Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigating longitudinal changes in real-ear to coupler difference measurements in infants Bingham, Kristina 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-0329.pdf [ 3.18MB ]
Metadata
JSON: 831-1.0101116.json
JSON-LD: 831-1.0101116-ld.json
RDF/XML (Pretty): 831-1.0101116-rdf.xml
RDF/JSON: 831-1.0101116-rdf.json
Turtle: 831-1.0101116-turtle.txt
N-Triples: 831-1.0101116-rdf-ntriples.txt
Original Record: 831-1.0101116-source.json
Full Text
831-1.0101116-fulltext.txt
Citation
831-1.0101116.ris

Full Text

INVESTIGATING LONGITUDINAL CHANGES IN REAL-EAR TO COUPLER DIFFERENCE MEASUREMENTS IN INFANTS by KRISTINA BINGHAM THESIS SUBMITTED IN PARTIAL FUFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in FACULTY OF GRADUATE STUDIES (Audiology and'Speech Sciences) THE UNIVERSITY OF BRITISH COLUMBIA August 2007 © Kristina Bingham 2007 11 Abstract The real-ear-to-coupler difference (RECD) measurement is a commonly-used clinical procedure that quantifies the difference in sound-pressure level between a 2-cc coupler and an individual's ear canal. The SPL levels in infant ears are highly variable and significantly higher than the SPL levels present in average adult ears, which makes the quantification of SPL levels in infant ears extremely important for threshold determination and the fitting of amplification. The purpose of this study was to examine longitudinal changes in RECD values in newborn infants, over a 1 -month interval, to determine whether a significant decrease in RECD values takes place, and whether that decrease is predictable from either the infant's corrected age or their initial RECD values. Using standard (226Hz) and high frequency (1000 Hz) tympanometry, measures of static admittance and ear canal volume were also evaluated to determine their predictive ability regarding frequency specific RECD values. Measurement error associated with the RECD was quantified and then compared to the magnitude of RECD changes over a 1 -month interval. A multi-variate analysis of variance (MANOVA) was performed and revealed an overall significant change in RECD values over a 1 month interval. Forward regression analysis further revealed corrected age as a predictor of RECD values at 3 frequencies and initial RECD values as a predictor at 1 frequency. Static admittance was a larger predictor of RECD values when compared to E C V , with the exception of the lowest and highest frequencies (0.25 and 6 kHz). Magnitude of the test-retest variability associated with the RECD measurement was not large but variability was large enough to obscure longitudinal RECD changes over a 1-month period. Consequently, it may be clinically unnecessary to re-measure an RECD within an infant's first month of life, to account for changes in ear canal acoustics between initial screening tests and follow-up assessments. The best measures of static admittance and ear canal volume in infants remains un-determined and a strong need exists for further longitudinal data in infant populations. iv Table of Contents Abstract i i Table of Contents iv List of Tables vi List of Figures vii Acknowledgements ix Dedication x Introduction 1 Utility of the R E C D Measurement 3 Sources of Variability in Fitting Amplification for Infants 7 Threshold Data and Prescriptive Targets 7 R E C D Measurement 9 Maturation of the External and Middle Ear 15 Maturation of the External Ear 15 Maturation of the Middle Ear 15 Measures of External and Middle Ear: Infant Tympanometry 17 Measures of External and Middle Ear: Static Admittance 19 Compensation for Ear Canal Volume 22 Middle Ear Pathologies 23 Average versus Individually Measured RECDs 25 Clinical Implications 28 Methods 29 Recruitment 30 Subjects 30 V Procedure 31 TEOAES 31 Tympanometry 31 RECD Measurement 32 Wide Band Reflectance 33 Results 33 Treatment of the Data 33 RECD Values at Visit 1 and Visit 2 34 RECD Predictors 36 Test-retest Reliability 49 Discussion 53 Significant Changes in RECD Values 53 Corrected Age and RECD values at Visit 1 55 RECD Predictors 57 Test-retest Reliability 60 Clinical Implications 61 Directions for Future Research 62 Conclusion 63 References .- 65 Appendix A • 74 Appendix B 76 Appendix C 78 Appendix D 79 Appendix E 80 vi List of Tables Table 1 Significant Change in RECD values between Visit 1 and Visit 2, across Frequency 35 Table 2 Corrected Age and RECD at Visit 1 as Significant Predictors by Frequency 36 Table 3 Pearson Correlations between RECD Predictors 37 Table 4 Variable Sets 1, 2 and 3 for determining best data transformations 38 Table 5 Significant SA and E C V Predictors by Frequency 40 Table 6 Pearson Correlations between Test 1 & Test 2 across Frequency at both Visits 50 Table 7 Test-retest Reliability at each Visit 51 Table 8 Average Test-retest Variability for each Visit and Combined Visits 52 Table 9 Visit 1, Visit 2 Changes Compared to Upper Confidence Interval for Test-retest Variability 53 vii List of Figures Figure 1 Polar and Rectangular Notation 21 Figure 2 250 Hz Correlation between RECD values and E C V 1000 Hz B+ 43 Figure 3 250 Hz Correlation between RECD values and E C V at 226 Hz B+ 43 Figure 4 500 Hz Correlation between RECD values and SA 1000 Hz +200 Ya 44 Figure 5 500 Hz Correlation between RECD values and E C V at 1000 Hz B+ 44 Figure 6 750 Hz Correlation between RECD values and SA 1000 Hz +200 Ya 45 Figure 7 750 Hz Correlation between RECD values and E C V at 1000 Hz B+ 45 Figure 8 1000 Hz Correlation between RECD values and SA 1000 Hz +200 Y a 46 Figure 9 1000 Hz Correlation between RECD values and E C V at 1000 Hz B+ 46 Figure 10 1500 Hz Correlation between RECD values and SA 1000 Hz +200 Ya 47 Figure 11 2000 Hz Correlation between RECD values and E C V at 1000 Hz B+ 47 viii Figure 12 4000 Hz Correlation between RECD values and SA at 1000 Hz +200 Ytm 48 Figure 13 6000 Hz Correlation between RECD values and E C V at 1000 Hz B+ 48 Figure 14 6000 Hz Correlation between RECD values and E C V at 226 Hz B+ 49 ix Acknowledgements Thank-you to The British Columbia Early Hearing Program for lending us their equipment and providing funding for this research. Thank-you also to the University of British Columbia Faculty of Graduate Studies, for providing funding for this research. I would also like to thank my advisor, Lorienne Jenstad, for her time, knowledge, and support. Thank-you also to my committee members Navid Shahnaz and Dreena Davies. Dedication " For O p a ! 1 Introduction With the implementation of universal newborn hearing screening, the identification of infant hearing loss is occurring earlier than ever before. The first few years of life are a critical period for the development of speech and language, and infants who receive early identification and intervention demonstrate significantly better vocabulary, general language abilities, speech intelligibility and social-emotional development than infants who are identified at a later age (Yoshinaga-Itano, 2003). The goal of Early Hearing Programs, including the British Columbia Early Hearing Program (BCEHP), is to screen all babies by 1 month of age, identify hearing loss by 3 months of age and ensure enrolment in appropriate intervention by 6 months (JCIH, 2000; BCEHP, 2006). Meeting these deadlines presents particular challenges for audiologists who must obtain accurate and timely results, when maturational changes are occurring rapidly and therefore affecting the accuracy of measured thresholds. Infants who fail the initial screening at birth will be referred for diagnostic assessment. The main challenge with infants is that threshold data are often incomplete and limited to a few frequencies (Seewald & Scollie, 1999). If a hearing loss is diagnosed, appropriate intervention usually includes the timely fitting of hearing aids that will provide an amplified speech signal that is both audible and comfortable. Amplification, usually behind the ear hearing aids (BTEs), is then selected based on output and gain targets which are calculated from the infant's audiological data. Due to the fact that infants are unable to communicate their impressions regarding the appropriateness of their amplification, audiologists also need to be diligent in obtaining objective measures for the verification of real-ear hearing aid performance. 2 Unfortunately a fair amount of variability is introduced as an infant moves through the assessment and fitting process. At the assessment stage, infant thresholds are usually measured with insert earphones; however, at the time of intervention, hearing aid fittings are done with the infant's personal earmold, resulting in two different coupling methods. In addition, the shape and size of the infant ear is changing rapidly and is reflected by changes in ear canal volume and impedance (Keefe, Bulen, Hoberg & Burns, 1993; Shanks, Stelmachowicz, Beauchaine, & Shulte, 1992). The changes in ear canal volume and impedance will impact the sound pressure level (SPL) present in the infant's ear and consequently, as the infant matures and his or her ear canal acoustics change, their apparent hearing level (dB HL) thresholds will also change. One way to account for some of the variability in real ear SPL levels is through the use of a real-ear to coupler difference (RECD) measure. The RECD measurement provides a way to monitor these SPL changes by comparing the SPL measured in a 2-cc coupler with the SPL measured in an individual's ear (Bagatto, Moodie, Scollie, Seewald, Moodie, Pumford & Liu, 2005). Due to the fact that infant ears differ markedly from the average adult ear, the use of individually measured RECD values is particularly important, especially when fitting hearing aids for young children and infants. It has been proposed that due to a shorter ear canal (Feigin, Kopun, Stelmachowicz, & Gorga, 1989), smaller ear canal volume (Dillon, 2001; Jirsa & Norris, 1978), and higher impedance (Keefe et al., 1993), the SPL levels present in infant ears are greater than those in adult ears. The presence of higher SPL levels in the infant ear requires measurement so that appropriate amplification can be selected and fit. 3 The changing ear canal volume and impedance in infant ears introduce variability into the RECD measurement. Variability in the RECD measure also exists due to acoustic leakage (Sinclair, Beauchaine, Moodie, Feigin, Seewald & Stelmachowicz, 1996) and the method and depth of probe tube placement (Bagatto, Scollie, Seewald, Moodie, & Hoover, 2002). The RECD has also demonstrated significant variance among individuals the same age (Bagatto et al.; 2002; Feigin et al., 1989), which suggests that a more accurate measure can be obtained by measuring an RECD for the individual as opposed to relying on an average, age-appropriate value. Overall, when care is taken in the measurements, reliable RECD values can be obtained for infants (Bagatto, Moodie, Scollie, Seewald, Moodie, Pumford & Liu, 2005; Moodie, Seewald & Sinclair, 1994; Sinclair et al., 1996). Utility of the RECD Measurement In general, RECD values are greater than or equal to zero (Pumford & Sinclair, 2001). A positive RECD value indicates by how much the levels measured in the real ear exceed the levels that are measured in a coupler (Bagatto, 2001). This measure is useful because once the difference between the 2-cc coupler and the real ear is known, this difference can be added to the coupler response to predict levels in the individual's ear. The use of RECD measurements is also important in order to convert hearing level thresholds (dB HL) to sound pressure level or dB SPL, which also helps to ensure an appropriate hearing aid can be selected and fit (Bagatto, 2001). Quite often the audiometric data an audiologist collects when testing a child may be from a variety of transducers. Data may be collected in dB H L using a supra-aural earphone, an insert earphone, or a sound-field loudspeaker. As Seewald (1991) points out, all of these transducers are calibrated using different instrumentation and different SPL reference 4 values. In theory, all dB H L thresholds should be the same regardless of the transducer used, provided that the transducer is calibrated properly and the acoustic properties of individual's ear are representative of an average adult (Seewald & Scollie, 1999). Unfortunately the acoustic properties of infant ears differ markedly from those of adults. Voss and Herrmann (2005) compared the sound pressure generated in an infant's ear compared with an adult's ear, for three different earphones; When sound pressure was measured at the earphone, insert earphones were found to produce pressures of up to 15 dB higher in infant ears compared with adults. However, circumaural and supra-aural earphones produced pressures that differed by less then 3 dB between infant and adult ears. When the same measurements were made at the tympanic membrane, insert earphones were found to produce pressures that were 5 to 8 dB higher in infants across all frequencies (Voss & Herrmann, 2005). Seewald and Scollie (1999) measured the thresholds of a 9-month-old infant and an adult, using TDH-series earphones, insert earphones, and a sound-field loudspeaker. They found that predicted threshold values for the 9 month old varied as a function of the signal transducer. They also predicted that the signal transducer effects would be larger in younger infants (Seewald & Scollie, 1999). Consequently in an infant with hearing loss, thresholds will appear to differ and the amplification characteristics that are selected will depend on which type of transducer is used (Seewald & Scollie, 1999). In the case of the 9-month-old infant, three different sets of amplification characteristics would result for the same infant (Seewald & Scollie, 1999). Unfortunately this can lead to an inappropriate hearing aid fitting. One way to avoid the problems associated with dB H L measures is to define threshold in dB SPL in the ear canal (Seewald & Scollie, 1999). The SPL at the eardrum 5 that is needed to hear a test signal is the same regardless of which transducer is used (Seewald & Scollie, 1999). Acoustic transforms, including the RECD, can be used to define thresholds in dB SPL at the eardrum, therefore removing a major source of variability in prescribing amplification characteristics. Due to the fact that the RECD is most feasible for use with infants, it has been recommended that insert earphones are used whenever possible when testing infants because they are calibrated in a 2-cc coupler (Seewald & Scollie, 1999). Consequently the corresponding coupler SPL for any dB H L setting on the audiometer is known (Seewald & Scollie, 1999). Because the Desired Sensation Level (DSL) fitting method also recommends using insert earphones to measure infant's RECD values, the RECD can be added to the coupler SPL to get a predicted dB SPL at the infant's ear canal (Seewald & Scollie, 1999). The advantage to using the RECD this way is that the acoustic transform used to predict the ear canal SPL at threshold is individualized for that particular infant and is not based on adult data (Seewald & Scollie, 1999). The R E C D is actively promoted for clinical use with infants and young children because it is the ear canal measurement that is most practical for this population for several reasons (Bagatto et al., 2005). First, the measurement of an RECD does not require the use of a sound field loudspeaker; therefore, a specific position does not need to be maintained by the child during testing and the variability associated with sound field probe tube microphone measurements can be eliminated (Moodie et al., 1994). The real-ear to dial difference (REDD) does not require the use of a soundfield loudspeaker either, however, the REDD measures the difference in real ear SPL between TDH-series headphones and the audiometer dial reading (Bagatto et a l , 2005). TDH-series headphones are heavier 6 compared with insert earphones and are often more intimidating and bulky for use with younger populations (Bagatto et al., 2005). Second, normative RECD data is described in more detail than real-ear unaided gain (REUG) data, and normative REDD data is not available (Bagatto et al., 2005). Third, the RECD can also be used later in the hearing aid fitting process with a coupler-based fitting approach that allows audiologists to perform electro-acoustic response shaping in the test box, rather than in the infant's or child's real ear. This ensures a controlled acoustic environment and reduces the amount of time and cooperation required from the infant or child (Moodie et al., 1994). Specifically, in the coupler-based approach, the RECD measurement is used to determine appropriate output limiting values to achieve suitable targets for fitting hearing aids. As a result of the higher SPL levels present in infant ears, many audiologists limit the output (saturation) sound pressure level (OSPL-90) in infant's hearing aids in order to avoid the possibility of over-amplification caused by the SPL levels being too high (Feigin et al., 1989). However, this cautious approach may lead to setting the maximum output of a child's hearing aid too low, which then results in less then optimal amplification that places conversational speech at a reduced sensation level (Feigin et al., 1989). A difference of only a few decibels in the output of a child's hearing aid may determine whether the child's residual hearing is used or abused (Seewald, 1991). Use of an individually-measured RECD in the coupler-based fitting approach also ensures that the real-ear aided gain (REAG)/ real-ear aided response (REAR) can be accurately predicted to within approximately 2 dB (Seewald, Moodie, Sinclair & Scollie, 1999). It is important to note that R E A R and derived OSPL90 values can be predicted with 7 a superior level of accuracy when the RECD is measured for the individual child, as opposed to relying on average RECD values (Bagatto et al., 2002). Sources of Variability in Fitting Amplification for Infants Threshold Data and Prescriptive Targets As a result of earlier identification of hearing loss and an increase in newborn hearing screening programs, the demand for accurate and objective measures to determine infant hearing thresholds has increased (Sininger & Cone-Wesson, 2004). The auditory brainstem response (ABR) is a common clinical procedure that provides accurate threshold data for the infant population (Stapells, 2000a). However, as Bagatto et al. (2005) point out, there are several issues surrounding the use of A B R threshold estimates for calculating hearing aid prescriptions. Threshold estimations based on A B R testing are referenced in dB normalized HL (nHL) and unfortunately, the method for obtaining the nHL reference has not been standardized (Bagatto et a l , 2005). Because the methods of calibration are different for audiometers and A B R systems, the thresholds obtained from each system also differ. The result is a discrepancy between behavioural and electrophysiologic results for a given patient (Stapells, 2000a). A B R threshold estimations have been found to be 10 to 15 dB higher than the behavioural thresholds of infants and young children with sensorineural hearing loss (Stapells, Gravel, & Martin, 1995). Significant research has been done regarding the difference between behavioural and A B R thresholds, and the majority of this research has shown that the tone A B R to provides good threshold accuracy (Stapells, 2002b). It is important, however, to keep in mind the discrepancy between A B R and behavioural thresholds, especially when fitting amplification on infants and young 8 children. Without this consideration the possibility of over and under-amplification is increased (Bagatto et al., 2005). Another issue to consider when using A B R threshold estimations is the effect of ear canal acoustics (Bagatto et al., 2005). If a child's A B R threshold estimations measured at 1 month of age are compared with their behavioural thresholds, measured at least 5 months later, a difference between the nHL values and the later measured H L values may not be the result of an actual change in hearing sensitivity (Bagatto et al., 2005). The difference between the two values may be due to a change in the infant's ear canal acoustics (Bagatto et al., 2005). By using an individually measured RECD, changes in ear canal acoustics between the initial assessment and follow up assessment can be accounted for. Bagatto et al. (2005) examined the relationship between A B R and behavioural threshold estimates. A n RECD was measured for each of the 30 adult participants and A B R threshold estimates and behavioural thresholds were obtained at the same age.. Behavioural thresholds originally measured in dB H L were converted to dB SPL by adding the subject's RECD plus the RETSPL to the original dB H L threshold. The A B R threshold estimation in nHL was also converted to SPL using the following equation: (ABR estimate in nHL - Behavioural Correction + RECD + RETSPL = A B R threshold in SPL) (Bagatto et al., 2005). This means that the changes in ear canal acoustics are accounted for by the R E C D and subsequently, the A B R threshold defined in SPL can be compared with the behavioural threshold, which is also expressed in SPL. The average difference between the A B R threshold estimates and the behavioural thresholds was found to be 6.5 dB with 85% of the A B R threshold estimates falling within 10 dB of the behavioural thresholds (Bagatto et al., 2005). This study found good agreement between 9 A B R and behavioural thresholds; however, Bagatto et al., (2005) point out that further investigation of this method and its feasibility with infants will require longitudinal measurements of the age related changes in infant ears. Despite the early stage of this research, it appears that appropriate RECD measurement can help to reduce some of the variability inherently associated with threshold estimation in infants. RECD Measurement The RECD measure has been shown to have fairly high test-retest reliability in both children and adults (Sinclair et a l , 1996; Munro & Davis, 2003). Research done by Munro and Davis (2003) has shown that test-retest differences for adult RECD measures were reported to be close to 0 dB with a standard deviation of 1 dB. Sinclair et al. (1996) reported that for both children and adults, mean test-retest variability was less then 2 dB. Measurement error has, however, been reported to cause increased RECD variability in both the lowest and highest frequencies (Bagatto et al., 2002; Sinclair et al., 1996). Variability in the low frequencies may be due to leakage of sound. Acoustic leakage will result in RECD values below zero in the low frequency region (Bagatto et al., 2001). If an earmold is used, the probe tube may cause some of the sound to escape from around the earmold. In the case of an immittance or foam tip, the size of the tip may be too small or it may not be inserted deeply enough into the ear canal, consequently allowing sound to escape (Bagatto, 2001). High frequency variability may be the result of insufficient probe tube insertion depth (Bagatto, 2001; Dirks & Kincaid, 1987). High frequency roll-off is considered to be present when the real-ear SPL values at 3, 4 and 6 kHz are lower than the value at 2 kHz by 3 dB or more (Dirks & Kincaid, 1987). This is attributed to insufficient depth of probe tube insertion. The recommended insertion depth 10 for infants has not yet been determined, although it is likely to be less than the 20 mm insertion depth recommended for children. Sinclair and colleagues (1996) varied their probe-tube insertion depth (15-20 mm) depending on the age of the child being tested and the clinician's judgment of ear canal size. Fikret-Pasa and Revit (1992) reported a notch in their RECD data in the 4000 to 6000 Hz range. They account for this by suggesting that the probe tube was placed in a standing wave pressure minimum for the 4000 to 6000 Hz range. They also suggest that the probe tube may have been too close to the sound outlet of the insert earphone, which resulted in inaccurate measures of SPL due to near-field effects (Fikret-Pasa & Rivet, 1992). As the distance between the probe tube and the eardrum increases, the highest frequency at which an accurate estimate of SPL at the eardrum can be obtained decreases (Dirks, Ahlstrom & Eisenberg, 1994). Moodie et al. (1994) recommend a standard insertion depth from the intertragal notch of between 20-25 mm for children, therefore ensuring probe tube placement within 5mm of the tympanic membrane (TM). If the probe tube is not deep enough, the measurement may be affected by standing waves (Tharpe, Sladen, Huta & Rothpletz, 2001). Standing wave patterns are created by sound that is reflected from the eardrum (Dirks et al., 1994). The impedance difference between the air in the ear canal and the eardrum causes reflected sound waves to travel outward from the eardrum and consequently, interact with incident acoustic waves (Dirks et al., 1994). The result is standing waves, which create areas of compression and rarefaction, which lead to pressure minima and maxima at different locations in the ear canal (Dirks et al., 1994). If a probe tube happens to be positioned at a pressure minima, the SPL measured at that frequency 11 would be 10 to 15 dB less then if the probe were positioned at a pressure maxima for the same frequency (Dirks et al., 1994). The importance of accurate probe tube positioning is obvious; however, in infants, determining the distance between the probe tube and the T M can be a challenging task. The raises the question of what is the best method for positioning a probe tube in an infant ear canal to ensure accurate probe placement. Techniques for probe tube placement that have been used in adult subjects, such as the tactile perception method, are inappropriate for use with infants (Tharpe et al., 2001). The tactile perception method involves inserting the probe tube until it touches the T M , and then withdrawing it a few millimeters. This may cause discomfort and upset an infant (Tharpe et al., 2001). Another method proposed by Chan and Giesler (1990), suggests the use of an acoustic method that uses the principles of wave theory to estimate the distance between the probe tube and the tympanic membrane. Once the standard wave minima notch is identified, the distance between the probe tip and the eardrum can be calculated. Pressure minima and maxima locations within the ear canal are determined by eardrum impedance and the frequency of the test signal, and are predictable (Dirks et al., 1994). The first pressure minimum is located at a distance that is one quarter the wavelength of the frequency (Dirks et al., 1994). The speed of sound, divided by four times the frequency of the notch, will give the distance of the probe tube form the eardrum (c/4f = d) (Chan & Giesler, 1990). Unfortunately this method is more time consuming, complicated, and is also not considered feasible for use with young children (Tharpe et al., 2001). Sullivan (1988) has described a more simplistic acoustic method that results in placement of the probe tube within 3 to 6 mm of the T M . When the distance of the probe 12 tube from the T M exceeds approximately 6 mm, the measured dB SPL response is decreased as a result of standing waves. This decrease in dB SPL is increased in magnitude above the lA wavelength resonant frequency of the ear canal (Sullivan, 1988). Standing waves are caused by the interaction between the incident acoustic wave and its reflection from the T M (Sullivan, 1998). The standing wave ratio (SWR) is the ratio of maximum to minimum amplitude caused by the interaction between the incident and reflected waves (Sullivan, 1988). It is also the point of lowest sound pressure in the ear canal (Tharpe et al., 2001). Sullivan presented a 6 kHz warble tone and then slowly advanced the probe tube into the ear canal paying attention to the lowest SPL level corresponding to the SWR node. He then marked this position on the probe tube and added another 10mm laterally, and made a second mark on the probe tube. By advancing the probe tube to the second mark, the probe tube should be located within 2 to 6 mm of the T M , without physical contact (Sullivan, 1988). Tharpe et al. (2001) compared Sullivan's acoustic method and the constant insertion depth method. The constant insertion depth method essentially involves inserting the probe tube to a constant depth, however, because of the infant's maturation, the length of the ear canal is increasing (Keefe et al., 1994). Consequently, inserting the probe tube to a constant insertion depth may result in a varying distance between the probe and the T M (Tharpe et al., 2001). The constant insertion depth method consequently raises questions surrounding the possibility that a measured SPL difference may not be a true effect of differences in maturation (Tharpe et al., 2001). Tharpe et al. (2001) took monthly RECD measures on 21 infants during their first year of life. Sullivan's acoustic method was modified by advancing the probe tube only 5mm beyond the SWR node, as opposed to the 13 recommended 10mm for adults. This was done to avoid contact with the infant T M (Tharpe et al., 2001). R E C D values were obtained twice using the acoustic method and twice using the constant insertion depth method for each ear at each visit. Results indicated that, despite concerns of increased distance between the probe tube and T M over time, with use of the constant insertion depth method, measurement error was found to be larger across both methods as the infants got older. In conclusion, they found no compelling evidence to suggest using the acoustic method over the constant insertion depth method for obtaining the RECD (Tharpe et al., 2001).1 The widely used RECD procedure described by Moodie et al. (1994) also supports use of the constant insertion depth method. The RECD procedure was first described by Seewald in 1991, and then reviewed in further detail by Moodie et al. (1994). First the HA-2 coupler is connected to the coupler microphone of the hearing instrument test system and a transducer is coupled to the other end of the HA-2 coupler. The HA-2 coupler is used instead of an HA-1 coupler for two reasons: 1) to reduce the chance of low-frequency acoustic leakage in the coupler measurement, and 2) to allow for the acoustic effects of the earmould coupling to be incorporated into the RECD measurement, i f applicable (Moodie et al., 1994). Then a signal is delivered from either an ER-3A insert phone or another equipment-specific R E C D transducer, into the coupler and the coupler response is measured and stored . Next a foam plug is connected to the transducer and together with a probe tube, is inserted into the child's ear. The probe tube should be placed at a standard insertion depth from the intertragal notch, approximately 20-25 mm for children (Moodie et al., 2004). 1 Note that Tharpe et al. (2001) did not report the within-child age-related changes in their RECD data. 2 Note that the new ER5 insert phone is not recommended for use with the R E C D procedure due to possible impedance mismatches (Sinclair, personal communication). 14 Findings from a recent study by Bagatto, Seewald, Scollie and Tharpe (2006), suggest a strategy where a probe tube and otoacoustic emission (OAE) tip are inserted into the infants ear together, instead of separately. The typical RECD method where the probe tube and tip are inserted separately may not be feasible for use with infants due to their very small ear canals and their position during measurement (Bagatto et al., 2005). In this alternative method an O A E tip is substituted in place of an insert earphone and is then coupled with a probe tube, using plastic film. The probe tube extends approximately 2 mm beyond the sound bore of the O A E tip, which is inserted into the infant's ear canal until the most lateral part of the O A E tip is flush with the opening of the canal. This method was found to be a reliable clinical technique for obtaining RECD measurements in infants between the ages of 2 to 6 months (Bagatto et al., 2006). Mean RECD values across frequencies from this study differed by no more than 1.8 dB when compared with average RECD values for infants who were four months old (Bagatto et al. 2006). Test-retest reliability was also high and the overall mean difference between repeated RECD measures was 2.61 dB (Bagatto et al., 2006). This method was also found to be equivalent to inserting the probe tube and immittance tip separately (Bagatto et al., 2006). Once the probe tube and OAE tip or insert phone are inserted into the ear, the same signal is presented into the ear canal and the SPL across frequencies is measured and stored by the probe microphone system (Seewald, 1991). The RECD is then calculated as the difference between the real-ear SPL and the SPL measured in the coupler across frequencies (Seewald, 1991). 15 Maturation of the External and Middle Ear Because the R E C D measure is thought to reflect and be affected by changes in the properties of the external and middle ear, it is important to understand how these systems change with maturation in infants. In addition, measurement of these properties in infants can be challenging and may require special consideration. Maturation of the External Ear By the time a child is nine years of age their external auditory canal is fully developed and has reached adult length (Northern & Downs, 2002). At birth, however, the infant ear canal is straighter, narrower and much shorter than an adult ear canal. In adults, the inner two thirds of the ear canal wall are bony while the outer third is composed of soft tissue (McLellan & Webb, 1957). In infants however, the ear canal is almost entirely surrounded by soft tissue (McLellan & Webb, 1957). During postnatal development the infant's canal wall and cartilage begin to thicken and gradually become less compliant (Northern & Downs, 1974). The characteristic impedance of the ear canal is largely determined by changes in ear canal area (Keefe, Bulen, Arehart & Burns, 1993). As area increases, the impedance decreases. The characteristic impedance of a 1-month old infant's ear canal is approximately 6 times larger than the characteristic impedance for adults (Keefe, et al., 1993). This means that the transfer of sound energy to the middle ear will be reduced in infants (Keefe et al., 1993). Maturation of the Middle ear Changes in the transfer of sound energy to the middle ear cannot be completely accounted for by the anatomical differences of the external ear between infants and adults (Holte, Margolis & Cavanaugh, 1991). The middle ear also undergoes change as infants 16 mature. The middle ear can be thought of as a system that is comprised of mass, stiffness and resistance. The resulting relationship between these different elements is what determines the impedance of the middle ear (Wiley & Stoppenbach, 2002). The volume of the middle ear cavity is affected by the expansion of the mastoid air sinuses and increasing size of the temporal bone in infants (Anson & Donaldson, 1981). Ikui, Sando, Haginomori and Sudo (2000) also found that the volume of the tympanic cavity is correlated with the degree of pneumatization of mastoid air cells. According to Ikui et al. (2000), the average adult middle ear cavity is approximately 1.5 times larger than those of infants less than one year of age. Middle ear cavity volume affects the stiffness of the middle ear meaning that a smaller middle ear volume results in increased stiffness. The effect of stiffness is most robust in the low frequencies therefore influencing the conduction of low frequency sounds to the middle ear. The admittance (Y), or ease of energy flow into the middle ear, is also affected by middle ear volume so an increase in admittance caused by an increasing middle ear cavity volume will lead to improvements in low frequency sound conduction as an infant matures (Saunders, Doant, & Cohen 1993). Changes in mass affect the conduction of high frequency sounds. The overall mass and resistive elements of an infant's middle ear are high at birth and then decrease as the infant matures. The reasons for this are not entirely understood but are thought to be the result of several different maturational changes (Shahnaz, Miranda, & Polka, submitted). Changes in bone density, tightening of the joints between the middle ear bones (Anson & Donaldson, 1981), the changing orientation of the tympanic membrane (Eby & Nadol, 1986), and incomplete development of the tympanic ring (Saunders et al., 1993), may all have an effect. Absorption of amniotic fluid and mesenchyme, a gelatinous embryonic 17 connective tissue that can be present in the middle ear for up to five months after birth (Paparella, Shea, Meyerhoff, & Goycoolea 1980), may also contribute to the decreasing mass component of the infant middle ear. The presence of this fluid and mesenchyme should therefore affect the conduction of high frequency sounds through the middle ear (Paparella et al., 1980), however recent research done by Hsu, Margolis & Schachern (2000) suggests this may not be the case. They investigated the middle ear status of chinchillas between 1 and 14 days of age and found that despite the absence of mesenchyme in the middle ear, neonatal chinchillas show the same chaotic results on measures of middle ear function as human neonates. In chinchillas, these findings rule out mesenchyme as factor in determining middle ear impedance (Hsu et al., 2000), however the possible effects of mesenchyme in the human neonate remain unclear. Measures of External and Middle Ear: Infant Tympanometry When sound waves reach the T M , the T M is set into motion and energy is transferred to the middle ear. The T M , however, will reflect some of the sound energy. The amount of energy reflected by the T M is directly related to impedance (Wiley & Stoppenbach, 2002). Sound pressure levels measured at the lateral surface of the tympanic membrane, help to describe the transfer of energy to the middle ear. Tympanometry is a measure in which the air pressure is changed in the ear canal while a probe tone is presented (Margolis & Shanks, 1991). The resulting sound pressure level measured at the probe tip gives a measure of admittance or the ease of energy flow into the middle ear (Margolis & Shanks, 1991). Admittance is determined by 1) compliance susceptance (the inverse of stiffness), 2) mass susceptance and 3) friction. When the total susceptance (Bt), comprised of compliance and mass, is positive, the middle ear is stiffness controlled. 18 However i f the total susceptance is negative, the system is mass controlled (Margolis & Shanks, 1991). Tympanometric data indicates that the impedance of an infant's middle ear is mainly dominated by mass and resistive elements at low probe tone frequencies (Holte et al., 1991). In young children and adults however, middle ear impedance is dominated by stiffness elements at low probe tone frequencies (Shahnaz & Polka, 1997). Consequently when conducting standard 226 Hz tympanometry on newborn infants, the tympanometric patterns are significantly different from the patterns observed in older infants and adults (Holte et al., 1991). Y-tympanograms and their rectangular components, susceptance (B) and conductance (G) that are obtained with a 226 Hz probe tone on newborn infants, are often multi-peaked (Holte et al., 1991, Shahnaz et al., submitted). At higher frequencies however, the tympanograms of newborns are single peaked (Holte et al., 1991, Shahnaz et al., submitted). As probe tone frequency increases, the tympanograms of newborns become less complex whereas in adults, as probe tone frequency increases tympanograms become more complex (Holte et al., 1991). Due to the fact that the infant's middle ear is mass dominated at low probe tone frequencies, it may be easier to measure the static admittance with the use of a higher probe tone frequency such as 1 kHz, as the effect of mass or mass reactance becomes more robust at higher frequencies (Shahnaz et al., submitted). Several studies indicate that tympanometry performed with higher probe tone frequencies in newborn infants are the most informative (Baldwin, 2006; McKinley, Grose, Roush, 1997; Rhodes, Margolis, & Hirsch, 1999). In a study conducted by Rhodes and colleagues (1999) on 87 NICU babies, infants who failed 226 Hz and 678 Hz tympanometry obtained pass rates ranging from 30% to 67% on other hearing screening 19 measures. This suggests a high false-positive rate for 226 and 678 Hz tympanometry in the infant ear. In contrast, the 3 ears that failed the 1000-Hz tympanogram also failed all other hearing screening methods, suggesting a low false-positive rate for 1000 Hz tympanometry in infants (Rhodes et al., 1999). In another study McKinley and colleagues (1997) found that in infants less than 24 hours old, a majority (53 out of 55) demonstrated either single or multi-peaked tympanograms at 226 Hz. However at a higher frequency (678 Hz), 34 of the same infants showed flat tympanograms. This suggests that for infants, high frequency tympanometry, is superior in detecting middle ear pathology when compared with 226 Hz. For the purposes of the current study, measures of static admittance were obtained with both 226 and 1000 Hz probe tone frequencies, however only tympanograms obtained with a 1000 Hz probe tone were included in the analysis. Measures of External and Middle Ear: Static Admittance Admittance, determined by compliance susceptance, mass susceptance and friction, can be thought of as the ease of energy flow into the middle ear (Margolis & Shanks, 1991). In order to quantify static admittance, the admittance at either the positive or negative tail of the tympanogram is subtracted from the peak admittance (Holte et al., 1991). At the peak of the tympanogram, the air pressure in the ear canal is nearly equivalent to the air pressure in the middle ear (Margolis & Shanks, 1991). By introducing pressure into the ear canal, the middle ear stiffens, which causes a drop in the admittance of the middle ear. Consequently, the measured admittance value reflects only the contribution of the ear canal volume because the admittance of the middle ear is assumed to be near zero (Margolis & Shanks, 1991). The value of the ear canal admittance can then be subtracted from the peak admittance value to get an estimate of middle ear admittance (Margolis & Shanks, 1991). This method is suitable for use with adults when using a 226 Hz probe tone, but it may not be suitable for use with infants (Holte et al., 1991). By subtracting the tail from the peak of the tympanogram, an assumption is made that the ear canal and middle ear behave as