responses for estimating an infant s audiogram.seewald RC, Bamford J, eds. A Sound Foundation Through Early Amplification Staefa: Phonak AG, Stapells DR, Oates P. (1997) Estimation of the pure-tone audiogram by the auditory brainstem response: a review. Audiol Neurootol 2: Stueve MP, O Rourke C. (2003) Estimation of hearing loss in children: comparison of auditory steady-state response, auditory brainstem response, and behavioral test methods. Am J Audiol 12: Swanepoel d.w, Ebrahim S, Friedland P, Swanepoel A, Pottas L. (2008) Auditory steady-state responses to bone conduction stimuli in children with hearing loss. Int J Pediatr Otorhinolaryngol 72: Swanepoel d.w, Steyn K. (2005) Short report: establishing normal hearing for infants with the auditory steady-state response. S Afr J Commun Disord 52: Tlumak A, Rubinstein E, Durrant JD. (2007) Meta-analysis of variables that affect accuracy of threshold estimation via measurement of the auditory steady-state response. Int J Audiol 46: Vander Werff KR, Brown CJ, Gienapp BA, Schmidt Clay KM. (2002) Comparisons of auditory steady-state response and auditory brainstem response thresholds in children. J Am Acad Audiol 13: Vander Werff KR, Johnson T, Brown C. (2008) In: Behavioral threshold estimation for auditory steady-state response. Rance G, ed. The Auditory Steady-State Response. Generation, Recording, and Clinical Application. San Diego: Plural Publishing, Vander Werff KR, Prieve BA, Georganatas LM. (2009) Infant air and bone conduction tone burst auditory brainstem responses for classification of hearing loss and the relationship to behavioral thresholds. Ear Hear 30: Van Maanen A, Stapells DR. (2009) Normal multiple auditory steady-state response thresholds to air-conducted stimuli. JAm Acad Audiol 20: Yoshinaga-Itano C, Gravel JS. (2001) Evidence for universal newborn hearing screening. Am J Audiol 10: Yoshinaga-Itano C, Sedley AL, Coulter DK, Mehl AL. (1998) Language or early- and later-identified children with hearing loss. Pediatrics 102: