UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Inter-aural difference norms for wideband absorbance (WBA) : potential for identifying otosclerosis Schlagintweit, Martine 2018

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

Item Metadata

Download

Media
24-ubc_2019_february_schlagintweit_martine.pdf [ 7.07MB ]
Metadata
JSON: 24-1.0375802.json
JSON-LD: 24-1.0375802-ld.json
RDF/XML (Pretty): 24-1.0375802-rdf.xml
RDF/JSON: 24-1.0375802-rdf.json
Turtle: 24-1.0375802-turtle.txt
N-Triples: 24-1.0375802-rdf-ntriples.txt
Original Record: 24-1.0375802-source.json
Full Text
24-1.0375802-fulltext.txt
Citation
24-1.0375802.ris

Full Text

Inter-Aural Difference Norms for Wideband Absorbance (WBA): Potential for Identifying Otosclerosis  by  Martine Schlagintweit   B.A., The University of Victoria, 2014  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Audiology and Speech Sciences)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   December 2018  © Martine Schlagintweit, 2018  ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:  Inter-Aural Difference Norms for Wideband Absorbance (WBA): Potential for Identifying Otosclerosis  submitted by Martine Schlagintweit in partial fulfillment of the requirements for the degree of Master of Science  in Audiology and Speech Sciences  Examining Committee: Dr. Navid Shahnaz  Supervisor  Ms. Sasha Brown  Supervisory Committee Member  Dr. Tony Herdman  Supervisory Committee Member  University Examiner  University Examiner    iii  Abstract Purpose: Wideband acoustic immittance (WAI) is a field of noninvasive diagnostic measures that are used to evaluate the status of the middle ear. Wideband absorbance (WBA) is a WAI tool that distinguishes normally-hearing from pathological middle ears with a high level of diagnostic accuracy. This work aims to further improve the diagnostic efficiency of WBA, by providing a set of normative data which represents the typical inter-aural difference of WBA in adults of Caucasian, Asian and Mixed ethnicities.  Design: This retrospective analysis examined records of WBA obtained from normally-hearing adults (N=122; mean age= 24.16 yrs) using the Titan by Interacoustics immittance equipment. Norms were established for individual ear measures, and the inter-aural difference of WBA measured within-subjects. A mixed model ANOVA was used to analyze how normative WBA measures are affected by factors of ethnicity, gender, and test pressurization method. A comparison of the diagnostic efficacy of the two WBA measurements was conducted with an ROC analysis, using WBA records from 28 subjects with surgically confirmed otosclerosis as a clinical group.  Results: The ROC analysis confirmed that diagnostic accuracy of WBA individual ear and inter-aural difference measures was equal for the detection of otosclerosis; however, these measurements may provide complementary information regarding middle ear status. WBA norms based on individual ear data varied significantly depending on ethnicity, gender, and test pressurization method across frequencies. WBA norms based on inter-aural difference values only varied significantly with test pressurization method across frequencies, though the difference was small and likely does not have any clinical relevance, at least for the detection of otosclerosis. iv  Conclusion: WBA norms based on inter-aural difference values did not depend on population-based factors of ethnicity or gender; the same set of normative data may be used reliably across these populations. Parameterization of norms for individual ear WBA data is recommended based on ethnicity, gender and test pressurization method. Inter-aural difference norms for WBA may be instrumental in interpreting immittance results for individuals with unilateral conductive pathology, or for populations for which other forms of normative data are not available.    v  Lay Summary Wideband absorbance (WBA) is a test that involves directing sound at the eardrum, and measuring how much is absorbed by the middle ear. The purpose of this test is to differentiate ears with normal middle ear structures, and ears with abnormalities that might cause conductive hearing loss. There is a high level of individual variability in WBA measures, partially due to factors of ethnicity and gender, as well as the test pressurization method used during the measurement. Using norms specific to these parameters has resulted in better test accuracy of WBA. Researchers have therefore suggested that further investigation into normative measures of WBA is valuable for the development of this tool. This study looks at the sources of variability in two types of normative WBA measures; it also compares the diagnostic accuracy of WBA for detecting a common middle ear disease, otosclerosis, when either type of norm is used.  vi  Preface The following thesis project was based on ongoing research at the University of British Columbia’s Middle Ear Laboratory, conducted by Dr. Navid Shahnaz, Chamutal Efrat, Shahab Ravanparast, Ainsley Ma, Esther Rhi, Sukaina Jaffer, Rae Riddler and myself, Martine Schlagintweit. This project was completed in partial fulfillment of graduation requirements for the degree of Master of Science in the Faculty of Graduate and Postdoctoral Studies of the University of British Columbia. Identification of the research question and design of the research project, as well as analysis of all research related data, was a product of collaboration between the principal investigator, Dr. Navid Shahnaz, and myself, the co-investigator. Data was originally collected by Sukaina Jaffer, Rae Riddler, and Chamutal Efrat for three separate thesis projects completed in partial fulfillment of graduation requirements for the degree of Master of Science in the Faculty of Graduate and Postdoctoral Studies of the University of British Columbia, and for ongoing projects at UBC in the latter case. I was responsible for organizing existing data into a database that was amenable to analysis for the current objective, carrying out the statistical analysis under the guidance of Dr. Navid Shahnaz, and the writing of the manuscript. Ms. Sasha Brown and Dr. Tony Herdman were thesis committee members who provided invaluable input into the writing of this paper.  This study titled, Inter-Aural Difference Norms for Wideband Absorbance (WBA): Potential for Identifying Otosclerosis, was approved by the University of British Columbia Clinical Research Ethics Board (UBC CREB) as an independent study. This study’s associated UBC CREB certificate number is H18-01238. There are no sponsoring agencies or conflicts of interest to disclose.  vii  Table of Contents Abstract ......................................................................................................................................... iii	Lay Summary .................................................................................................................................v	Preface ........................................................................................................................................... vi	Table of Contents ........................................................................................................................ vii	List of Tables ..................................................................................................................................x	List of Figures ............................................................................................................................... xi	List of Abbreviations ................................................................................................................. xiv	Acknowledgments ........................................................................................................................xv	Chapter 1: Introduction ................................................................................................................1	1.1	 Evaluation of the Middle Ear .......................................................................................... 1	1.2	 Principles of Wideband Absorbance ............................................................................... 5	1.3	 Normative Data: Why it is Important to Identify Sources of Variability ....................... 6	1.3.1	 Variability due to Test Pressurization Method ....................................................... 8	1.3.2	 Variability due to Ethnicity and Gender ................................................................. 9	1.3.3	 Variability due to Maturation ................................................................................ 11	1.3.4	 Variability due to Test Ear .................................................................................... 13	1.3.5	 Summary of Sources of Variability for WBA in Normally-Hearing Populations 13	1.4	 Norms for the Between-Ear difference of WBA: A Less Variable Alternative? ......... 17	1.4.1	 An Inter-Aural Difference Measure of WBA Could be a Useful Variable to Identify Otosclerosis ............................................................................................................. 20	1.5	 The Rationale for the Current Study ............................................................................. 24	1.6	 Aims of the Current Investigation ................................................................................. 25	viii  1.7	 Hypotheses of the Current Investigation ....................................................................... 26	Chapter 2: Methods .....................................................................................................................27	2.1	 Data Sources ................................................................................................................. 27	2.2	 Subject Records ............................................................................................................ 27	2.2.1	 Inclusion Criteria .................................................................................................. 28	2.2.2	 Exclusion Criteria ................................................................................................. 29	2.2.3	 Ethnicity of Participants ........................................................................................ 29	2.3	 Instrumentation ............................................................................................................. 31	2.3.1	 Calibration ............................................................................................................. 34	2.4	 Statistical Analysis ........................................................................................................ 35	2.4.1	 Analyses of Variance ............................................................................................ 35	2.4.2	 Receiver Operating Characteristic Analyses ........................................................ 36	2.4.3	 The justification for Combining ANOVA and ROC Analyses ............................ 38	Chapter 3: Results ........................................................................................................................40	3.1	 Wideband Absorbance: Individual Ear Data ................................................................ 40	3.1.1	 Descriptive Statistics: Individual Ear Data ........................................................... 40	3.1.2	 ANOVA: Individual Ear Data .............................................................................. 42	3.2	 Wideband Absorbance: Inter-Aural Difference Data ................................................... 48	3.2.1	 Descriptive Statistics: Inter-Aural Difference Data .............................................. 48	3.2.2	 ANOVA: Inter-Aural Difference Data ................................................................. 49	3.3	 Receiver Operating Characteristic Curve Analysis ...................................................... 50	Chapter 4: Discussion ..................................................................................................................55	4.1	 Sources of Variability Affecting WBA Norms in this Study ....................................... 56	ix  4.1.1	 Individual Ear WBA Norms ................................................................................. 56	4.1.1.1	 Individual Ear WBA data Compared to Published Norms ............................... 56	4.1.1.2	 Sources of Variability for Individual Ear Norms .............................................. 59	4.1.1.2.1	 Ethnicity Effects .......................................................................................... 60	4.1.1.2.2	 Gender Effects ............................................................................................ 64	4.1.2	 Inter-Aural Difference WBA Norms .................................................................... 66	4.1.2.1	 Comparison of Inter-Aural Difference Data to Published Norms .................... 66	4.1.2.2	 Sources of Variability for Inter-Aural Difference Norms ................................. 68	4.2	 Test Performance of WBA with Either Style of Norm Applied ................................... 75	4.3	 Clinical implications of the Current Findings ............................................................... 83	4.4	 Study Limitations .......................................................................................................... 85	4.5	 The Direction of Future Research ................................................................................. 88	Conclusions ...................................................................................................................................91	Bibliography .................................................................................................................................92	Appendices ..................................................................................................................................105	Appendix A Supporting Documents for Methods .................................................................. 105	Appendix B Supporting Documents for Results ..................................................................... 106	Appendix C Supporting data for the Discussion Section ........................................................ 114	 x  List of Tables Table 1.1 Summary of studies that have investigated sources of variability in WBA obtained from normally-hearing adults. ...................................................................................................... 14	Table 1.2. Basic patterns of audiological test results obtained for otosclerotic ears. This table is recreated from Danesh, Shahnaz, and Hall (2018). ...................................................................... 22	Table 2.1. Demographic summary of normal and otosclerosis groups. ........................................ 30	Table 3.1. Descriptive summary of individual ear WBA data by test frequency, pressurization method, and ethnicity. ................................................................................................................... 41	Table 3.2. Descriptive summary of individual ear WBA data at each test frequency and pressurization method, organized by subject gender. ................................................................... 42	Table 3.3. Descriptive summary of inter-aural WBA difference data at each test frequency and pressurization method. .................................................................................................................. 48	Table 3.4. Summary of AUROCs of three best univariate predictors of otosclerosis from each type of normative data (individual ear data, inter-aural difference). ............................................ 53	Table 3.5. Summary of AUROCs of six best univariate predictors of otosclerosis across both types of normative data (individual ear data, inter-aural difference). ........................................... 54	Table 4.1. Pairwise comparisons of AUROCs of WBA measured with the Amb_WBA, Amb_3DT, and Peak_3DT pressurization methods from 1000-1250 Hz, applying difference norms............................................................................................................................................. 74	Table 4.2 Otosclerosis patient results which were identified as abnormal by ambient WBA results that were outside the 80% normative range for either individual ear measures or inter-aural difference measures, and a battery combining both measures. ............................................ 81	xi  List of Figures Figure 1.1 Wideband absorbance (2D and 3D) result for a healthy Caucasian adult middle ear, as measured by the Titan by Interacoustics immittance system. The grey shaded region in the right panel indicates the normative 80% range (from 10-90%); the red line indicates the individual’s result. ............................................................................................................................................... 6	Figure 2.1 Titan by Interacoustics Handheld Unit connected to a PC laptop computer. .............. 33	Figure 2.2 Measurement of a normal middle ear taken with the Wideband Tympanometry (3DT) tab of the Titan Suite (IMP440 module). ...................................................................................... 33	Figure 2.3 Right ear WBA of a normally-hearing ear with induced positive ME pressure, measured with the Titan IMP440 module wideband absorbance tab (left) and 3DT tab (right). . 34	Figure 3.1 Mean WBA magnitudes from 250-8000 Hz for Caucasian, Asian and Mixed ethnicity normally-hearing adults, as a function of frequency and pressurization method. Current Effect: [F(60, 7170)=2.501, p=.0000]. Vertical bars denote 95% confidence intervals. .......................... 45	Figure 3.2 Mean WBA magnitudes from 250-8000 Hz for Caucasian, Asian and Mixed ethnicity normally-hearing adults measured at Peak_3DT, Amb_3DT and Amb_WBA pressurization methods, as a function of frequency and ethnicity. Current Effect: [F(60, 7170)=2.501, p=.0000]. Vertical bars denote 95% confidence intervals. ............................................................................ 46	Figure 3.3 Mean WBA magnitudes from 250-8000 Hz for male and female normally hearing adults as a function of frequency. Current effect: [F(15, 3585)=10.833, p=.0000]. Vertical bars denote 95% confidence intervals. ................................................................................................. 47	Figure 3.4 Mean inter-aural WBA values from 250-8000 Hz for adults with normal hearing, measured with 3 pressurization methods. Vertical bars denote 95% confidence intervals. ......... 50	xii  Figure 3.5 Summary of AUROCs for all univariate predictors of otosclerosis (n=96). Numeric labels across the top pane denote the rank order of highest AUROC values from 1-6. Data labels provide the associated AUROC value. ......................................................................................... 51	Figure 3.6 ROC comparisons of univariate predictors with the three highest AUROC values as a function of test frequency, pressurization method, and normative data type (individual ear vs. inter-aural difference). .................................................................................................................. 52	Figure 3.7 ROC comparisons of univariate predictors with the six highest AUROC values as a function of test frequency, pressurization method, and normative data type (individual ear, inter-aural difference). ........................................................................................................................... 54	Figure 4.1 Comparison of mean ±SD ambient individual ear WBA data to published norms. .... 58	Figure 4.2 Comparison of ambient WBA data from the current study to norms published by Polat et al. (2015). The black dashed line and grey shaded region indicate the mean and 80% range of the current study, respectively. The dashed and solid blue lines indicate the mean and 80% range published by Polat et al. (2015), respectively. .............................................................................. 59	Figure 4.3 Mean ± SD of inter-aural difference data in this study (not absolute values). ............ 68	Figure 4.4 Mean individual ear WBA data as a function of the pressurization method and frequency (Hz). Vertical bars denote 95% confidence intervals. ................................................. 69	Figure 4.5 ROC comparison of WBA measured with the Amb_WBA, Amb_3DT, and Peak_3DT pressurization methods from 1000-1250 Hz. ................................................................................ 73	Figure 4.6 Mean and individual ear ambient WBA results for the otosclerosis group (n ears=30) plotted with the normative data. The grey shaded region shows the 80% normative range. ........ 82	xiii  Figure 4.7 Mean and individual absolute inter-aural difference ambient WBA results for the otosclerosis group (n=28) plotted with the normative data. The grey shaded region shows the 80% normative range. ................................................................................................................... 82	xiv  List of Abbreviations ABG – Air Bone Gap ANOVA – Analysis of Variance AUROC – Area under the receiver operating curve  CHL – Conductive Hearing Loss ME – Middle Ear  MEMR – Middle Ear Muscle Reflex MFT – Multi-frequency Tympanometry  peSPL – Peak equivalent SPL RF – Resonance Frequency  ROC – Receiver Operating Characteristic Curve  SCD – Superior Canal Dehiscence SPL – Sound Pressure Level TM – Tympanic Membrane TPP – Tympanometric Peak Pressure  Tukey HSD – Tukey-Kramer Honestly Significant Difference  WAI – Wideband Acoustic Immittance  WBA – Wideband Absorbance WBR – Wideband Reflectance xv  Acknowledgments Thanks are owed to my thesis supervisor, Dr. Navid Shahnaz, Ph.D., AUD (C), whose probing questions and endless knowledge of acoustic immittance supported my learning throughout this project. His wisdom, enthusiasm for the field, and generosity with his time have made the journey to completion of this thesis both inspiring and enjoyable. I extend gratitude to my committee members Dr. Sasha Brown, Ph.D., AUD(C) and Dr. Tony Herdman, Ph.D., for their support and thoughtful suggestions throughout this process. Special recognition is due to my colleagues at the UBC Middle Ear Lab, Chamutal Efrat, Samaneh Yekta, Esther Rhi, and Shahab Ravanparast for their encouragement, support, and patience in carrying out this endeavor. Thank you also to my classmates at the UBC School of Audiology and Speech Sciences for their encouragement.  Finally, I offer my heartfelt thanks to my family members, who have all provided unique, and substantial, support throughout my years of education. It is due to the moral and financial support of my parents, Robert and Emily Schlagintweit, that my education was made possible. Encouragement from my grandparents, Johannes and Puck Kerkhoven, has kept me enthusiastic and able to enjoy life’s distractions. And last, but never least, to my sister, Anna, and her loving partner (also Anna): your generosity and kindness have allowed me to find a home in this city through the past two years, thank you for your endless support.    1 Chapter 1: Introduction Wideband acoustic immittance (WAI) is a middle ear analysis technique that is starting to enter clinical practice. This method uses a broadband stimulus as opposed to single or multiple pure tones to evaluate the reflectance, transmittance, and absorbance of the middle ear structures (Hunter & Shahnaz, 2014). It is valued for its high degree of test-retest stability (Sun, 2016), and relative insensitivity to factors such as probe tip placement in the ear canal (Voss & Allen, 1994). Researchers have demonstrated the clinical utility of WAI parameters in distinguishing between normally-functioning middle ear systems, and those with a variety of abnormalities (Nakajima et al., 2013).  It is important to consider the factors that may impact measures of middle ear function when interpreting the results of acoustic immittance tests. The current paper contributes a set of normative data obtained from healthy adults, for individual ear data and the inter-aural difference of wideband absorbance (WBA) measures. Both variables were analyzed to determine the influence of ethnicity, gender, frequency and test pressurization method on normative values, to determine whether parameterization for these factors is necessary. A comparison of the diagnostic accuracy of the two WBA measures for the identification of otosclerosis, was also conducted to establish the clinical utility of difference norms.  1.1 Evaluation of the Middle Ear  The tympanic membrane (TM), ossicles, and middle ear cavity are essential for the transduction of sound energy in its journey from the environment to the inner ear. Sound waves propagating through the air as regions of condensation and rarefaction strike the TM, vibrating the connected ossicles. The rocking of the smallest of the ossicles, the stapes, in the oval window   2 of the cochlea sets cochlear fluid into motion, ultimately leading to the biochemical and neural mechanisms that contribute to the perception of sound (Martin & Clark, 2000).  The middle ear (ME) compensates for an impedance mismatch between air and cochlear fluid by intensifying the incoming signal in a frequency dependent manner. As such, changes to the mechanical properties of the structures of the middle ear affect the transduction of sound, resulting in frequency-specific attenuation, or filtering (Norrix et al., 2013). Natural differences in the frequency-specific transmission of sound through the ME arise from between-subject and population-based factors such as ethnicity, gender, and maturation (Shahnaz & Davies, 2006; Mazlan et al., 2015; Beers et al., 2010). Body size and composition likely underlie these sources of variability (Shahnaz & Davies, 2006). Alternatively, middle ear pathology or injury to the structures of the conductive pathway cause abnormal filtering of acoustic signals; often resulting in conductive hearing loss.  Acoustic immittance testing is the part of the audiological test battery that evaluates the status of the middle ear. Tympanometry is a well-established acoustic immittance test; it is used to measure the admittance of the middle ear, as a function of pressure in the outer ear canal (ANSI, 1987). Typically, this measure is conducted using a single 226-Hz probe tone; a pliable ear tip seals the outer ear canal, and sound pressure level (dB SPL) is measured while pressure is varied in the canal. A reduction of SPL measured in the canal, which coincides with the equalization of pressure in the ear canal with that of the middle ear cavity, indicates maximum admittance of the middle ear. In a normally hearing ear, maximum admittance occurs at approximately ambient pressure (0 daPa). In ears with abnormally increased stiffness, due to pathologies such as otosclerosis, little to no change in admittance is observed across the pressure   3 sweep. Ears in which stiffness has been abnormally decreased, as in the case of ossicular discontinuity, present with elevated admittance (Hunter & Shahnaz, 2014). Tympanometry has been used reliably to differentiate some pathologies of the middle ear, however, it falls short in its ability to detect pathology of the ossicular chain (Liden et al., 1970). The insensitivity is believed in part to reflect the dominance of the status of the TM in these measures; it is especially notable for conditions in which the TM is intact, and the ME cavity is aerated (Nakajima et al., 2012). For example, Feldman (1974) demonstrated that tympanometry failed to detect otosclerosis in a sample of ears with concurrent healed TM perforations; high admittance tympanograms were consistent with a history of perforation and did not show any sign of the increased stiffness due to stapes fixation.  Alternatively, multi-frequency tympanometry (MFT) provides superior diagnostic accuracy over conventional 226-Hz tympanograms (Colletti, 1974; Shahnaz & Polka, 1997; Wada, Koike & Kobayashi, 1998). Resonant frequency (RF), a parameter estimated from these measures, gives an indication of the mass and stiffness characteristics of the middle ear. For example, abnormally stiffened middle ears exhibit an elevated RF, whereas middle ears affected by mass loading pathologies (e.g. ossicular discontinuity) are characterized by subnormal values (Colletti, 1974; Wada, Koike & Kobayashi, 1998). In this manner, RF can be used to distinguish normally-functioning from abnormal ME systems, while also giving an indication as to the etiology of the problem.  Despite encouraging results, MFT had several issues which hindered its acceptance into clinical use. The likelihood of inducing standing waves in the ear canal limited measurable frequencies from 200-2000 Hz - a problematic limitation given that RF in abnormally stiffened ears may exceed this range (Margolis, Saly & Keefe, 1999; Hunter & Shahnaz, 2014).   4 Measurements were also highly susceptible to probe tip placement in the ear canal, necessitating a precise and deep fit (Hunter & Shahnaz, 2014). Moreover, susceptance tympanograms obtained with MFT required considerable familiarity to interpret, which discouraged clinical audiologists from adopting this technology. In a survey conducted by Emanuel, Henson and Knapp (2012) seventy-seven percent of audiologists reported they never attempt multi-frequency tympanometry, primarily due to a lack of confidence in interpreting results. In contrast to obtaining a tympanogram with a single or multiple frequency pure tones, the broadband nature of wideband acoustic immittance (WAI) stimuli allow for evaluation of the ME across a wide frequency spectrum (250-8000 Hz) (Shahnaz et al., 2014). Comparability of WAI results to other broadband measures, such as behavioral audiometry, otoacoustic emissions, and auditory brainstem response make this technique useful as both a stand-alone measure as well as a diagnostic cross-check (Hunter et al., 2010). Precise probe tip placement in the ear canal is not essential, provided a good seal is achieved; Voss and Allen (1994) have demonstrated WAI parameters measured at the probe tip are equivalent to measurements taken at the TM. Using commercially available WAI instrumentation it is possible to simultaneously measure additional acoustic immittance functions, including single frequency tympanometry, multi-frequency tympanometry, middle ear muscle reflexes (MEMRs), and otoacoustic emissions (Hunter & Shahnaz, 2014). In addition to these practical benefits, WAI measures have been shown to have improved diagnostic accuracy, over MFT and single frequency tympanometry, for the detection and differentiation of middle ear disorders (Nakajima et al., 2013; Shahnaz et al., 2009).  A WAI protocol typically evaluates aural acoustic transfer functions of admittance, impedance, or reflectance (Keefe et al., 2017). Commonly analyzed parameters include   5 wideband absorbance (WBA), wideband reflectance (WBR), transmittance, acoustic impedance (Z), and group delay at either ambient or tympanic peak pressure (TPP) (Hunter & Shahnaz, 2014). The current paper analyzes WAI measures in terms of wideband absorbance (WBA).  1.2 Principles of Wideband Absorbance Wideband absorbance (WBA) is the proportion of acoustic energy that is absorbed by the middle ear when a calibrated broadband stimulus is directed toward the TM in a sealed ear canal. This dimensionless quantity ranges between one, where all the energy from the stimulus was absorbed, and zero, where all the energy was reflected (Liu et al., 2008). WBA is plotted as a function of frequency; the typical bandpass shape of WBA in normally-hearing adult ears is a dual peaked pattern, with local maxima around 1000 Hz and 4000 Hz (Figure 1.1). Values tend to increase steeply as a function of frequency to the first maximum, and decline moderately with increasing frequency above 4000 Hz (Keefe et al., 1993; Polat et al., 2015; Rosowski et al., 2012; Shahnaz & Bork, 2006; Voss & Allen, 1994). It is common to observe a slight ‘dip’ between 1000-4000 Hz. WBA is related to a complementary measurement, wideband reflectance (WBR), by conservation of energy (WBA = 1- WBR). Liu et al. (2008) state that although both WBA and WBR are equally representative of the mechanical properties of the ME, WBA lends itself to more intuitive interpretation due to its similarity with traditional tympanometric functions. Middle ear researchers at the Eriksholm Workshop on Wideband Acoustic Absorbance of the Middle Ear have also identified WBA as the preferred variable of these two quantities (Feeney et al., 2013). Consequently, this study reports values in terms of WBA.    6  Figure 1.1 Wideband absorbance (2D and 3D) result for a healthy Caucasian adult middle ear, as measured by the Titan by Interacoustics immittance system. The grey shaded region in the right panel indicates the normative 80% range (from 10-90%); the red line indicates the individual’s result.  1.3 Normative Data: Why it is Important to Identify Sources of Variability  The interpretation of audiological test results usually involves a comparison of values obtained from the patient to a range of values expected from normally-hearing individuals. This process describes a fundamental concept in evidence-based practice – the use of normative data. By knowing the range of values that are expected from normally-functioning individuals, clinicians can characterize pathological deviations from these norms. The knowledge of what is ‘normal’ thereby helps to identify results that represent ‘diseased’ or ‘injured’ states (Turner, Robinette & Bauch, 1999). Normal ranges obtained from one subpopulation may not be equivalent to the range of values obtained from another. This is the case for differences between groups relating to gender, ethnicity, and age for WBA, as well as for the same population measured with different   7 pressurization methods (Shahnaz, Feeney & Schairer, 2013). The use of separate norms relating to these variables is often justified, as more precise representation of WBA characteristics relating to these factors can improve the diagnostic accuracy of this measure.  A clinical study in Hong Kong reported a false positive rate of up to 48% when American Speech and Hearing Association (1990) specified tympanometry norms were used, highlighting the importance of using population-specific normative data (Wan & Wong, 2002). Individuals were incorrectly identified as abnormal, when in fact their results were being compared to normative values that were not representative of the typical acoustic immittance profile of their ethnicity. By contrast, Shahnaz and Bork (2006) demonstrated that diagnostic accuracy for the detection of otosclerosis with WBR (1-WBA) was improved when ethnicity-specific norms were used. Corresponding research by Jaffer (2016) showed that WBA at 800 Hz detected otosclerosis with greater diagnostic accuracy when instrument-specific norms were considered, demonstrating that norms specific to WAI equipment may also be beneficial.   The ability of a test to distinguish between results from normally-functioning and diseased populations is related to the degree of separation between the distributions of each population’s results, wherein greater separation provides better accuracy of the test (Turner, Robinette & Bauch, 1999). The use of separate sets of normative data for distinct subpopulations within normal groups can produce a narrowing of the normal range because results are more homogeneous within these subgroups. This so-called narrowing can lead to greater separation between the distributions for each group, which is evidenced by improved diagnostic accuracy with the application of the homogeneous norm (Turner, Robinette & Bauch, 1999). This occurrence was evidenced by the WAI research on otosclerosis (e.g. Shahnaz & Bork, 2006; Jaffer, 2016).    8  In light of encouraging findings, investigations into parameterization of norms for WAI are ongoing. At the 2013 Eriksholm Workshop on Wideband Acoustic Absorbance of the Middle Ear, researchers recommended that the collection and sharing of normative WAI data was a critical research priority, as the existing norms were largely published post-hoc to evaluations of clinical populations (Feeney et al., 2013). This recommendation has been addressed with the creation of the Wideband Acoustic Immittance Database (WAI-Database), an initiative of Dr. Susan Voss’ hosted by Smith’s College. As stated by Voss (2014, Jun 5), the WAI-Database is an online repository of published WAI norms, the purpose of which is to “determine how normative WAI measures depend on the parameters: instrumentation, age, gender, and race.” As well as “determine whether or not normative data would need to be parameterized for any of these quantities.” The following sections describe how sources of variability affect WBA results for populations of normally-hearing individuals. Table 1.1 provides a summary of this portion of the literature review. 1.3.1 Variability due to Test Pressurization Method As shown in Figure 1.1, WBA can be measured under pressurized (left) or ambient pressure (right) conditions. Measurement of WBA taken at compensated pressures are referred to as wideband tympanometry (WBT). By sweeping pressure in the ear canal through extreme positive and negative pressures, the TM and ossicular chain are put into positions of minimum and maximum mobility. At extreme pressures, the TM and ossicular chain are stiffened and are characterized by low WBA values. At tympanic peak pressure (TPP), a condition in which the pressure in the ear canal approximates pressure in the ME cavity, the ME shows its highest WBA, because the TM is at its maximum mobility (Hunter & Shahnaz, 2014). Readers are   9 referred to Riddler’s (2017) manuscript for an in-depth discussion of the impacts of pressurization on the structures of the ME, and the corresponding outcome on WAI measurements.   It is important to note herein, however, that slight differences are observed between WBA measures taken at ambient and TPP conditions for normally-hearing subjects, even if their middle ear pressure does not vary significantly from 0 daPa. As described by Margolis, Saly and Keefe in their 1999 normative data study, “Compared to the tympanometric peak pressure condition, average [absorbance] at ambient ear canal pressure is [lower] at low frequencies, [higher] just above the resonant frequency, and unchanged at high frequencies.” Furthermore, the absorbance peak usually shifts to a slightly higher frequency when measured at TPP.  Researchers Sanford, Hunter, Feeney and Nakajima (2013) conducted a review of the literature surrounding WAI measured at ambient and TPP. As discussed in their paper, research findings suggest that the high level of diagnostic accuracy of WAI measures appears to create a ceiling effect, wherein tests using either style of pressurization method cannot out-perform the other. However, they note that the differences between studies, such as the clinical populations analyzed, instrumentation, and the age of the subjects, reveal benefits of the application of either style of WAI pressurization particular to these parameters/populations. For this reason, the collection of normative WAI data at both ambient and TPP is recommended for further research.  1.3.2 Variability due to Ethnicity and Gender The acoustic transfer of the middle ear, as measured by WBA, appears to be influenced by ethnicity and gender effects relating to anatomical and physiological differences of the conductive pathway between these groups (Hunter & Shahnaz, 2014; Shahnaz, 2010). In a normative data study investigating population-based effects on mean WBR (1-WBA), Shahnaz   10 and Bork (2006) found significant frequency-dependent differences between Caucasian and Chinese young adult participants (aged 18-35 yrs). The researchers found that the Chinese group had higher mean WBR values (lower WBA) than the Caucasian group in the low frequencies. Conversely, Caucasian subjects had higher mean WBR (lower WBA) in the high frequencies. These results have since been verified in other studies, which also found interactions between ethnicity and frequency for mean WBA with other groups of adult subjects (Jaffer, 2016; Kenney, 2011; Riddler, 2017; Shaw, 2009), as well as with pediatric subjects of varying ethnicities (Abbott, 2018; Aithal et al., 2014; Beers et al., 2010).  To determine whether the use of ethnicity-specific norms is justified for WAI measures, Shahnaz and Bork (2006) compared the diagnostic accuracy of WBR when homogeneous versus composite norms were applied. Overall test performance was improved for the detection of otosclerosis when results from diseased ears were compared to norms that were representative of the patient’s ethnic group. Furthermore, successful identification (true positive rate) for the detection of otosclerosis increased (false positive rate decreased) with the application of ethnicity-specific norms for the Chinese group. This research thereby rationalizes the use of ethnically homogeneous sets of normative data for WBA (Shahnaz & Bork, 2006).  Middle ear researchers have also found significant gender effects in mean WBA measures. Combined data from Shahnaz and Bork (2006) and Shaw (2009) revealed an interaction between frequency and gender; female subjects had higher WBA values in the mid-frequency range than males (Shahnaz, Feeney & Schairer, 2013). Jaffer (2016) found females to have higher mean WBA values than males in the high-frequency range (4000-5000 Hz). Mazlan et al. (2015) and Polat et al. (2015) found males to have higher mean WBA values than females in the low frequencies, whereas females had significantly higher mean WBA values in the high   11 frequencies. Research into the diagnostic performance of WAI with the application of gender -specific norms is lacking, however, most researchers recommend separate sets of normative data for these populations.  The population-based variability of WBA could reflect variation in body size and composition between these groups. Males across ethnicities generally have larger body size indices than females, while across genders individuals of Caucasian descent have a higher mean height and weight than Chinese populations (Bell, Adair & Popkin, 2002); Shahnaz and Davies (2006) demonstrated that these body size differences are reflected in ear canal volume (Shahnaz, Feeney & Schairer, 2013). Shahnaz and Davies (2006) and Shahnaz and Bork (2006) observed that an increase in middle ear compliance coincides with an increase in overall body size; this relationship is expressed in the relatively higher low-frequency WBA for groups with larger mean body size as indexed by height, weight, and equivalent ear canal volume (i.e. Caucasians).  1.3.3 Variability due to Maturation  Significant maturational changes to WBA have been documented over the first six months of life, leading to recommendations of separate norms for neonates, and infants at one, and six to 15 months of age (Hunter, 2016). WBA values also vary between adult and pediatric populations. Beers et al. (2010) compared WBA results from school-aged children from their own study to adult data from Shahnaz and Bork (2006) and found that WBA differed between the age groups in a frequency dependent manner. Adults across ethnicities had lower WBA than children in the mid frequencies (2500 - 5000 Hz), whereas Caucasian children had lower absorbance than adults in the low frequencies (315 – 1250 Hz). Maturational differences in WBA between pediatric and adult populations are consistent with data relating to body size and acoustic immittance measures (Shahnaz, Feeney & Schairer, 2013).    12 Maturational changes to WBA throughout the adult lifespan are less clear. In a longitudinal study with adult participants, Feeney et al. (2014) observed slightly, yet significantly, higher low-frequency WBA in middle-aged adults (30-39 yrs) compared to older (40-49 yrs) and younger adults (20-29 yrs) at baseline. No significant WBA differences were observed between younger adults and older adult groups at baseline. Age-related changes to WBA were not, however, significant when the same subject was measured repeatedly at different time-points. That is to say, maturation had a significant effect as a between-subject factor in the cross-sectional design, however did not exert a significant effect as a within-subject measure in the longitudinal design. This was perhaps a consequence of the relatively short intervals between measurements (5 yrs at maximum), though this is conjecture on the part of the researcher.   In a cross-sectional study that examined the effect of age on WBA values, Mazlan et al. (2015) observed higher mean WBA values in the young adult group in the high frequencies (~2000-5000 Hz) compared to older and middle-aged adults. This pattern was reversed in the low frequencies; WBA was significantly lower in the young adult group than the middle-aged group at from 280 to 500 Hz. These findings are in direct contrast to expected WBA trends with increasing age, as Ruah et al. (1991) have shown that anatomical changes to the ME through adulthood trend toward a progressive increase in rigidity.  Conflicting data regarding maturational changes of WBA through the adult lifespan prevent generalizations about specific age groups within adult populations. As evidence in this area is lacking, separate norms are usually recommended for use between pediatric and adult groups, but not for adults belonging to different age categories.    