a pure acoustic compliance (Holte et al., 1991). If the assumption that admittance approaches zero at the tympanic membrane when the canal is pressurized to 200 daPa is false, then ear canal volume will be overestimated and consequently the admittance of the middle ear will be underestimated (Shanks & Lilly, 1981). When subtracting the tail from the peak of the tympanogram it is also assumed that variations in ear canal pressure will not affect the ear canal volume or change the admittance phase angle, however this is not always the case (Holte et al., 1991, Shanks and Lilly, 1981). Admittance is the vector sum of conductance (G) and total susceptance (Bt), and it can be expressed in either rectangular notation or polar notation (see Figure 1). Figure 1 Polar and Rectangular Notation (Shahnaz, 2005) 21 POLAR RECTANGULAR CONDU CTANCE (G) IY | =2.1 mmhos $ = 65° W U WS D 40 = Bs+ B r CONDUCTANCE (G) Btotai= 1.5 mmho G - 1.5 mmhos In polar notation, admittance is expressed by its magnitude and phase angle (Margolis & Shanks, 1991). The admittance phase angle refers to the angle formed by the admittance vector and the horizontal axis. Subtracting the tail from the peak admittance of a tympanogram will only provide an accurate estimate of middle ear admittance when the phase angle associated with the peak admittance and the phase angle associated with the tail are equal (Shanks & Margolis, 1991). In adults, this condition is usually met at 226 Hz; however, at higher frequencies, significant changes in phase angle occur due to changes in ear canal pressure (Margolis & Shanks, 1991). Data from Holte et al. (1991) indicate that in infants 4 months of age or younger, there are significant changes in phase 22 angle and the ear does not behave as a pure compliance. Consequently, in these cases static admittance must be calculated by using the rectangular components of admittance. In rectangular notation, admittance is equal to the sum of conductance (G) and susceptance (Bt) elements but conductance and susceptance are vectors that operate in different directions and can therefore not be added together (Margolis & Shanks, 1991). In order to obtain the admittance value, Pythagorean Theorem is used in the following formula: Ytm = V Bt + G The total susceptance and conductance components are squared and added together and then the square root of that sum is taken to provide the admittance Ytm. For the purpose of this study, static admittance was calculated using the rectangular admittance components as well as by subtracting the tail value from the peak admittance. Compensation for Ear Canal Volume When ear canal admittance at the tail of the tympanogram is subtracted from peak admittance, the question of whether the positive or the negative tail provides a better estimate of ear canal admittance arises. Shanks and Lilly (1981) reported that the negative tail value is the most accurate estimation of actual ear canal admittance in adults, but the mean diameter changes in the ear canals of newborns have been demonstrated to be much larger for negative pressures than for positive pressures (Holte, Cavanagh, & Margolis, 1990). Due to the compliant nature of the infant ear canal, negative tail compensation is likely to result in collapse of the ear canal in infant populations (Shahnaz et al., submitted). Consequently it may be more feasible to compensate using the positive tail in infants. For the purpose of this study, positive tail compensation to estimate ear canal volume was used. 23 The developmental changes in the external and middle ear are likely a large factor in the associated changes seen in RECD values as infants mature. According to Voss and Herrmann (2005), the sound pressure measured at the T M is the result of many different elements including ear canal length, ear canal diameter, and the impedance of the T M . As the volume of the middle ear cavity increases and mass decreases as infants mature, the impedance of the middle ear and tympanic membrane also decreases. Sound is therefore more easily transmitted through the middle ear resulting in lower SPL measurements in the canal. The presence of higher SPL levels measured in infant ear canals, compared to adult canals, can in part be accounted for by lower admittance/higher impedance in the infant ear. This is supported by the finding that infant RECD values are significantly higher than the values for adults (Feigin et al., 1989). Middle Ear Pathologies Pathologies of the middle ear have been shown to have a significant impact on the RECD (Martin, Munro & Lam, 2001; Martin, Westwood & Bamford, 1996). By substituting average age appropriate RECD values for an individually measured RECD, in the case of an individual with middle ear pathology, more harm will be done than good. If average RECD values are used to calculate hearing aid targets, depending on the pathology, the targets will either provide too much gain or insufficient gain. Martin et al. (2001) studied a group of 24 adults, 12 whom had T M perforations and the other 12 who had normal middle ear function. RECD measurements were taken from all subjects and the results indicated that although the RECD values at high frequencies were similar across both groups, in the low frequencies, the mean RECD was approximately 1 OdB lower for the group with T M perforations (Martin et al., 2001). The T M perforation effectively acts 24 as a path for sound to escape from the ear canal into the middle ear. The SPL levels are more affected in the low frequencies due to the fact that the impedance of the leakage path increases with frequency (Martin et al., 2001). Consequently, i f average RECD transform values were used with an individual who has a T M perforation, the predicted low frequency SPL will be overestimated by approximately 10 dB (Martin et a l , 2001). In a similar study Martin et al. (1996) investigated the effects of otitis media with effusion (OME) on RECD measurements in children. The experimental group consisted of 14 children with a mean age of 6 years. The presence of middle ear fluid was confirmed by flat tympanograms and otoscopy. The probe frequency used for tympanometry was not reported. Flat tympanograms were defined by the absence of peaks (greater than 0.1 ml) in the compliance in the pressure range between -200 to +200 daPa. The control group contained 14 age matched peers with normal middle ear status. From 200 to 3000 Hz the mean RECDs for the children with O M E exceeded the RECD values of the control group by 0.8-3.5 dB. The increase in the low and mid frequency RECD values within the OME group is likely the result of the increased mass and stiffness of the middle ear, caused by the presence of fluid (Martin et al., 1996). The fluid also causes an increase in middle ear impedance, which results in higher SPL levels in the low and mid frequencies (Martin et al., 1996). Consequently, the use of average age-appropriate RECDs for children with OME will result in larger than expected ear canal SPLs in the low and mid frequencies. Clinically, a few dB of error resulting from the presence of O M E may be viewed as insignificant, however Martin et al. (1996) point out that it is important to reduce as much variation as possible as quickly as possible to ensure that appropriate amplification for children can be fit in a timely manner. Furthermore middle ear problems are highly 25 prevalent in children and Mora et al. (2002) predict that 80% of children experience at least 1 episode of OME. In OME, fluid is present behind the eardrum but there are no signs of active infection, making it particularly difficult to detect (Sagraves, Maish & Kameshka, 1992). The data from Martin et al. (1996), supporting the presence of higher RECD measures in children with OME, combined with the difficulty in detecting OME, once again supports the importance of individually measured RECDs whenever possible. Average versus Individually Measured RECDs It is known that infant ears differ from adult ears in somewhat predictable ways. For a given H L value, the SPL in a child's ear may be up to 20 dB greater than in the adult's ear (Bagatto et al., 2005). Consequently, in comparison with adults the infant RECD values will be significantly higher. Average adult RECD values can be seen in Appendix C and Appendix D. As a child grows however, their ear canal acoustics change and the H L thresholds that are needed to generate a given ear canal SPL also increase (Bagatto et al., 2005). Variability between the SPL levels in infant and adult ears can partly be accounted for by changes in ear canal volume (Feigin et al., 1989; Westwood & Bamford, 1995) as well as by the impedance of the middle ear (Holte et al., 1991). The question is whether or not this variability can be accounted for by an infant's age alone and whether the use of an average age-appropriate RECD value will provide a sufficiently accurate estimate of ear canal SPL. Previous research done by Feigin et al. (1989) was the basis for most age-related RECD predictions that were used clinically up until quite recently. These predicted values were included in DSL v4.1, most hearing instrument programming modules, and probe microphone systems (Seewald, Cornelisse, & Ramji, 1997). Feigin et al. (1989) measured 26 RECDs in thirty-one children that ranged in age from 1 month to 5 years. The measurements were then divided into 12 to 24 month age categories. Unfortunately because the age categories were 12 months to 2 years apart, the normative data set may not describe the changes in ear canal acoustics in enough detail to differentiate an infant who is 3 months of age from one who is 6 months of age (Bagatto et a l , 2005). The same predicted RECD value would be applied to a 1-month-old infant and to an 11-month-old infant (Bagatto et al., 2002). This was problematic because the infant who is one month of age has a much smaller ear canal volume in comparison to an infant who is 11 months of age. The result is a higher SPL level in the younger infant's ear that could result in over-amplification and damage of any residual hearing. More recent research by Bagatto et al. (2002) is the basis for the predicted RECD values in DSL v5.0 (Bagatto et al., 2005). RECD measurements were taken in 392 children ranging from 1 month of age to 16 years. Two coupling procedures, acoustic immittance tips and personal earmolds were used and cross-sectional data is available in 1-month age increments. Results indicated significant between-subject variability across frequencies (Bagatto et al., 2002). RECD values for infants and young children have been found to vary significantly among infants and children of the same age (Feigin et al., 1989; Seewald & Scollie, 1999; Westwood & Bamford, 1995). Bagatto's 2002 data showed the range of variability was 16dB at 500 Hz, and 14 dB at 2000 Hz, for twenty-two infants less then 6 months of age. The RECD predictions were poorly associated with subject age, and using these equations to predict RECDs may result in values that vary by more than 14 dB from the actual RECD in more then 95% of cases (Bagatto et al., 2002). In order to improve the predictions, the 27 data were re-analyzed and errors in probe microphone measurement, resulting from sound leakage and shallow probe tube placement, were accounted for (Bagatto et al., 2002). The data showed a limited age related change in the high frequencies and also suggested that the largest changes, around 2000 and 4000 Hz, take place within the first 10 to 12 months of life (Bagatto et al., 2002). Based on the fact that predicted RECDs can vary by more than 14 dB from actual measured RECDs, it has been recommended that individually measured RECDs should not be replaced by age-appropriate predictions (Bagatto et al., 2005). This variability is not limited to infants. Fikret-Pasa & Revit (1992) investigated the effect of individual variability in RECD values within a clinical population of 15 adults. RECD measurements were made on each subject as well as on K E M A R . Individuals with RECDs that differed from K E M A R measurements by more than 4dB, at two or more frequencies between 500 and 4000 Hz, were considered to be significantly different from average. Half of the subjects tested were significantly different from average. Fikret-Pasa and Rivet (1992) attribute some of this variance to known middle-ear conditions in these subjects; however, subjects with normal middle ear status were also found to have significantly different from average RECDs. Due to the high percentage of subjects with notable differences from the K E M A R RECD, measuring individual RECDs routinely in the hearing aid selection process is advised, even in adults (Fikret-Pasa & Revit, 1992). However, in cases where this is not attainable, such as with an unco-operative child, age appropriate RECD predictions from Bagatto et al. (2002) can be substituted. Audiologists are strongly advised, whenever possible, to individually measure RECDs for infants and children as part of the hearing aid fitting and selection process (Moodie et al., 1994) Unfortunately, it appears that most audiologists do not routinely measure the RECD. When 425 practicing audiologists who regularly fit hearing aids on infants and children were surveyed regarding the methods they typically use to verify the gain/frequency specifications of a hearing instrument, only 50% of the respondents indicated using some type of probe microphone measure (Tharpe, 2000). Respondents from the survey did indicate that they most often use average age-appropriate RECDs and measured RECDs to select output limiting characteristics (Tharpe 2000). The results of Tharpe's (2000) survey suggest that use of probe microphone measures with young children is increasing; however, improvement is still needed. Clinical Implications A fair amount of data exists on RECDs for adults, but there is little published information on infants and young children. It has been suggested that RECD values become adult like by approximately 7 years of age (Feigin et al., 1989), although Dillon (2001) suggests that once a child is older than 5 years the RECD becomes similar to adult values. Other research has suggested that this point may be as early as two years of age (Kruger, 1987; Voss & Herrmann, 2005). Data from Feigin et al. (1989) suggest that infants under 12 months of age show the largest RECDs, and Bagatto et al.'s (2002) data indicate that the largest changes in the RECD measurement seem to occur during the first 10 to 12 months of life (Bagatto et al., 2002). To date, however, there is no data that quantifies how the R E C D changes over time in an individual infant. The question is how quickly and by how much does the RECD change during the first few months of life in 29 individual infants. Cross-sectional data indicates RECD values show the greatest amount of change in the first year of life, so by obtaining RECD measurements shortly after birth and then again at a 1 -month interval, longitudinal data will help quantify change in the measured ear canal SPL as the infant matures. These data will also provide information regarding the development of the ear canal and middle ear as the infant matures. If there are significant changes in individual RECD data, then these individual changes can be compared across infants to identify any possible patterns of RECD measurement change due to infant maturation. This information can then be used to improve the accuracy of infant hearing assessment as well as hearing aid fittings. If, however, there are no significant RECD changes over a 1 -month interval, then this knowledge can be used to improve clinical efficiency by reducing the number of tests and procedures needed at each visit. The main questions this study will address include: 1) Is there a significant decrease in RECD values for an individual infant over a 1 month period? If yes, is this decrease predictable from the infants corrected age, and are the RECD values at the time of the second measurement predictable from the RECD values obtained at the first measurement? 2) Can measures of static admittance and ear canal volume be used to predict RECD values at any or all frequencies? 3) How much measurement variability exists within a single test session, .and are measured RECD changes over a 1-month interval large enough to detect within the quantified measurement error? Methods In this study we measured RECD values for 14 infants over a 1-month interval. RECD measurements were taken on two separate visits, first when the infant was 30 approximately two to three weeks old and then again approximately 1 month later. This project was approved by the University of British Columbia's Clinical Research Ethics Board, as well as the Children's and Women's Health Centre of British Columbia Research Review Committee. Recruitment Subjects were recruited through pre-natal classes offered by St. Paul's Hospital and The Douglas College Pre-natal Program. A brief talk regarding the importance of hearing for speech and language development as well as the benefits of early intervention was provided. The study was also explained and parents were given the opportunity to ask questions. Interested parents were then given a copy of the invitational letter and parental consent form and asked to contact researchers involved, one to two weeks after their child was born. Subjects were also recruited through the Children's and Women's Health Centre of British Columbia. Researchers explained the study to hospital nurses, who then discussed the study with new parents and distributed invitational letters. Additional recruitment was conducted through the UBC health clinic, local physicians' offices, and word of mouth. Subjects Subjects consisted of 14 infants (7 females, 7 males) ranging in age from 7 to 25 days at the time of the initial visit. For statistical purposes gestational age was calculated for all participants. All subjects had normal middle ear status confirmed by tympanometry (1000 Hz) and transient evoked otoacoustic emissions (TEOAEs). Tympanometry was also used to obtain a measure of equivalent ear canal volume. Otoscopy was also performed to ensure no wax or residual vernix was present in the canal. Subjects were 31 tested regardless of sleep state, however parents were advised to feed babies 30 minutes prior to testing in order to facilitate sleep. Procedure This study was part of a larger longitudinal study of external and middle ear maturation, so at each visit, four tests were completed: TEOAEs, tympanometry, RECD measurements and wide band reflectance (WBR). Testing was conducted in the subject's home in a quiet room. In most instances the infant was held by a parent and the exposed ear was tested first. A quick otoscopic check was preformed prior to testing to ensure there was no residual vernix or excessive earwax present in the canal. TEOAES In order to be included in this study all infants had to pass TEOAE screening criterion to confirm normal middle ear status using the QuickScreen system on the ILO-292 otodynamics. Stimuli consisted of 84 dB SPL clicks. A 3 dB signal to noise ratio (SNR) was required at 1 and 1.5 kHz and a 6dB SNR was required at 2, 3 and 4 kHz. In accordance with Norton et al. (2000), to pass, a response at 3 out of the 5 frequency bands was required, along with 70% reproducibility and a minimum of 50 sweeps. Tympanometry Using the GSI-TympStar version 2, a 226 Hz Ya tympanogram was recorded followed by 226 Hz Ba and Ga tympanograms. Next a 1000 Hz Ya tympanogram was obtained followed by 1000 Hz Ba and Ga tympanograms. The sweep pressure method was used where the probe frequency is fixed and the air pressure is swept from positive to negative values. The pressure rate used was 200 daPa/sec. Decreasing pressure (positive to negative) was used for all recordings because this approach results in fewer irregular 32 tympanograms (Wilson, Shanks & Kaplan, 1984) and a reduced chance of collapsing canals (Holte et al., 1991), when compared with an increasing direction of pressure change. RECD Measurement A coupler measurement was made at the beginning of each testing session. Using the Verifit system, the coupler microphone and RECD transducer were attached to the H A -2 coupler, and the signal was introduced and stored. For measurement of the real-ear response, the RECD transducer was coupled to an immittance tip. Two different methods of constant-insertion depth were used, depending on what worked best for an individual child. In the first method, the probe tube was taped to an immittance tip according to the method described in Bagatto et al. (2006). The probe tube extended 2 mm beyond the end of the immittance tip to ensure probe tube placement within 5 mm of the tympanic membrane. Both the immittance tip and probe tube were then inserted into the infant's ear canal simultaneously. In the second method, the probe tube was marked at 15 mm and then inserted until the marker was at the point of the tragus. The immittance tip, which was coupled to the RECD transducer, was then inserted separately. These different methods of constant insertion depth have proven to be equivalent because in both cases, the probe tube is inserted to a depth of within approximately 5 mm of the tympanic membrane (Bagatto et al., 2006). With the probe tube and transducer in place, a 50 dB broadband noise signal was delivered by the transducer. The measurement was then recorded and stored. The Verifit automatically calculates the difference between the real-ear measure and the coupler and the resulting RECD values are displayed at 250, 500, 750, 1000, 1500, 2000, 3000, 4000 and 6000 Hz. To ensure a reliable measure, the probe tube and immittance tip were 33 completely removed after each measure and reinserted. Two RECD measures were made on each infant's ear, whenever possible. In some cases, e.g. fussy or crying infants, only one measure per ear was obtained. Wide Band Reflectance WBR was also measured as part of the larger longitudinal study, but the results are not reported here. Results The main objective of this study was to determine whether RECD values follow a predictable pattern of change over a 1-month period, as infants mature. If a significant decrease in RECD values exists, then can this decrease be predicted by the infants corrected age and can the RECD values at the second time of measurement be predicted from the RECD values at the first measurement? We also want to determine if measures of static admittance and ear canal volume can be used to predict the RECD at any or all frequencies. Lastly, we want to quantify the measurement variability that exists within a single test session and determine whether changes in RECD values over a 1 -month interval are large enough to detect within RECD measurement error. For each infant, two RECD measurements were taken for each ear at each visit. Infants participated in 2 visits each. Treatment of the Data Missing values were replaced with a series mean that was obtained from the data of other subjects. Over the course of this study 112 RECD measurements were attempted; seven of these measurements could not be completed. At visit 1 a second RECD measure to quantify test-retest reliability on the right ear was not obtainable in four subjects. For the left ear a second RECD measurement was not obtainable in three subjects. No RECD 34 values were obtained for the left ear of one subject at visit 1; however, two measures were obtained for the right ear. At the second visit RECD values were unable to be measured on the right ear of one subject however, 2 measures were obtained for the left ear. Two RECD measures were obtained for the left ear on all subjects at visit 2. Data from this study is also being used as pilot data for a larger longitudinal study on maturational changes so, for the purpose of this study, a 90% confidence interval, along with an alpha level of .1, was used for most statistical tests. The purpose of this lenient criterion was to minimize the risk of missing a significant trend that would be important to follow up in future studies. RECD Values at Visit 1 and Visit 2 To determine if there was a significant change in RECD values between visits 1 and 2 at each of the nine measured frequencies, RECD values at each visit were compared using a multi-variate analysis of variance (MANOVA). Overall there was a significant decrease in RECD values between visit 1 and visit 2, (F(9, 19) = 5.469, p = .001. Individual ANOVAs revealed significant changes at only 3 frequencies. At 0.25 kHz there was a significant increase in RECD value, F(l, 27) = 4.479, p = .044, at 1.5 kHz there was a significant decrease in RECD value, F(l, 27) = 3.290, p = .081, and at 2.0 kHz there was also a significant decrease in RECD value, F(l, 27) = 12.506, p = .001. The p values for the remaining 6 non-significant frequencies can be seen in Table 1. Individual RECD values for each subject at visit 1 can be viewed in Appendix A. Individual RECD values for each subject at visit 2 can be viewed in Appendix B. 35 Table 1 Significant Change in RECD values between Visit 1 and Visit 2, across Frequency Frequency (Hz) Mean RECD at Vist 1 Mean RECD at Visit 2 df F P value * = p < .1 (significant) 250 .423 3.00 1 4.479 .044* 500 7.885 8.630 1 .569 .457 750 11.923 11.259 1 .683 .416 1000 13.038 12.519 1 .621 .438 1500 11.808 10.444 1 3.290 .081* 2000 11.385 9.296 1 12.506 .001* 3000 12.654 12.000 1 1.678 .206 4000 9.808 10.370 1 1.217 .280 6000 14.423 13.333 1 1.613 .215 A forward regression analysis was then done to determine if the RECD at visit 2 could be predicted from either the RECD value at visit 1 or the infant's corrected age. It was found that corrected age was a significant predictor at 250, 3000 and 4000 Hz and the RECD value at visit 1 was a significant predictor of the RECD value at visit 2 at 3000 Hz only. Table 2 shows significant predictors at each frequency along with their F and R-square values. 36 Table 2 Corrected Age and RECD at Visit 1 as Significant Predictors by Frequency Frequency Predictors Included in the Regression Model R-square Change F Change dfl df2 Sig.F Change 250 Corr. Age .108 3.143* 1 26 .088 3000 RECD Visit 1 .205 6.700* 1 26 .016 Additional variance accounted for by Corr. Age .144 5.536* 1 25 .027 Total Variance R Square F dfl df2 Sig.F RECD Visit 1 and Corr Age .349 6.702 2 25 .005 4000 Corr. Age .119 3.524* 1 26 .072 * Significant F change Frequencies with no significant predictors are not included RECD Predictors A second forward regression analysis was performed to determine whether static admittance (SA) and ear canal volume (ECV) are able to predict RECD values. The specific variables we examined were; 1) ECV at 226 Hz B+; 2) ECV at 1000 Hz B+; 3) SA computed from a 1000 Hz + 200 Ya tympanogram and 4) SA 1000 Hz + 200 Ytm, computed from the rectangular components, Ba and Ga. Due to the fact that many of the static admittance and ear canal volume measurements were highly correlated (see Table 3) a transformation of variables was performed. Three different sets of transformed variables were created to determine which transformations best dealt with the problem of 37 collinearity: 1) ECV at 1000 Hz B +, SA 1000 Hz +200 Ytm, and SA 1000 Hz +200 Ya were all centered, and ECV at 226 B+ was not transformed; 2) SA 1000 Hz +200 Ytm, and SA 1000 Hz +200 Ya were centered, the log of ECV at 1000 Hz B+ was taken and ECV at 226 B+ was not transformed; 3) SA 1000 Hz +200 Ya was centered, the log of ECV at 1000 Hz B+ was taken, and SA 1000 Hz +200 Ytm and ECV at 226 B+ were not transformed. The sets of transformed variables were then compared and were found to be highly similar (see Table 4). Variable sets 2 and 3 were found to be identical in the amount of variance accounted for at each frequency, therefore, variable set 1 was excluded and variable set 2 was chosen for the regression analyses. Table 3 Pearson Correlations between RECD Predictors ECV 226 B+ ECV 1000 B+ SA 1000 +200 Ya SA 1000 +200 Ytm ECV 226 B+ 1 .195 .233 .251 ECV 1000 B+ .195 1 .482** .420** SA 1000 +200 Ya .233 .482** 1 .827** 1000 SA +200 Ytm .251 .420** .827** 1 * Correlation is significant at the 0.01 level. 