13 1.3.4 Variability due to Test Ear  Inconsistent data regarding the frequencies at which differences due to test ear exert their effects, and small differences across frequencies, imply that this is likely not an important factor for WAI measures (Werner et al., 2010). Several studies have found that the right ear is slightly, but significantly, more stiff than the left, as indexed by higher WBR values (Feeney & Sanford, 2004; Rosowski et al., 2012; Werner et al., 2010), whereas others have reported the opposite (Feeney et al., 2014). Feeney and Sanford (2004) reported that between-ear differences were significant only at 6350 Hz in a group of older adults, whereas no significant effect of test ear was observed for their group of young adult subjects. Rosowski et al. (2012) reported that right ear WBR significantly exceeded left ear values at 300 Hz. Werner et al., (2010) found significant WBR differences between the ears and stated that on average the right ear was marginally stiffer than the left. The conclusion of Werner et al.’s (2010) study was that, “the observed difference between left and right ears in WBR is small and likely to be of little consequence to establishing norms for WBR.” 1.3.5 Summary of Sources of Variability for WBA in Normally-Hearing Populations  Table 1.1 summarizes the studies that have evaluated sources of variability which affect WBA measured from normally-hearing, adult subjects; papers are organized alphabetically within a year of publication. Several studies reported on WBR as opposed to WBA. This table presents the results of all the studies in terms of WBA for clarity.   14 Table 1.1 Summary of studies that have investigated sources of variability in WBA obtained from normally-hearing adults. Study Methods Subjects Summary of Findings Feeney & Sanford (2004) WBA measured at ambient pressure with custom research immittance equipment, Etymotic ER-1 inserts, and ER 7-C Microphone.  70 normally-hearing adults divided into two age groups (18-28 and 60-85 yrs).  Left ear WBA was higher than right ear at 6350 Hz, for older adults only.  The older age group had higher mean WBA from 794-2000 Hz, and lower at 4000 Hz. Females had lower WBA from 794-1000 Hz, but higher WBA than males at 5040 Hz in the younger group. Shahnaz & Bork (2006) WBA measured at ambient pressure with the Mimosa Acoustics RMS-system v. 4.0.4.4.  126 normally-hearing adults aged 18-32 yrs.  WBA was influenced by ethnicity; the Chinese group had a lower mean WBA from 469-1500 Hz than the Caucasian group, however, had higher mean WBA from 3891-6000 Hz. The effects of gender and ear were not significant. Liu et al. (2008) WBA measured at ambient and TPP with custom research immittance equipment with the Titan, AT235 probe assembly. 48 normally-hearing adults aged 33.8±10 yrs.  Below 2000 Hz measurements of WBA at TPP were higher than ambient values; at about 4000 Hz WBA at TPP was slightly lower than values taken at ambient pressure.  Werner et al. (2010) WBA measured at ambient pressure with custom research immittance equipment, Etymotic ER-1 inserts, and an ER 7-C Microphone.  210 adults aged 18-30 yrs, 198 infants aged 2-3 months, and 260 infants aged 5-9 months; all subjects had normal hearing. Gender effects were not significant for adults. Left ear WBA was higher than right ear values across frequencies. WBA test-retest correlations were higher within subjects than across subjects. Within-subject right-left cross-correlation of WBA was significantly higher (0.83) than the same correlation across subjects (0.73). Rosowski et al. (2012) WBA was measured at ambient pressure with the Mimosa Acoustics HearID system, software v 3.4.45.1. 29 normally-hearing adults aged 22-64 yrs.  Left ear WBA was higher than right ear values at 300 Hz. Females had lower WBA than males below 2000 Hz, and higher WBA from 3000-6000 Hz; these differences reached significance at 4000 Hz only. Slight age effects at 1000 Hz were significant and implied that WBA increases with age. Feeney et al. (2014) WBA measured at ambient pressure with custom research immittance equipment with the ER-10C probe assembly. 112 normally-hearing adults aged 20-49 yrs, divided into 3 age groups (20-29 yrs, 30-39 yrs, 40-49 yrs) Left ear WBA was lower than right ear WBA for baseline and longitudinal measures. The effect of age was significant at baseline; middle-aged adults had higher mean WBA than young and old adults from 500 to 1600 Hz. Age effects were not significant within-subjects between time-points. Females had lower mean WBA than males below 3000 Hz.    15 Study Methods Subjects Summary of Findings Mazlan et al. (2015) WBA measured at ambient pressure with an Interacoustics prototype immittance system combined with ReflWin software (v 3.2.1).  101 normally-hearing adults aged 20-82, divided into three age groups (20-38 yrs, 40-64 yrs, 65-82 yrs).  The bandpass shape of WBA varied with age; young adults showed a single peaked pattern, while middle and older adults had a dual peaked pattern. Between 280-500 Hz WBA of young adults was lower than middle adults, though their values were higher from 2000-5040 Hz. Young adults had significantly higher WBA from 2250-5000 Hz than the older adult group.  Across age groups females had lower WBA from 280-790 Hz, and higher WBA from 2830-4490Hz than males. When age group was analyzed as a between-subject variable, middle-aged women were demonstrated to have significantly lower WBA than males from 500 to 790 Hz.  Polat et al. (2015) WBA measured at ambient and TPP with the Titan by Interacoustics (v 3.1).  109 normally-hearing Turkish adults aged 18.3-26.2 yrs.  The effect of gender was significant; in the bandwidth of 3100 to 6900 Hz the female participant group had higher mean WBA compared to the male participant group. Feeney et al. (2016) WBA measured at ambient and TPP with prototypical Interacoustics immittance software combined with an Interacoustics AT235 Tympanometer.  33 normally-hearing adults aged 19-46 yrs.  WBA varied with pressurization method. WBA was higher when measured at TPP using upswept and down-swept pressurization than at ambient below 2200 Hz and 2500 Hz, respectively. Ambient WBA was greater than down-swept TPP measures from 3600-5700 Hz, and upswept TPP measures at frequencies >3600 Hz. WBA measured at upswept TPP were significantly greater than down-swept measures from 250-1410 Hz but significantly lower at frequencies >4000 Hz.  Jaffer (2016) WBA measured at ambient and TPP with the ReflWin and Titan systems by Interacoustics and the Otostat and HearID systems by Mimosa Acoustics.  80 normally-hearing adults aged 18-35 years.  Caucasian groups had significantly higher WBA from 630-1250 Hz, while Chinese groups had significantly higher absorbance from 5000-6300 Hz. Female subjects had significantly higher WBA from 4000-5000 Hz than male subjects. Effect of test ear was not significant. Sun et al. (2016) WBA measured at ambient and TPP with custom research immittance equipment with the Titan, AT235 probe assembly. 84 normally-hearing adults aged 18-35 yrs.  WBA varied depending on pressurization method. The study analyzed WBA spectra extracted from a purely ambient recording mode, as well as WBA spectra extracted at ambient (0 daPa) and TPP from a WBT sweep. WBA at TPP was substantially higher in the low frequencies (+0.15) and was significantly lower in the high frequencies (-0.05), compared to WBA obtained from a purely ambient pressure recording mode.   16 Study Methods Subjects Summary of Findings Riddler (2017) WBA measured at ambient and TPP with the Titan by Interacoustics immittance system (v 3.4.0).  105 normally-hearing adults aged 18-53 yrs.  Caucasian groups had significantly higher WBA values from 250-2500 Hz, and Chinese participants had significantly higher WBA values from 3150-8000 Hz. Across frequency and gender, significant differences were found between ethnic groups for mean WBA values in both ambient and pressurized conditions; Caucasians had the highest mean WBA (ambient/TPP) (0.39/0.43), followed by the ‘Other’ group (0.39/0.42), with the Chinese group having the smallest mean WBA value (0.37/0.42). Males had higher WBA values at all frequencies except for the range of 3150-5000 Hz. Effect of test ear was not significant.   17  1.4 Norms for the Between-Ear difference of WBA: A Less Variable Alternative? The use of separate norms for distinct subpopulations and data collection methods can act to improve test performance, however, this approach may prove impractical in clinical settings. As discussed by Shahnaz, Feeney and Schairer (2013), clinicians often do not apply separate norms, despite there being evidence of the benefits of parameterization. Reasons preventing clinicians from using separate norms may vary, though we speculate that inconvenience features as a common justification. Accordingly, recommendations for research of the normative data approach must be balanced with the clinical perspective that a single normative data set which produces optimal diagnostic efficacy is ideal. With that in mind, the current study seeks to investigate a style of normative data for WBA which may satisfy both the research recommendations outlined in the consensus statement from the 2013 Eriksholm Workshop on Wideband Absorbance Measures of the Middle Ear and the clinical motivation for a stable norm that can be used across subpopulations. Instead of collecting normative data based on the individual ears of different subjects, employing measures that reflect the expected difference between two ears of the same subject may have potential in improving the precision of WBA measures. If individual variability affecting the acoustic transfer though the middle ear is anatomically and physiologically based, then the two ears of the same individual likely share the characteristics that account for such variability. Therefore, immittance characteristics measured from the two ears of one subject can be expected to be highly similar (Moller, 1960). In agreement with this line of reasoning, Werner et al. (2010) reported a significantly higher between-ear cross-correlation value for within-subject measures (0.84) of WBR, compared to randomly paired measures from different subjects (0.73). Similarly, a longitudinal study assessing sources of variability in WBA measures by 18  Feeney et al. (2014) showed that WBA measures from the two ears of one individual were more similar than WBA measured from ears belonging to different individuals.  Provided that anatomical variation between-subjects account for significant sources of variability in WBA measures of normally hearing adults, and the WBA values of the two ears of an individual are highly similar, it follows that measuring WBA in terms of the difference between the two ears of an individual could eliminate much of the variability in these measures. As such, WBA norms that represent the typical between-ear difference for an individual would be expected to show fewer sources of variability than norms based on WBA of individual ears from different subjects. Moreover, the effect of equipment and pressurization methods would ‘cancel out’ as these settings would likely be shared by an individual’s two ears. If this is indeed the case, then this style of normative data would be stable across subpopulations and data collection methods. Using the contralateral ear for reference may then offer clinicians a quick and reliable check for normality.  Preliminary investigations into the average between-ear difference of ambient WBA values suggest these values are small and highly consistent between individuals. Rosowski et al. (2012) found that mean WBA inter-aural differences ranged from -0.041 to 0.044, where absolute differences were greatest at 1000 and 2000 Hz. Werner et al. (2010) also found small right-left differences in WBA across adult subjects (0.02 on average), and reported a high degree of test-retest reliability of these measures. Population-based effects on inter-aural difference values of WBA were not analyzed in either of these studies, despite having been found significant for individual ear data; as such, it is not known if these measures share the same sources of variability as individual ear WBA measures.   19  Knowing the expected variation of WBA between the two ears could be particularly helpful in interpreting mechano-acoustic changes induced by pathologies with unilateral onset, as the individual could act as their own control (Norrix et al., 2013). Situations in which the WBA of the two ears of an individual are vastly different, though both fall into the normative range are not uncommon. This is largely a product of the wide ranges of ‘normal’ WBA observed in the normative data published by many researchers (Nakajima et al., 2013). Few clinicians would be willing to accept these findings as a ‘normal’ result, however, even in the absence of a significant unilateral conductive hearing loss. By using the contralateral ear for reference, clinicians could differentiate results that represent between-ear differences that occur in normally-functioning ME systems and those with a unilateral abnormality, with greater precision than is possible with the conventional individual ear norms.  A foreseeable query regarding the value an inter-aural difference measure concerns the potential confound introduced by bilateral pathologies. It is possible that the inter-aural difference of WBA measured from a participant with two affected ears would be indistinguishable from normal values. Having a norm for inter-aural differences does not, however, preclude the use of the conventional individual ear norms; it simply provides an additional metric upon which to base judgements of normality. Moreover, WBA is only one component of the audiological test battery. It is highly unlikely that an individual conductive hearing loss would not be detected by air bone gaps exceeding 10 dB, absent middle ear muscle reflexes, or WBA (Nakajima et al., 2012).  20  1.4.1 An Inter-Aural Difference Measure of WBA Could be a Useful Variable to Identify Otosclerosis Abnormal attenuation of air-conducted sound caused by lesions to the structures of outer and ME is referred to as a conductive hearing loss (CHL) (Martin & Clark, 2000). Often, the cause of a CHL is readily identified using an audiological test battery including otoscopy, audiometry, tympanometry and middle ear muscle reflex (MEMR) testing. This is generally the case for disorders such as tympanic membrane perforation or otitis media with effusion, where the abnormality is characterized by a relatively obvious change to the anatomy/physiology of the structures involved. By contrast, conditions in which the TM is intact and the middle ear cavity is aerated, such as superior canal dehiscence (SCD), ossicular discontinuity, and otosclerosis, are considerably more elusive. In fact, there are currently no diagnostic tests in widespread clinical use that can reliably distinguish disorders of this nature; though, investigations into the use of WAI techniques have been met with promising results (Nakajima et al., 2012).  The current practices used for the assessment of otosclerosis, superior canal dehiscence (SCD) and ossicular discontinuity do not allow for a differential diagnosis. The inconsistent patterns and severity of air bone gaps (ABG) on the audiogram within and among these conditions prevents audiometry from providing a conclusive etiological indication (Rappaport & Provencal, 2002). The diagnostic performance of single frequency tympanometry is tenuous for the detection and differentiation of ossicular disorders (Feldman, 1974). Vestibular evoked myogenic response (VEMP) and middle ear muscle reflex (MEMR) testing can differentiate ossicular disorders from SCD; though the results of these assessments do not distinguish pathologies affecting the ossicles (Minor, 2005). Moreover, absent MEMRs are relatively common in populations with normal ME function, so SCD can potentially be missed with this 21  test (Mujeri et al., 2010). High resolution computed tomography (CT) scanning can be used to detect these disorders, however, the associated radiation risks to the patient, and poor sensitivity prevent widespread clinical use of this imaging technique for most patients with CHL (Naumann, Porcellini & Fisch, 2005).  Otosclerosis is the most common cause of CHL that presents with an intact TM and aerated ME cavity (Nakajima et al., 2012); it is also among the most common causes of acquired hearing loss in adults (Shahnaz & Polka, 1997). Otosclerosis is more common among females than males; the incidence ratio of female to male expression is currently estimated at 3.2 (Sziklai, 2016). Prevalence of otosclerosis is highest among Caucasian individuals at approximately 3 per 1000, though up 12% of the Caucasian population show histopathological (pre-symptomatic) signs of the disease (Rudic et al., 2015). Other ethnicities are less affected, with prevalence in South American and Japanese populations half that of their Caucasian counterparts.  The onset of otosclerosis is typically unilateral, however, 80% of individuals with otosclerosis in one ear eventually develop the disease on the contralateral side (Foster & Backous, 2018). The disease characterized by abnormal deposition of bone in the otic capsule, typically at the footplate of the stapes in its junction with the oval window (Foster & Backous, 2018). As new bone forms and hardens, the stapedial footplate becomes fixed in varying degrees, resulting in increased stiffness of the middle ear system, accompanied by a CHL of up to 60 dB (Keefe et al., 2017). The typical configuration of audiological results for otosclerosis is summarized below, in a table that was recreated from Danesh, Shahnaz and Hall (2018) (Table 1.2).   22  Table 1.2. Basic patterns of audiological test results obtained for otosclerotic ears. This table is recreated from Danesh, Shahnaz, and Hall (2018). Procedure Findings Pure Tone Audiometry Air and Bone Conduction  Air conduction thresholds show hearing loss across frequencies that is more pronounced in the low frequencies. Bone conduction thresholds are typically normal, although with a characteristic ‘dip’ at 2000 Hz. Air-bone gap apparent across most frequencies.  Weber Test The pure tone should lateralize to the affected ear.  Speech Audiometry Speech Reception Threshold  Consistent with pure tone average.  Word Recognition Score  Consistent with minimal difficulty understanding speech, given the absence of a concurrent sensorineural hearing loss.   Acoustic Immittance Measures 226-Hz Tympanometry Normal or shallow type A tympanogram due to increased stiffness.  Acoustic Reflexes  Absent in clinical forms, and elevated (probe effect) in histopathological forms.  Otoacoustic Emissions  Absent, as they cannot be detected by the probe due to reduced ‘back transfer’ of sound.   Given the challenging nature of obtaining a differential diagnosis of otosclerosis with other means of assessment, the current diagnostic gold standard for this disease is surgery (Probst, 2007). For reasons mentioned earlier in this section, patients and surgeons are often unaware of the diagnosis preoperatively. Having a noninvasive test that identifies and differentiates among conductive disorders would be of value, as it would aid in preoperative planning and counselling, and may prevent unnecessary surgery for inoperable cases (Nakajima et al., 2012).    WAI techniques have been used successfully to differentiate etiologies of CHL. The patterns of WBA generally reflect the status of the TM and ossicular chain; pathologies that affect the conductive pathway are characterized by unique WBA contours (Danesh, Shahnaz & Hall, 2018). Nakajima et al. (2013) conducted a review of the WAI literature that has reported on 23  clinical populations, and summarized the characteristic patterns that have been observed for various pathologies. As discussed by the researchers, the WBA pattern of ossicular discontinuity obtained from individual ears is easily identified by its sharp low-frequency peak. On the other hand, otosclerosis remains a challenging to identify, despite several studies having reported on the high level of diagnostic accuracy of WBA for this disease (Shahnaz et al., 2009; Keefe et al., 2017). Measures of central tendency for otosclerosis show subnormal low-frequency WBA values, but many researchers comment that individual WBA results for otosclerotic ears overlap considerably with the normal range (Shahnaz et al., 2009). Nakajima et al. (2013) explained that this problematic finding is likely a product of the wide range of normal observed for WBA measures. Overlap of normally-functioning and diseased distributions represent the potential for error of clinical tests, namely poor sensitivity and specificity (Turner, Robinetee & Bauch, 1999). Distribution overlap also presents a familiar dilemma to the clinician: population means may show vast differences between normal and diseased groups, but the clinician must base their decision on the results of the individual (de Jonge & Valente, 1979). Attempts at using the normative data approach to separate the distributions of normally hearing and otosclerotic WBA results have increased the degree of separation between the groups, as indexed by increased diagnostic accuracy (e.g. Shahnaz & Bork, 2006). However, it may be unfeasible in clinical situations to have separate norms available that are specific to the demographic characteristics of all patients.  Inter-aural difference measures of WBA could address the challenges to pre-surgical diagnosis of otosclerosis with WAI from two vantages. Firstly, this measure likely omits the sources variability introduced to normative data via intra-subject measurements (e.g. ethnicity 24  and gender), ultimately providing a more homogeneous norm. Accordingly, a practical benefit would be that parameterization of norms for this measure would theoretically be unnecessary.  Secondly, wide ranges of normal for individual ear WBA seems to interfere with the detection of otosclerosis; results from affected ears fall largely within normal values (Nakajima et al., 2013). Otosclerosis presents primarily as a unilateral disease, which progresses to bilateral in the later stages (Foster & Backous, 2018). It follows that having a metric for the normal between-ear WBA difference, may then more readily detect the presence of this disease. This is because, in the absence of any abnormality, the immittance characteristics of the two ears of one individual are expected to be largely the same, resulting in an inter-aural difference of approximately zero (Moller, 1960). However, if one ear was affected by a CHL-causing pathology, then inter-aural difference values obtained from that individual would be considerably higher. Similar with the individual ear measures of WBA, the frequency-specific patterns of the inter-aural difference of WBA could inspire suspicion indices for the various pathologies. For example, otosclerosis would likely present with greater mean inter-aural difference values at frequencies below 2000 Hz, consistent with the abnormal stiffening effects observed with this pathology (Allen et al., 2005).  To investigate this hypothesis, the current study analyzed the diagnostic efficacy of each type of WBA measure (individual ear, inter-aural difference) for the detection of otosclerosis.  A new set (i.e. not previously analyzed) of WBA records from 28 subjects with surgically confirmed otosclerosis contributed to the analysis. 