38 Table 4 Variable Sets 1, 2 and 3 for Determining Best Data Transformations Frequency Variable Set 1 Variable Set 2 Variable Set 3 #Of predictors Variance accounted for #Of predictors Variance accounted for #Of predictors Variance accounted for 250 Hz 2 .142 2 .144 2 .144 500 Hz 2 .168 2 .168 2 .168 750 Hz 2 .204 2 .191 2 .191 1000 Hz 2 .274 2 .262 2 .262 1500 Hz 1 .166 1 .166 1 .166 2000 Hz 0 0 1 .058 1 .058 3000 Hz 0 0 0 0 0 0 4000 Hz 1 .066 1 .066 1 .066 6000 Hz 1 .201 2 .236 2 .236 Regression analyses indicated that static admittance was a predictor of RECD values at 5 out of the 9 analyzed frequencies including: 500, 750, 1000, 1500 and 4000 Hz. Static admittance was also the largest predictor at all of these frequencies. At 500 Hz, SA 1000 Hz + 200 Ya accounted for 6.8% of RECD variance, at 750 Hz, SA 1000 Hz +200 Ya accounted for 12.5 % of the variance, and at 1000 Hz, SA 1000 Hz +200 Ya accounted for 18.4% of the variance. At 1500 Hz, SA 1000 Hz +200 Ya accounted for 16.6% of the variance, and at 4000 Hz, SA 1000 Hz +200 Ytm accounted for 6.6% of the variance. Ear canal volume was also a significant predictor at 6 frequencies: 250, 500, 750, 1000, 2000 and 6000 Hz, however it was only the largest predictor at 3 frequencies: 250, 2000 and 39 6000 Hz. At 250 Hz, ECV at 1000 B+ accounted for 8.2% of the variance, at 2000 Hz, ECV at 1000 B+ accounted for 5.8% of the variance, and at 6000 Hz, ECV at 1000 B+ accounted for 17.9% of the variance. Static admittance therefore appears to be a larger predictor of RECD values compared with ECV due to the fact that S A accounted for more variance at more frequencies. It is also noted that static admittance at 1000 Hz +200 Ya accounted for more variance than SA at 1000 Hz Ytm and ECV at 1000 B+ accounted for more variance than ECV at 226 B+. Table 5 shows significant predictors at each frequency, along with their F and R-square values. Table 5 Significant SA and ECV Predictors by Frequency 40 Frequency (Hz) Predictor R Square Change F Change dfl df2 Sig.F Change 250 ECV 1000 B+ .082 4.728 Additional variance accounted for by ECV 226 B+ .062 3.787 Total Variance R Square dfl ECV 1000 B+and ECV 226 B+ .144 4.382 53 52 df2 52 .034 .057 Sig. F .017 500 Predictor R Square Change F Change dfl SA 1000 +200 Ya .068 3.842 Additional variance accounted for by ECV 1000 B+ .101 6.290 Total Variance R Square dfl SA 1000 +200 Ya and ECV 1000 B+ .168 5.258 df2 53 52 df2 52 Sig.F Change .055 .015 Sig.F .008 Predictor R Square Change F Change dfl df2 Sig.F Change SA 1000 .125 7.600 +200 Ya 1 53 .008 41 Additional .065 4.181 1 52 .046 variance accounted for by ECV 1000 B+ Total R Square F dfl df2 Sig.F Variance SA 1000 .191 6.119 2 , 52 .004 +200 Ya and ECV 1000 B+ ' Predictor R Square F Change dfl df2 Sig.F Change Change SA 1000 T84 11.954 I 53 !001 +200 Ya Additional .078 5.479 1 52 .023 variance accounted for by ECV 1000 1000 B+ Total R Square F dfl df2 Sig.F Variance SA 1000 .262 9.221 2 52 .000 +200 Ya and ECV 1000 B+ Predictor R Square F Change dfl df2 Sig.F Change Change 1500 SA 1000 .166 10.558 1 53 .002 +200 Ya 2000 ECV 1000 .058 3.271 1 53 .076 B+ 3000 NS 42 SA 1000 .066 3.732 1 53 .059 +200 Ytm ECV 1000 .179 11.593 1 53 .001 B+ Additional .056 3.817 1 52 .056 variance accounted for by ECV 226 B+ Total R Square F dfl df2 Sig.F Variance ECV 1000 .236 8.013 2 52 .001 B+and ECV 226 B+ NS: no significant predictor. Scatter plots illustrating the correlation between RECD values and their significant predictors at each frequency can be seen in Figures 2-14. Generally as ear canal volume increases, a corresponding decrease in RECD values is expected which will result in a negative correlation. Positive correlations are seen at 250 and 500 Hz. It is likely that the significant correlations for ECV at 226 B+ are due to the single outlier and may not represent a true relationship, but further data need to be collected to determine whether this is the case. A negative correlation is also expected between static admittance and RECD values. As RECD values decrease, static admittance should increase. This was the case at all frequencies except 4000 Hz where a positive correlation was observed. Figure 2 250 Hz Correlation between RECD values and ECV 1000 Hz B+ N X o i n CN Kfl <u £ . -io-cs > Q o a -20 2~ Z\ 00 A 2 3 A .5 Variance due to ECV at 1000 Hz B+ Rsq = 0.0819 Figure 3 250 Hz Correlation between RECD values and ECV at 226 Hz B+ N O m CN td > Q U 10 -10 -20 0 fl 0 0 a o 0 mo 0 0 a a aa mo 0 0 a 0 ^ ^ * * — * L > D a 0 D ^ 0 0° 0 % 0 (o 031 00 0 0 a DO 0 0 0 0 Rsq = 0.0467 .2 .4 .6 .8 1.0 1.2 1.4 1.6 Variance due to ECV at 226 Hz B+ Figure 4 500 Hz Correlation between RECD values and SA 1000 Hz +200 Ya Rsq = 0.0676 -1.0 -.5 0.0 .5 1.0 1.5 2.0 2.5 Variance due to SA 1000 Hz +200 Ya Figure 5 500 Hz Correlation between RECD values and ECV at 1000 Hz B+ Rsq = 0.0327 Variance due to ECV at 1000 Hz B+ Figure 6 750 Hz Correlation between RECD values and SA 1000 Hz +200 Ya 20 N X o m t> > Q O 8 10H on -1.0 • • • • D O • • fm i mi i -o~ _^g__^  • • • • • mQ"ts—Q. • • • • • Er-~-^ • • • • • • • • • • • • • • • • • Rsq = 0.1254 -.5 0.0 .5 1.0 1.5 2.0 2.5 Variance due to S A at 1000 Hz +200 Ya Figure 7 750 Hz Correlation between RECD values and ECV at 1000 Hz B+ N o > Q O 20 10 m • • • • no • • • • • • • • • • • • • • D E D • , " • • D • a a a a HE • • o • -.1 0.0 .1 .2 .3 .4 Variance due to ECV at 1000 Hz B+ Rsq = 0.0073 Figure 8 1000 Hz Correlation between RECD values and SA 1000 Hz +200 Ya N O o o <Ji ii J3 "c3 > O u Rsq = 0.1840 -1.0 -.5 0.0 .5 1.0 1.5 2.0 Variance due to SA at 1000 Hz +200 Ya Figure 9 1000 Hz Correlation between RECD values and ECV at 1000 Hz B+ N O o o ta <Si CD > Q U a 22 20 18 16 14 12 10 8 6 • • • • • • • • • • m a a a a •—o a a n a • • • m • • • • • • • • • • -.2 -.1 0.0 .1 .2 .3 .4 Variance due to ECV at 1000 Hz B+ Rsq = 0.0058 / Figure 10 1500 Hz Correlation between RECD values and SA 1000 Hz +200 Ya 30 T 1 N X o o in <u 13 > Q U 20 101 0 • • a a an a o—B-_jp^ • a • •a • cm o • • • • • • • • • • • • • • Rsq = 0.1661 -1.0 -.5 0.0 .5 1.0 1.5 2.0 2.5 Variance due to SA at 1000 Hz +200 Ya Figure 11 2000 Hz Correlation between RECD values and ECV at 1000 Hz B+ N X o o o (N t3 (U 13 > Q O 18 16 14 12 10 8 • m • • • • • • • • • • "0 •cm • • • a • m • m m • m a n • • • • • • -.2 -.1 0.0 .1 .2 .3 .4 Variance due to ECV at 1000 Hz B+ Rsq = 0.0581 Figure 12 4000 Hz Correlation between RECD values and SA at 1000 Hz +200 Ytm Figure 14 6000 Hz Correlation between RECD values and ECV at 226 Hz B+ 49 Test-retest Reliability A Pearson Correlation was conducted to determine whether RECD values for the within-visit test and retest were significantly related at both visits. Data from both visits were included in the analysis. As expected RECD values between test 1 and the immediate retest were significantly correlated at all nine measured frequencies (see Table 6). Table 6 Pearson Correlations between Test 1 & Test 2 across Frequency at both Visits Test 1/Test2 at Frequency (Hz) Pearson Correlation Significance N 250 .559 .000 56 500 .546 .000 56 750 .462 .000 56 1000 .491 .000 56 1500 .621 .000 56 2000 .543 .000 56 3000 .535 .000 56 4000 .530 .000 56 6000 .555 .000 56 A single sample T-test was also conducted which took the absolute value of the mean difference between test 1 and test 2 at each visit and compared this value to a test value of zero. Mean differences according to frequency were relatively small between test 1 and test 2 at both visits (see Table 7 for means and confidence intervals), but were significantly different from zero. Average test-retest variability was also calculated for all frequencies at each visit and then again across both visits to get an overall value. Average test-retest variability across both visits was 2.03 dB with a SD of 2.28 dB (see Table 8). 51 Table 7 Test-retest Reliability at each Visit Frequency (Hz) Visit t df Sig Mean Difference 95% Confidence Interval of the Difference Lower Upper 250 1 5.054 27 .000 3.1786 1.8882 4.4690 500 1 5.318 27 .000 2.4918 1.5303 3.4532 750 1 5.484 27 .000 1.9516 1.2215 2.6818 1000 1 6.282 27 .000 1.6777 1.1297 2.2258 1500 1 5.098 27 .000 1.8255 1.0908 2.5603 2000 1 5.375 27 .000 2.0382 1.2602 2.8162 3000 1 5.426 27 .000 1.5247 .9482 2.1013 4000 1 4.752 27 .000 1.7077 .9703 2.4451 6000 1 4.750 27 .000 2.5321 1.4384 3.6259 250 2 5.479 27 .000 3.3889 2.1198 4.6580 500 2 5.020 27 .000 2.7249 1.6110 3.8387 750 2 4.081 27 .000 . 2.2275 1.1076 3.3474 1000 2 4.964 27 .000 1.9365 1.1361 2.7369 1500 2 4.661 27 .000 1.6111 .9019 2.3204 2000 2 4.823 27 .000 1.2857 .7388 1.8326 3000 2 4.588 27 .000 1.6720 .9243 2.4197 4000 2 4.965 27 .000 1.0767 .6317 1.5217 6000 2 4.501 27 .000 1.6085 .8752 2.3418 52 Table 8 Average Test-retest Variability for each Visit and Combined Visits Visit 1 Visit 2 Overall Average test-retest 2.103 1.948 2.026 SD 2.237 2.318 2.277 The mean difference between visit 1 and visit 2 according to frequency was compared to the upper confidence interval for test-retest variability. A l l of the mean differences at each frequency between visit 1 and visit 2 were less then the upper limit of the 95% confidence interval for test-retest variability (see Table 9). 53 Table 9 Visit 1, Visit 2 Changes Compared to Upper Confidence Interval for Test-retest Variability Frequency (Hz) V1/V2 Mean 95% Confidence Interval Test-Retest Difference Lower Upper 250 -2.57 1.8882 4.4690 500 -.745 1.5303 3.4532 750 .664 1.2215 2.6818 1000 .520 1.1297 2.2258 1500 1.363 1.0908 2.5603 2000 2.088 1.2602 2.8162 3000 .654 .9482 2.1013 4000 -.563 .9703 2.4451 6000 1.090 1.4384 3.6259 Discussion Significant Changes in RECD Values In response to the first question posed in this investigation, overall there was a statistically significant change in RECD values between visit 1 and visit 2. This finding was expected and in agreement with Feigin and colleagues, (1989) data which suggested that between 1 kHz and 3 kHz, a systematic decrease in RECD values was observed in infants 0-12 months with increasing subject age. However, when RECD values at each visit were examined according to frequency, significant changes in RECD values were 54 only present at 3 frequencies (0.25, 1.5 and 2 kHz). Previous research conducted by Bagatto et al. (2002) has demonstrated that RECD values can be predicted with greater accuracy in the mid-frequency range from 750 to 3000 Hz. It is therefore not surprising that our data show a significant decrease in RECD values between visit 1 and visit 2, at 1500 and 2000 Hz. It is surprising however, that a significant change between visits was found at 250 Hz. The direction of the change at 250 Hz was positive meaning that the RECD values actually increased between visit 1 and visit 2. Previous research has consistently demonstrated that RECD values at 250 Hz show a high degree of variability (Bagatto et al., 2002; Feigin et al., 1989; Westwood & Bamford, 1995). This variability is thought to be the result of acoustic leakage during the RECD measurement (Bagatto et al., 2002). The 95% confidence intervals for Bagatto and colleagues (2002) RECD data were, on average, twice as large at 250 Hz when compared to mid-frequency values; therefore, the presence of low frequency acoustic leakage may provide an explanation for the significant increase found in our data at 250 Hz. As a result of infant movement during the RECD measurement, there was occasionally difficulty in maintaining an acoustic seal. Much previous research including that of Tharpe et al. (2001) has indicated increased infant movement at older ages as contributing to measurement error. Tharpe's data also indicate higher measurement error at 250 Hz. This issue is also a likely contributing factor to acoustic leakage at 250 Hz in the present data. In Westwood and Bamford's (1995) study, mean RECD values at 250 Hz were not reported due to a high degree of variability, which they suggest may have been caused by acoustic leakage. Another reason underlying the significant change in our data at 250 Hz may be related to experience in taking RECD measurements on infants. Bagatto et al., (2006) 55 strongly recommend that clinicians develop their skills regarding RECD measurements before attempting the procedure on infants. Troubleshooting the measurement is also particularly important for obtaining accurate results, evidenced by the fact that nearly one-third of the infants in Bagatto et al.'s (2006) study required repositioning of the probe-tube/eartip in order to achieve an accurate measurement. In comparison to Bagatto and colleague's (2002) cross-sectional data for foam tips, the magnitude of change in RECD values between visit 1 and visit 2 in this study is comparable with their findings. Bagatto et al. (2002) report no difference in average age-appropriate RECD values for foam tips between 1 and 2 months of age, except at 4 and 6 kHz where there is a difference of 1 dB. The RECD value at 2 months is therefore 1 dB less than the value at 1 month at 4 and 6 kHz. Corrected Age and RECD values at Visit 1 We also wanted to determine whether the infant's corrected age could predict the decrease in RECD values between visit 1 and 2 and whether RECD values at visit 2 are predictable from the RECD values obtained at the first visit. Corrected age was found to be a significant predictor of RECD values at 3 frequencies: 0.25, 3 and 4 kHz, although it did not account for a large amount of the variance, ranging from 10.8% to 14.4 %. One reason for this finding is likely the result of our small sample size. Only 14 babies were tested in this study, which made for a fairly small range of corrected age. The finding that corrected age was only a predictor of the RECD at 3 frequencies and subsequently only accounted for a small amount of the variance, is consistent with Bagatto and colleagues 2002 data. In their data r 2 values indicated at best a weak and in some cases non-existent relationships between subject age and measured RECD values. This finding points again 56 to the importance of individually measured RECD values. Significant variability has been demonstrated between subjects of the same age and Bagatto et al. (2005) suggest that average predicted RECDs can vary by more than 14 dB from actual measured RECDs. Such individual differences at each age range are consistent with the variability associated with infant growth in general. Much pediatric research has concluded that individuals grow by different rates during similar time frames (Lampl & Thompson, 2007). Growing is an individual process that is often characterized by nonlinear increments. This results in variations in the size relationships between infants of the same age, over short time intervals. Data from longitudinal studies of infants illustrate the poor performance of growth chart curves as representations of individual growth (Lampl & Thompson 2007). This finding helps explain why corrected age was not a significant predictor of RECD values across most frequencies. Infants grow differently and at different rates, and it can be assumed that this likely holds true for external and middle ear maturation as well. RECD values at visit 1 were found to significantly predict the RECD values at visit 2 at only 1 frequency: 3000 Hz. This lack of predictability may also be explained by nonlinear growth rates across individual infants and/or the magnitude of test-retest variability in this study. Predictability at 3000 Hz is consistent with previous research that indicates RECD measurements in the mid-frequency range (750 - 3000 Hz) are measured with greatest accuracy (Bagatto et al. 2002). 