1.5 The Rationale for the Current Study  WBA middle ear analysis describes an emerging field of measurement that has shown promise in the detection of middle ear diseases and injuries. While this measurement provides 25  superior diagnostic efficacy over traditional 226-Hz and multi-frequency tympanometry, the overlap between normally hearing populations and those with conductive pathology limits the sensitivity of this test (Nakajima et al., 2013; Shahnaz et al., 2009). Researchers have successfully demonstrated that a reduction of intra-subject variability in normative samples, brought about by using ethnicity-specific norms, improves diagnostic efficacy of WBA for ossicular chain pathology (Shahnaz & Bork, 2006). The investigation of sources of variability within normal groups is therefore warranted, as separate norms can improve the accuracy of these measures.   WBA values obtained from the two ears of the same subject likely do not exhibit as many sources of variability as WBA measured from two different subjects, because of general body symmetry and the elimination of intra-subject factors such as ethnicity and gender. Accordingly, norms based on the between-ear difference of WBA for each subject may provide better separation of normal and diseased groups. This may be especially true for pathologies that present unilaterally. The current investigation evaluated sources of variability affecting a normative data set obtained by measuring WBA of individual ears between-subjects (individual ear norms), and a normative data set representing the absolute WBA difference between the ears of each subject (inter-aural difference norms). A comparison of the diagnostic efficacy of WBA in distinguishing normal and otosclerotic ears was also conducted, to evaluate whether there are clinical benefits to the use of one style of WBA measure over the other.   1.6 Aims of the Current Investigation  The current study aims to: 1) establish WBA norms for individual ears in an adult population; 2) establish norms for the inter-aural difference of WBA in the same population; 3) evaluate the sources of variability affecting normative WBA measures for either style of 26  measurement; 4) evaluate the test performance of the two types of WBA measures for the detection of otosclerosis.  1.7 Hypotheses of the Current Investigation  There are two central null hypotheses to be investigated in the current study: H01) The same sources of variability will be significant in difference norms, that are significant in individual ear norms; H02) The diagnostic accuracy of WBA parameters will be equal when difference versus individual ear norms are applied.   27  Chapter 2: Methods  The current investigation was approved by the University of British Columbia Clinical Research Ethics Board (CREB) as an independent study, receiving the CREB certificate # H18-01238. Data was obtained, analyzed and stored in compliance with the regulations of the CREB.  2.1 Data Sources The current study provided a retrospective analysis of WAI data from three research studies conducted in the Middle Ear Laboratory at the University of British Columbia. Wideband acoustic immittance data was originally collected for the studies entitled, “Effects of Race, Caucasian, East Asian on Middle Ear Function and Hearing Sensitivity Norms” (CREB # H02-70609), “Wideband Energy Reflectance in Surgically Confirmed Middle Ear Ossicular Chain Abnormalities” (CREB #H04-70098, C04-0098), and “Effects of Middle Ear Pressure Compensation on Evoked Otoacoustic Emissions and Power Absorbance in Adults” (CREB # H15-03494).  2.2 Subject Records  Records from a total of 181 normally-hearing subjects were obtained from the three data sources. Records from 28 subjects (30 ears) with surgically confirmed otosclerosis were obtained from the study “Wideband Energy Reflectance in Surgically Confirmed Middle Ear Ossicular Chain Abnormalities”; this group contributed to the AUROC analysis, described in a subsequent section. The data from the otosclerosis group has not been analyzed in any previously published work. The records, stored in the Otoaccess V.1.2.1 database, were extracted using a custom Excel spreadsheet provided by Interacoustics and compiled into a single Excel database for analysis. 28  2.2.1 Inclusion Criteria  Normally Hearing Group Subject records were included in the current study if they were determined to have had ‘normal hearing’ binaurally at the time of primary data collection. Normal hearing was evidenced by the following criteria, as noted by the three previous studies from which data was obtained for the current analysis: 1) Case history details that were taken prior to audiometric and immittance testing confirmed no history of head trauma or sensorineural hearing loss. 2) Canals were relatively clear of cerumen, and the tympanic membrane free of abnormalities, as evidenced by otoscopic examination. 3) Pure tone thresholds no greater than 20 dB HL at octaves between 0.25k and 8 kHz. 4) No air-bone gaps exceeding 10 dB noted at any octave between 0.5k and 4 kHz. 5) A discernable tympanometric peak must have been noted at ambient middle ear pressure, as measured by 226-Hz tympanometry. In cases where WAI was used to evaluate the status of the middle ear, wideband absorbance must have resided within the 90% for the range for normal ears, as provided in the Titan IMP440 module. 6) Middle ear muscle reflexes were present at expected levels at 0.5k, 1k, 2 kHz and with a broadband noise stimulus. 7) Distortion product otoacoustic emissions must have been present from 1k through 8 kHz, with at least a 3-dB emission to noise ratio and a minimum of -6 dB for DP absolute amplitude. Note, no attempt was made to exclude records based on calculated inter-aural WBA differences due to outlying values, if the subject met inclusion criteria binaurally. Otosclerosis Group The inclusion criteria for individuals with otosclerosis were as follows. 1) Case history revealed no history of middle ear pathology or injury, prior to the presentation of otosclerotic symptoms. 2) Pre-operative audiometric findings indicative of otosclerosis, as described by 29  Danesh, Shahnaz and Hall (2018) (summarized in Table 1.2). 3) Surgical confirmation of clinical otosclerosis, as evidenced by the surgical report. 4) The absence of any additional, confounding conductive pathology, as evidenced by the surgical report. 2.2.2 Exclusion Criteria  Normally Hearing Group  As binaural data was required for the present study 32 subject records with only unilateral data collected or with one ear which fell outside of inclusion criteria were excluded (17.67%). Twelve subject records were excluded due to bilateral abnormalities noted in the immittance results, which were consistent with abnormal middle ear pressure (6.62%). An additional 15 subject records were excluded because WBA results were consistent with a pattern indicative of equipment failure (i.e. blocked probe-tip, loss of probe seal, excessive test noise) (8.28%). After exclusions, a total of 122 normally-hearing subject records (244 ears) were included in the study.  Otosclerosis Group   In total, records from 28 subjects (30 ears) contributed to the otosclerosis group. The method of mean substitution was used to accommodate missing data points for 4 subjects, in the purely ambient (Amb_WBA) pressurization method. Missing data occurred at several frequencies in either ear and did not encompass entire pressure conditions. See Table A.1 in Appendix A  for a detailed summary of records that underwent mean substitution.  2.2.3 Ethnicity of Participants Data was coded with the gender and ethnicity of the participants, where provided. Data belonging to both the normal and otosclerosis group was coded as male or female and Caucasian, Asian, or Mixed. The Asian group consisted of all participants who had self-reported their ethnic background as Filipino, Malaysian, Korean, Singaporean, Indonesian, Vietnamese, Bruneian, or 30  Chinese at the time of primary data collection. Chinese refers to Canadian or Chinese born participants whose parents have lineage from mainland China, Hong Kong, or Taiwan with no distinguishable foreign descent. The Caucasian group was comprised of participants whom self-reported being of European descent with light pigmentation of the skin, not identifying as Chinese, South/East/West Asian, Aboriginal, Arab, Black, Filipino, and/or Hispanic (Statistics Canada, 2004 as cited by Jaffer, 2016). The Mixed group was comprised of data from participants whose lineage was not aptly described by the previous two categories, was a combination or ‘mix’ of two or more ethnicities, or who chose not to self-report their ethnicity.  Across the Normal and Otosclerosis groups, records from 74 Caucasian, 57 Asian and 21 Mixed subjects were included. Gender groups were comprised of 83 females and 67 males. Full demographic characteristics of the sample analyzed in the current study are detailed in Table 2.1.  Table 2.1. Demographic summary of normal and otosclerosis groups. Subject Characteristics: Normal Otosclerosis Male Female All Male Female All Asian        N 23 30 53 2 2 4 Mean Age (yrs) 24.30 24.28 24.23 63.00 42.5 52.75 SD (yrs)  3.61 4.44 4.10 4.00 7.5 11.88 Range 18-32 18-34 18-34 59-67 35-50 35-67 Caucasian        N 28 32 60 5 8 13 Mean Age (yrs) 24.64 23.31 23.93 50.60 40.88 44.61 SD (yrs) 4.67 3.72 4.24 7.17 11.16 10.90 Range 18-35 18-34 18-35 43-63 30-60 30-63 Mixed       N 5 4 9 4 7 11 Mean Age (yrs) 25.20 25.00 25.11 42.25 53.43 49.36 SD (yrs) 8.0 5.39 6.97 5.93 12.09 11.61 Range 18-35 20-34 18-35 36-50 33-77 33-77 Total       N 56 66 122 11 17 28 Mean Age (yrs) 24.56 23.80 24.15 49.82 46.24 47.64 SD (yrs) 4.69 4.22 4.46 9.58 12.73 11.7 Range 18-35 18-34 18-35 36-67 30-77 30-77 31  2.3 Instrumentation All records included in the current study were recorded by the original researchers using the Titan by Interacoustics acoustic immittance system (software version 3.4.0, build number 3.4.6246.26391). The system includes a handheld immittance device and a PC computer in which integrated audiological software has been installed. Figure 2.1 shows the Titan system connected to a laptop PC, with the Titan software interface open to the IMP440 module, and set to record a measurement in the wideband absorbance tab. The handheld device includes a probe assembly, which is inserted into the subject’s ear during measurement, once a pliable plastic tip has been fit (shown in red in Figure 2.1).  The Titan IMP440 module uses a broadband click stimulus with a bandwidth of 226 to 8000 Hz. The WAI measurement yields 107 frequency data points (1/24th-octave bands), which were averaged into 16 1/3rd octave bands to reduce inflated type I errors for the current analysis; Table A.2 in Appendix A  contains the bandwidth of each center frequency. For adults, the click presentation level is maintained at 100 dB peSPL, and is presented at a rate of 21.5/sec; the total duration of measurement is typically less than 10 seconds (Interacoustics, 2017). Pressure is automatically controlled, though the pump speed is preferentially selected in the protocol set-up window; all studies from which data was analyzed used a ‘medium’ pump speed, which corresponded to 200 daPa/sec, and pressure boundaries were set at -400 to +300 daPa.  