57 RECD Predictors The second question posed in this investigation was whether measures of static admittance and ear canal volume could be used to predict RECD values at any or all of the nine tested frequencies. Static admittance measures included 1000 Hz +200 Ya and SA 1000 Hz +200 Ytm, computed from the rectangular components Ba and Ga. Measures of ear canal volume were obtained from positive tail compensation tympanograms at 226 Hz B+ and 1000 Hz B+. Static admittance was the predictor that accounted for the most variance in the RECD measure at 5 frequencies, whereas E C V was the largest predictor at 3 frequencies. At the five frequencies where SA was a predictor the SA value computed from 1000 Hz +200 Y a accounted for more variance than the 1000 Hz +200 Ytm SA computed from the rectangular components. Reasons for this are unclear. It is not surprising that SA is a significant predictor of RECD values as the status of the middle ear has been proven to affect external ear canal acoustics (Martin et al., 1996). Increased mass and stiffness components due to an effusion in the middle ear will result in greater sound pressure level reflected by the tympanic membrane. This will lead to an increase in RECD values in the low and mid-frequencies (Martin et al., 1996). Holte and colleagues (1991) found that the admittance magnitude in the infant middle ear increased with age at frequencies above 226 Hz. Calandruccio, Fitzgerald and Prieve, (2006) also found that with the use of a 630 Hz probe tone, infants aged 11-19 weeks had lower middle ear admittance than infants aged 6 months to 2 years. Infants aged 20 -26 weeks also had lower admittance compared with 2-year-old infants. Similar findings were observed for probe tone frequencies of 226 and 1000 Hz. A l l age groups in the study had significantly lower middle ear admittance than adults at 226, 630 and 1000 Hz probe tones 58 (Calandruccio et a l , 2006). This would therefore suggest that as infants mature, admittance of the middle ear increases and consequently, the measured SPL levels in the ear canal become lower with increasing age. This trend of decreasing SPL levels in the infant ear was supported by the overall decrease in infant RECD values between visit 1 and visit 2, observed in the present study. Data from this study indicated that static admittance was a larger predictor of RECD values across frequencies when compared to E C V , with the exception of the lowest and highest frequencies (0.25 and 6 kHz), as well as 2000 Hz. Feigin et al. (1989) reported that E C V was, in general, negatively correlated with RECDs; however, the correlation coefficients from their infant data were not significant at any frequencies, leading to their conclusion that E C V does not appear to be a clinically useful predictor of RECD values for infants (Feigin et al., 1989). Data from this study supports this conclusion, although E C V was found to be a significant predictor of RECD values at 250, 2000 and 6000 Hz. At all 3 frequencies where E C V was a significant predictor, E C V at 1000 Hz B+ accounted for the most variance. Feigin and colleagues did not report the probe tone frequency that they used to obtain estimates of ear canal volume but i f 226 Hz was the only probe tone frequency used, this would help to explain the findings in the current study. Scatter plots indicate that at 250 Hz, ECV at 1000 Hz B+ (the larger predictor) actually shows a positive correlation with RECD, meaning that as E C V increases, the RECD also increases. This is not consistent with previous research and likely reflects measurement error at this frequency. Previous research has consistently demonstrated that RECD values in the lowest and highest frequencies are the most variable and are prone to increased measurement error (Bagatto et al., 2002, Sinclair et al., 1996; Westwood & 59 Bamford, 1995). At 2000 Hz and 6000 Hz, ECV at 1000 Hz B+ shows a more expected negative correlation with the RECD. As discussed in the introduction, significant error can result when obtaining ear canal volume readings in infants. Shanks & Lilly (1984) reported that negative tail compensation provides a more accurate estimate of ear canal volume when compared with positive tail compensation. Positive tail compensation will overestimate ear canal volume and consequently underestimate the admittance of the middle ear (Shanks & Lilly, 1984). On the other hand, as a result of the compliant nature of the infant ear canal, negative tail compensation is likely to result in collapse of the ear canal (Holte et al., 1990; Shahnaz et al., submitted). Currently a majority of studies are using positive tail compensation (Margolis & Goycoolea, 1993, Shahnaz et al., submitted); however, the limitations of ECV and SA measures must be kept in mind. In infant populations it is not yet known which measures provide the best estimate of SA and ECV. Consequently, a limitation of this study is that it is not certain whether we have quantified the true contributions ECV and SA may have with regards to the RECD measurement. These limitations may be circumvented by the use of a relatively new middle ear technique called wide band reflectance (WBR). When a sound is presented to the ear some of the sound is absorbed and transferred to the middle ear, and some is reflected back out of the ear canal (Shahnaz & Bork, 2006). WBR quantifies the absorbed and reflected sound energy in the external ear canal and measures middle-ear function at a constant pressure, across a significantly wider frequency range than tympanometry. With the use of constant pressure issues regarding the compliant nature of the infant ear canal are avoided and information is available across a wider range of frequencies. WBR data was collected 60 from infants in this study as part of the larger longitudinal study on external and middle ear development. W B R data will be analyzed and presented as part of this larger study conducted by Shahnaz. Test-retest Reliability The third question this study addressed was a quantification of the measurement variability that exists within a single test session and then a determination of whether the measured RECD changes over a 1-month interval are large enough to detect within the measurement error. If the mean difference in RECD values between visit 1 and visit 2 is less then the upper limit of the 95% confidence interval for the test-retest variability, then a change in the RECD values between visit 1 and 2 will be undetectable amidst the test-retest variability of the measure. Our data indicated that all of the mean RECD differences at each frequency, between visit 1 and visit 2, were less then the upper limit of the 95% confidence interval for test-retest variability. Test-retest reliability in this study ranged from a maximal difference between tests of 3.39 dB at 250 Hz, and a minimal difference of 1.08 dB at 4000 Hz. Although the magnitude of these differences is not huge, they were large enough to obscure the changes in RECD values in the first month of life. Test-retest differences reported in Westwood and Bamford's (1995) data are encouraging as they correspond well with the findings of this study. Data at 250 Hz was not reported in the Westwood and Bamford study because of the extremely high variability, but at 0.5 kHz, their test-retest differences were 3.1 dB (SD = 2.7) and their overall test-retest mean was 2.8 dB (SD = 2.7 dB). The overall test-retest mean in the present study, at 2.03 dB (SD = 2.28 dB), was slightly lower than the Westwood and Bamford test-retest variability. 61 Data from our study are also consistent with Bagatto and colleagues 2006 data. The overall mean difference between repeated RECD measurements in their study was 2.6ldB (SD = 2.61), which is comparable to our mean difference of 2.03 dB (SD = 2.28). Data from Bagatto et al., (2006) also indicates higher mean differences at 250 and 6000 Hz, consistent with the findings from the current study. Despite these small test re-test differences, the change in RECD values between visit 1 and visit 2 was even smaller than the test-retest differences, therefore making changes in RECD undetectable amidst test-retest variability. This finding may be in part due to the small sample size in this study but it may also suggest that that in the first month or so of life, the change in RECD values may not be large enough to warrant re-measuring. Clinical Implications Based on the findings from this study, it is known that the mean change in RECD values over a 1-month interval, corresponding to an infant's first month of life, are not large enough to detect amidst the test-re-retest variability of the RECD measure, despite the statistically-significant difference in RECD values. This suggests that in the clinic, audiologists may not need to re-measure the RECD during the infant's first month of life to account for changes in ear canal acoustics between initial screening tests and follow-up assessments. There are small changes in the measured SPL in infant canals over the first month of life but, despite the statistical significance of these changes, they are not likely to be detected within the expected test-retest variability of the RECD measurement. The potential for measurement error with the RECD can be high, so it is particularly important for clinicians to be comfortable in taking the measurement, and to minimize potential sources of error. For example, in our study, often infant movement 62 caused subsequent movement of the probe tube after it had already been positioned. This resulted in RECD measurements with significant high frequency roll off; however, this issue was recognized and the probe tube was re-positioned. Other sources of error include low frequency acoustic leakage. Often times in this study it was hard to find an appropriate sized immittance tip to couple to the infant's ear. In most cases the tip was too small and the next available size was too big for the infant's ear. By using a tip that was smaller then the infant's canal it was more difficult to obtain a good seal and there was a risk of low frequency sound leakage. One way to circumvent this was the use of otofirm cream around the edge of the immittance tip, which helped to seal the canal. Data from this study further enforce the finding that RECD values can vary significantly across infants of the same age. This points to the importance of using measured RECDs whenever possible as opposed to average age appropriate values which can result in considerable error. Directions for Future Research Findings from this study cannot be extended past the first month of life. It is not known how the RECD changes in individual infants over an extended period of time, creating a strong need for more longitudinal RECD data in the infant population. Another point to consider is that at the assessment stage, infant thresholds are usually measured with insert earphones but at intervention, hearing aid fittings are done with the infant's personal earmold. The result is two different coupling methods that can introduce variability into RECD measurements obtained from the same individual (Munro, 2004). The variability is the result of the earmold insertion depth, length of the sound bore, and differences in acoustic leakage (Munro, 2004). It is necessary to account for the 63 effects of the earmold acoustics on the resulting RECD values and at present, the BCEHP recommends that an RECD measurement be taken and applied to the prescriptive targets of the child's hearing aid each time a new earmold is made (BCEHP, 2006). More research is needed in order to quantify the additional variance brought on by the use of personal earmolds. Wide band reflectance measurements in combination with longitudinal RECD measures are another area for future research. WBR will help to provide more information about the contributions of the middle ear across a wider frequency range compared with tympanometry. Lastly, the question of the best measures of static admittance and ear canal volume in infants has also yet to be answered. Conclusion The purpose of this study was to address 3 main questions. 1) Is there a significant decrease in RECD values for individual infant over a 1 month period? If yes, is this decrease predictable from the infants corrected age, and are the RECD values at the time of the second measurement predictable from the RECD values obtained at the first measurement? Overall there was a statistically significant decrease in RECD values between visit 1 and visit 2. Examination of individual frequencies revealed significant changes in RECD values at 3 frequencies. At 250 Hz an increase in RECD values was observed and at 1500 Hz and 2000 Hz a decrease in RECD values was observed. Corrected age was found to be a significant predictor of RECD values at 3 frequencies: 250, 3000 and 4000 Hz, while RECD values at visit 1 were found to significantly predict the RECD values at visit 2 at one frequency, 3000 Hz. 64 2) Can measures of static admittance and ear canal volume be used to predict RECD values at any or all frequencies? Data from this study indicated that static admittance was a larger predictor of RECD values across frequencies when compared to ECV, with the exception of the lowest and highest frequencies (0.25 and 6 kHz). Static admittance was the predictor that accounted for the most variance in the RECD measure at 5 frequencies, whereas ECV was the largest predictor at 3 frequencies. 3) How much measurement variability exists within a single test session, and are measured RECD changes over a 1-month interval large enough to detect within the quantified measurement error? The overall test-retest mean in the present study was found to be 2.03 dB with a standard deviation of 2.28 dB. Despite the relatively small magnitude of this test-retest variability, it was still large enough to obscure the measured changes in RECD values between visit 1 and visit 2. In other words, the mean change in RECD values over a 1-month interval, corresponding to an infant's first month of life, is not large enough to detect amidst the test-re-retest variability of the RECD measure, despite the statistically-significant difference in RECD values. Consequently, re-measuring the RECD during the infant's first month of life to account for changes in ear canal acoustics between initial screening tests and follow-up assessments may not be necessary. Further research to investigate the longitudinal changes in RECD values within the infant population is needed. 65 References Anson, B. J., & Donaldson, J. A. (1981). Surgical anatomy of the temporal bone (3rd ed.). Philadelphia, Pennsylvania: W. B. Saunders. Bagatto, M. P. (2001). Optimizing your RECD measurements. The Hearing Journal, 54(9), 32-36. Bagatto, M. P., Moodie, S., Scollie, S. D., Seewald, R. C, Moodie, S., Pumford, J., et al. (2005). Clinical protocols for hearing instrument fitting in the desired sensation level method. Trends in Amplification, 9(4), 199-226. Bagatto, M. P., Scollie, S. D., Seewald, R. C, Moodie, S., & Hoover, B. M. (2002). Real-ear-to-coupler difference predictions as a function of age for two coupling procedures. Journal of the American Academy of Audiology, 13(8), 407-415. Bagatto, M. P., Seewald, R. C, Scollie, S. D., & Tharpe, A. M. (2006). Evaluation of a probe-tube insertion technique for measuring the real-ear-to-coupler difference (RECD) in young infants. Journal of the American Academy of Audiology, 17(8), 573-581. Baldwin, M. (2006). Choice of probe tone and classification of trace patterns in tympanometry undertaken in early infancy. International Journal of Audiology, 45(7), 417-427. 66 BC Early Hearing Program. (2006). Hearing equipment training and resource manual, sections 1,2 & 4. Unpublished manuscript. Calandruccio, L., Fitzgerald, T. S., & Prieve, B. A. (2006). Normative multifrequency tympanometry in Infants and toddlers. Journal of the American Academy of Audiology, 17(1), 470-480. Chan, J C Geisler,C D. (1990). Estimation of eardrum acoustic pressure and of ear canal length from remote points in the canal. [New York, etc.]: American Institute of Physics for the Acoustical Society of America. Dallos, P. (1973). The auditory periphery biophysics and physiology. New York: Academic Press. Dillion, H. (2001). Hearing aids. New York: Thieme. Dirks, D. D., Ahlstrom, J. B., & Eisenberg, L. S. (1994). Effects of probe tube insertion depth on real ear measurements. Otolaryngology - Head and Neck Surgery, 110(1), 64-74. Dirks, D. D., & Kincaid, G. E. (1987). Basic acoustic considerations of ear canal probe measurements. Ear & Hearing, 8(5) (SUPPLEMENT), 68s. 67 Eby, T. L., & Nadol, J. B. (1986). Postnatal growth of the human temporal bone implications for cochlear implants in children. Annals of Otology, Rhinology and laryngology, 95(4 Part 1), 356-364. Feigin, J. A., Kopun, J. G., Stelmachowicz, P. G., & Gorga, M. P. (1989). Probe-tube microphone measures of ear-canal sound pressure levels in infants and children. Ear and Hearing, 10(4), 254-258. Fikret-Pasa, S., & Revit, L. J. (1992). Individualized correction factors in the preselection of hearing aids. Journal of Speech & Hearing Research, 35(2), 384-400. Holte, L., Cavanaugh, Jr. R. M., & Margolis, R. H. (1990). Ear canal wall mobility and tympanometric shape in young infants. The Journal of Pediatrics, 117(\ Part 1), 77-80. Holte, L., Margolis, R. H., & Cavanaugh, Jr. R. M. (1991). Developmental changes in multifrequency tympanograms. Audiology, 30, 1-24. Hsu, G. S., Margolis, R. H., & Schachern, P. A. (2000). Development of the middle ear in neonatal chinchillas. I. birth to 14 days. Acta Oto-Laryngologica, 120(8), 922-932. Ikui, A., Sando, I., Haginomori, S., & Sudo, M. (2000). Postnatal development of the tympanic cavity: A computer-aided reconstruction and measurement study. Acta Oto-Laryngologica, 120(3), 375-379. 68 Jirsa, R. E., & Norris, T. W. (1978). Relationship of acoustic gain to aided threshold improvement in children. Journal of Speech and Hearing Disorders, 43, 348-352. Joint Committee on Infant Hearing. (2000). Year 2000 position statement: Principles and guidelines for early hearing detection and intervention programs. Pediatrics, 106(4), 798-817. Keefe, D. H., Bulen, J. C, Arehart, K. H., & Burns, E. M. (1993). Ear-canal impedance and reflection coefficient in human infants and adults. Journal of the Acoustical Society of America, 94(5), 2617-2638. Kruger, B. (1987). An update on the external ear canal resonance in infants and young children. Ear & Hearing, 8(6), 333-336. Lampl, M., & Thompson, A. L. (2006). Growth chart curves do not describe individual growth biology. American Journal of Human Biology, Special Issue: 2006 Alaska Symposium on Adaptation, Growth and Health. Margolis, R. H., & Shanks, J. E. (1991). Tympanometry: Basic principles and clinical applications. In W. F. Rintelmann (Ed.), (2nd ed., pp. 179-245). Massachusetts, USA: Allyn and Bacon. Martin, H. C, Munro, K. J., & Lam, M. C. (2001). Perforation of the tympanic membrane and its effect on the real-ear-to-coupler difference acoustic transform function. British Journal of Audiology, 35, 259-264. 69 Martin, H. C, Westwood, G. F. S., & Bamford, J. M. (1996). Real ear to coupler differences in children having otitis media with effusion. British Journal of Audiology, 50,71-78. McLellan, M. S., & Webb, C. H. (1957). Ear studies in the newborn infant. Journal of Pediatrics, 57(6), 672-677. Moodie, S., & Moodie, S. (2004). An approach to defining the fitting range of hearing instruments for children with severe-to-profound hearing loss. Retrieved January 11, 2006, from http://www.phonak.com/com proceedings chicago04 19.pdf Moodie, K. S., Seewald, R. C., & Sinclair, S. T. (1994). Procedure for predicting real-ear hearing aid performance in young children. American Journal of Audiology, 3, 23-31. Mora, R., Barbieri, M., Passali, G. C, Sovatzis, A., Mora, F., & Cordone, M. P. (2002). A preventive measure for otitis media in children with upper respiratory tract infections. International Journal of Pediatric Otorhinolaryngology, 63(2), 111-118. Munro, K. J. (2004). Update on RECD measures in children. A Sound Foundation through Early Amplification: Proceedings of the 3rd International Conference. Stafa, Switzerland. 71-89. 70 Munro, K. J., & Buttfield, L. M. (2005). Comparison of real-ear to coupler difference values in the right and left ears of adults using three earmold configurations. Ear & Hearing, 26(3), 290-298. Munro, K. J., & Davis, J. (2003). Deriving the real-ear SPL of audiometric data using the "coupler to dial difference" and the "real ear to coupler difference". Ear & Hearing, 24(2), 100-110. Northern, J. L., & Downs, M. P. (2002). In Met T. L. (Ed.), Hearing in children (5th ed.). USA: Lippincott Williams & Wilkins. Paparella, M. M., Shea, D., Meyerhoff, W. L., & Goycoolea, M. V. (1980). Silent otitis media. The Laryngoscope, 90, 1089-1098. Pumford, J., & Sinclair, S. (2006). Real-ear measurement: Basic terminology and procedures. Audiology Online, May 2001, March 25th, 2006 Rhodes, M. C, Margolis, R. H., Hirsch, J. E., & Napp, A. P. (1999). Hearing screening in the newborn intensive care nursery: Comparison of methods. Otolaryngology—head and neck surgery, 120(6), 799-808. Sagraves, R., Maish, W., & Kameshka, A. (1992). Update on otitis media, part 1. epidemiology and pathophysiology. American Pharmacy, NS32(\2), 27-31. 71 Saunders, J. C, Doant, D. E., & Cohen, Y. E. (1993). The contribution of middle-ear sound conduction to auditory development. Comparative Biochemistry & Physiology, 106A(\), 7-13. Seewald, R. C. (1991). Hearing aid output limiting considerations for children. Proceedings of the 1991 National Conference, Omaha, NE: Boys Town National Research Hospital Press. 19-35. Seewald, R. C, Moodie, K. S., Sinclair, S. T., & Scollie, S. D. (1999). Predictive validity of a procedure for pediatric hearing instrument fitting. American Journal of Audiology, 8(2), 143-152. Seewald, R. C, & Scollie, S. D. (1999). Infants are not average adults: Implications for audiometric testing. The Hearing Journal, 52(10), 65-72. Shahnaz, N. (08/25/05). Multi-frequency tympanometry immittance principles. Retrieved 06/21, 2007, from http://www.audiospeech.ubc.ca/school/facultv/navid/mv lab.htm Shahnaz, N., & Bork, K. (2006). Wideband reflectance norms for Caucasian and Chinese young adults. Ear and hearing, 27(6), 774-788. Shahnaz, N., Miranda, T., & Polka, L. ((submitted) Journal of the American Academy of Audiology). Multi-frequency tympanometry in neonatal intensive care unit and well babies. Unpublished manuscript. 72 Shanks, J. E., & Lilly, D. J. (1981). An evaluation of tympanometric estimates of ear canal volume. Journal of Speech and Hearing Research, 24(4), 557-566. Shanks, J. E., Stelmachowicz, P. G., Beauchaine, K. L., & Schulte, L. (1992). Equivalent ear canal volumes in children pre- and post-tympanostomy tube insertion. Journal of Speech and Hearing Research, 55(4), 936-941. Sinclair, S. T., Beauchaine, K. L., Moodie, K. S., Feigin, J. A., Seewald, R. C., & Stelmachowicz, P. G. (1996). Repeatability of a real-ear-to-coupler difference measurement as a function of age. American Journal of Audiology, 5(3), 52-56. Sininger, Y. S., & Cone-Wesson, B. (2002). Threshold prediction using auditory brainstem response and steady-state evoked potentials with infants and young children. In J. Katz (Ed.), Handbook of clinical audiology (5th ed., pp. 298-322). United States: Lippincott Williams & Wilkins. Stapells, D. R. (2000a). Frequency-specific evoked potential audiometry in infants. In Seewald R. C, Ed. A Sound Foundation through Early Amplification: Proceedings of an International Conference. Stafia, Switzerland. 13-32. Stapells, D. R. (2002(b)). The tone-evoked ABR: Why it's the measure of choice for young infants. The Hearing Journal, 55(11), 14-18. Sullivan, R. F. (1988). Probe tube microphone placement near the tympanic membrane. Hearing Instruments, 39(1), 43-44. 73 Tharpe, A. M. (2000). Service delivery for children with multiple impairments: How are we doing? A Sound Foundation through Early Amplification: Proceedings of an International Conference, Chicago, Illinois. 175-187. Tharpe, A. M., Sladen, D., Huta, H. M., & Rothpletz, A. M. (2001). Practical considerations of real-ear-to-coupler difference measures in infants. American Journal of Audiology, 70,41-49. Voss, S. E., & Herrmann, B. S. (2005). How does the sound pressure generated by circumaural, supra-aural, and insert earphones differ for adult and infant ears? Ear & Hearing, 26(6), 636-650. Wiley, T. L., & Stoppenbach, D. T. (2002). Basic principles of acoustic immittance measures. In J. Katz (Ed.), Handbook of clinical audiology (5th ed., pp. 159-174). Baltimore, Maryland: Lippincott Williams & Wilkins. Wilson, R. H., Shanks, J. E., & Kaplan, S. K. (1984). Tympanometric changes at 226 Hz and 678 Hz across 10 trials and for two directions of ear canal pressure change. Journal of Speech and Hearing Research, 27(2), 251. Yoshinaga-Itano, C. (2003). Early intervention after universal neonatal hearing screening: Impact on outcomes. Mental Retardation & Developmental Disabilities Research Reviews, 9(4), 252-266. Appendix A Individual RECD Values at Visit 1 Subject Ear 250 500 750 1000 1500 2000 3000 4000 6000 1 R 2 12 14 13 13 12 13 10 16 L -1 8 14 15 13 11 14 11 9 2 R -1 10 13 13 12 8 13 10 18 L -1 8 12 14 13 11 12 11 19 3 R -12 -4 3 8 9 10 14 11 13 L 0 7 11 11 9 11 12 9 15 4 R -4 5 10 11 9 13 14 10 11 L -4 4 5 8 8 11 14 9 19 5 R 999* 999* 999* 999* 999* 999* 999* 999* 999* L -5 3 9 11 12 11 13 11 15 6 R -1 7 12 16 19 17 15 10 15 L -7 2 8 10 9 13 14 10 12 7 R 9 12 15 13 13 16 17 13 17 L 7 13 15 17 18 16 16 13 19 8 R 0 7 12 13 13 14 16 11 21 L -2 7 12 13 11 9 14 9 17 9 R 8 14 17 18 16 14 10 13 17 L I 9 14 16 15 13 7 5 16 10 R 0 7 11 13 11 10 10 7 7 L 0 7 12 13 11 11 9 8 12 11 R -2 6 10 13 12 15 16 15 17 L 4 10 13 13 12 14 14 10 14 12 R 4 12 12 12 10 9 13 11 17 L 8 12 14 13 8 6 8 7 8 13 R -4 5 9 12 9 6 15 12 15 L 4 11 15 14 10 8 15 12 16 14 R 3 8 12 12 10 9 7 7 8 L I 7 11 12 10 9 8 9 11 *Missing values are coded as 999 L 7 11 14 14 10 8 10 10 14 11 R 8 13 11 12 9 8 13 10 17 L -1 4 9 10 7 9 12 9 15 12 R 9 11 14 14 10 8 11 11 12 L 2 7 8 8 5 4 8 10 12 13 R 0 7 10 12 13 8 12 12 12 L -1 6 8 11 9 9 17 11 12 14 R 7 10 11 11 8 10 13 12 10 L 6 10 13 12 8 10 13 10 10 * Missing values are coded as 999 Appendix B Individual RECD Values at Visit 2 Subject Ear 250 500 750 1000 1500 2000 3000 4000 6000 1 R 5 12 12 12 10 7 11 11 18 L 4 10 12 13 12 9 11 9 17 2 R 999* 999* 999* 999* 999* 999* 999* 999* 999* L 4 11 14 14 11 10 11 9 12 3 R -1 4 9 11 8 10 14 11 19 L I 6 8 9 6 8 12 8 9 4 R 4 10 14 15 13 12 9 • 6 14 L -15 -5 2 7 10 9 12 9 13 5 R -1 7 11 13 13 12 13 12 19 L 2 9 13 13 12 10 13 12 18 6 R 9 14 14 14 11 10 13 9 11 L 7 12 12 17 16 14 13 14 11 7 R 5 6 8 9 7 11 16 13 16 L 5 9 12 16 14 11 14 15 18 8 R -3 4 9 13 8 5 12 11 9 L 5 13 19 21 21 12 12 8 9 9 R 4 12 13 13 10 8 11 9 14 L 2 8 9 9 9 9 12 12 11 10 R 7 12 15 15 12 10 6 7 8 78 Appendix C Average Adult RECD Values from Moodie and Moodie (2004) Frequency (Hz) 250 500 750 1000 1500 2000 3000 4000 6000 RECD (dB) 4 4 4 4 4 4 7 10 9 79 Appendix D Mean Adult RECD Values using Foam Tips from Munro and Buttfield (2005) Frequency (Hz) Ear 250 500 750 1000 1500 2000 3000 4000 6000 RECD (dB) R 2.8 5.1 5.6 6.4 7.6 6.7 8.5 12.1 5.9 L 3.4 5.0 5.7 6.9 7.7 7.0 8.4 12.3 4.8 Appendix Clinical Research Ethics Board Approval 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0101116/manifest

Comment

Related Items