Immittance data is available from two tabs within the Titan IMP440 module: the wideband tympanometry tab (3DT), or the ambient wideband absorbance tab (WBA). The 3DT tab provides a 3-dimensional measure of pressure, and frequency on the x-axes and WBA on the y-axis (Figure 2.2). Absorbance at TPP, defined in terms of the absorbance tympanogram averaged from 0.376 to 2 kHz (Hunter & Shahnaz, 2014), and ambient pressure (corresponding 32  to 0 daPa, within the pressure sweep) can be extracted and graphically displayed on the same panel in the 3DT tab. The right panel of Figure 2.3 is an example of an absorbance graph measured with the 3DT tab for a normally-hearing ear with positive pressure induced by the Valsalva manoeuver, for illustrative purposes. The second tab, called the wideband absorbance tab, provides measures of WBA at an ambient pressure only; no accommodation is made for disparities in pressure between the middle ear cavity and that of the ear canal (Figure 2.3, left). The three pressure conditions analyzed in the current study arise from the two tabs described herein, and are summarized as follows: TPP and ambient pressure extracted from the 3DT tab (Peak_3DT, Amb_3DT), and ambient pressure extracted from the ambient wideband absorbance tab (Amb_WBA).  33   Figure 2.1 Titan by Interacoustics Handheld Unit connected to a PC laptop computer.  Figure 2.2 Measurement of a normal middle ear taken with the Wideband Tympanometry (3DT) tab of the Titan Suite (IMP440 module). 34   Figure 2.3 Right ear WBA of a normally-hearing ear with induced positive ME pressure, measured with the Titan IMP440 module wideband absorbance tab (left) and 3DT tab (right).  2.3.1 Calibration In each of the studies from which data was obtained for the current investigation, calibration was performed daily prior to the start of testing. Calibration of the Titan system was achieved by placing the probe assembly in each of four metal waveguide calibration units of 0.2, 0.5, 2 and 5 cc volumes. Based on acoustic measurements in the waveguide units, the source reflectance and incident pressure were determined for the probe assembly and its transducers. The calibration procedure of the Titan system was recently updated to incorporate modifications which provide more accurate and stable results. This procedure, researched by Nørgaard et al. (2017) for Interacoustics, involves placing the probe tip into the waveguide units without the ear tip attached. Consistent positioning of the probe assembly into the waveguide input aperture and use of the defined dimensions of the waveguides rather than estimation from the probe response make this procedure more precise than its predecessor.  35  Discussion of the calibration of the Titan system is important as it represents a limitation of the current study. Two of the studies which served as data sources for the current investigation performed the previous method of calibration, whereas the third employed the updated method. The previous calibration method involved placing the probe tip into the waveguide units with the ear tip attached and used estimations of the waveguide dimensions based on the probe response, introducing greater variability in results. To date, no publications have established whether the calibration method has a significant impact on measurements of WBA obtained from real ears.  2.4 Statistical Analysis 2.4.1 Analyses of Variance  A mixed model analysis of variance (ANOVA) approach was used to determine the effects of ethnicity and gender on the mean individual ear and inter-aural difference values of WBA. In this model, mean individual ear and inter-aural difference WBA magnitudes were investigated outcome measures; separate ANOVA analyses were carried out for each type of measurement. Ethnicity (Asian, Caucasian, Mixed) and gender (male, female) were between-subject factors. Frequency (16 1/3rd octave bands from 226-8000 Hz), pressurization method (Peak_3DT, Amb_3DT, Amb_WBA) and test ear (left, right) were treated as within-subject factors for the individual-ear norms analysis. Test ear was not a factor in the inter-aural difference norm ANOVA, as this analysis used between-ear difference values of absorbance (absolute value of the right ear WBA minus the left ear WBA). A Greenhouse-Geisser (GG) Correction was applied to all outcome measures, to accommodate the violation of the sphericity assumption characteristic of this experimental design, and to avoid inflated Type I errors. Where statistically significant differences were observed in the outcome measures between groups (i.e. the null hypothesis was rejected), a post hoc test, Tukey’s Honestly Significant Difference 36  (HSD), was administered to identify the comparisons in which significant differences existed. This post-hoc analysis was selected because it is considered an acceptable and conservative method of controlling for multiple comparisons, where sample sizes are unequal (Bender & Lange, 2001).  For the current analyses, a p-value of .05 was selected to determine statistical significance. Results were reported in the format [F(df, df)=F, p=.x], where the p-value indicates significance, the degrees of freedom (df) indicates the number of data points contributing to the sample, and the F-test describes the ratio of variance between and within groups. The combination of a p-value less than .05 and a relatively large F-test value are indicative of significance. As the p-value represents the probability of the observed difference being due to chance, and the F-test is a measure of the variability between group means, a test result with a larger F-test value combined with the smallest p-value is desirable (Riddler, 2017). Statistica Software version 13 (TIBCO Inc., 2017) was used to conduct this portion of the statistical analysis.  2.4.2 Receiver Operating Characteristic Analyses A ROC curve is a valid method of evaluating the diagnostic accuracy of a clinical test or comparing the diagnostic accuracy of multiple tests. It is a plot of the sensitivity (true positive rate, valued between 0-1) of a test as a function of 1-specificity (false positive rate, valued between 0-1) for all possible criterion values of a parameter; each point on a ROC curve represents a sensitivity/specificity pair corresponding to a criterion threshold (Turner, Robinette & Bauch, 1999). The interpretation of a ROC curve is based on the calculation of the area under the curve. The area under a ROC curve (AUROC) varies from 0.5, indicating test performance at chance levels, and 1, indicating perfect test performance (Turner, Robinette & Bauch, 1999). A 37  significance level or p-value is reported; it represents the probability that the resulting AUROC would be obtained if the actual (true) area under the curve is 0.5. As such, the p-value denotes the probability that results were due to chance and not an indication of the test’s performance. A p-value of less than .05 (p<.05) is an indication that the AUROC is significantly different from 0.5, indicating that the diagnostic test operates above change levels, and is effective in discriminating diseased from normal populations (Turner, Robinette & Bauch, 1999). A ROC curve analysis was used to determine the clinical utility of WBA for the identification of otosclerotic ears, as a function of WBA measurement type (individual ear, inter-aural difference). The AUROC was calculated for each WBA variable as a univariate predictor, using the nonparametric U-test statistic (DeLong et al., 1988). The variables were selected at each test frequency (16 center frequencies of 1/3rd-octave bands from 250-8000 Hz) from all pressurization methods (Peak_3DT, Amb_3DT, Amb_WBA) and normative data types, totaling 96 ROC curves. These univariate predictors were selected because researchers have reported significant differences between WBA results obtained from normal and otosclerotic ears below 2000 Hz (Feeney, Grant & Maryott, 2003; Keefe et al, 2017; Merchant et al., 2016; Nakajima et al., 2012; Shahnaz et al., 2009; Voss et al., 2012), as well as at 4000 Hz (Keefe et al., 2017). As such, WBA values measured within this frequency range may serve as good candidate predictors of otosclerosis.  The AUROCs of the three most accurate WBA univariate predictors obtained with each type of norm, and the six most accurate predictors across normative data type, were compared using DeLong et al.’s (1988) method. The outcome of this method of comparison is independent of the test criterion. In clinical situations, the optimal criterion would be determined by the prevalence of the disease, and the associated costs of false positive or false negative 38  identification (Turner, Robinette & Bauch, 1999). However, given a lack of published otosclerosis prevalence data for the region in which this study was carried out, and the purpose of this analysis being to compare the efficacy of the two styles of WBA measurement (regardless of disease prevalence), the Youden Index cut-off was automatically selected. Youden’s Index is a value which corresponds to the point on the ROC curve that has the maximum vertical distance from the diagonal (chance) line; this is the optimal cut-off only when disease prevalence is 50% (Youden, 1950). When disease prevalence is unknown, the MedCalc software states that this decision threshold may favor specificity (MedCalc Software, 2018). The sensitivity and specificity of each WBA univariate predictor reported herein corresponds to the Youden Index criterion. All ROC analyses were calculated using MedCalc ® (MedCalc Software, 2018). No correction was applied to address familywise error rate introduced by multiple comparisons. 2.4.3 The justification for Combining ANOVA and ROC Analyses  The mixed model ANOVA and AUROC analyses provide complementary information about the clinical utility the two WBA test variables. The ANOVA identifies significant sources of variability affecting the normative WBA measurements. A reduction in the sources of variability affecting a measurement for a normal group could act to separate the distributions of normal and abnormal results, thereby increasing diagnostic accuracy (indexed by a higher AUROC) (Shahnaz, Feeney & Schairer, 2013; Turner, Robinette & Bauch, 1999). Conversely, a reduction in the sources of variability may not affect the spread within the normal distribution or the overlap between normal and abnormal distributions, in which case the AUROCs of the compared parameters would not differ significantly. As such, the AUROC analysis in the current investigation delivers suggestive evidence as to whether the reduction of intra-subject variability 39  provided by difference norms translates to improved sensitivity and specificity of WBA measures.  40  Chapter 3: Results This chapter details the results of the present investigation. Sections are organized as follows: (3.1) Wideband Absorbance: Individual Ear Data, (3.2) Wideband Absorbance: Inter-Aural Difference Data, and (3.3) Receiver Operating Characteristic Curve Analysis.  3.1 Wideband Absorbance: Individual Ear Data   The following sections contain the results of the individual ear analysis. Sections are organized as follows: (3.1.1) descriptive statistics of individual ear data, (3.1.2) ANOVA results for individual ear data.  3.1.1 Descriptive Statistics: Individual Ear Data   Table 3.1 and Table 3.2 contain the descriptive results, mean and standard deviation, for the individual ear data analyzed in the present study. The data is organized by ethnicity, pressurization method and frequency in Table 3.1, and by gender, pressurization method and frequency in Table 3.2. Separate descriptive tables for these groups are justified, as significant differences in the mean WBA existed between them (see section 3.1.2). Across all groups, mean WBA was lowest at 250 Hz, increased steeply with increasing frequency to about 1250 Hz, rose to a second maxima at about 4000 Hz, and decreased moderately in the high frequencies.  41  Table 3.1. Descriptive summary of individual ear WBA data by test frequency, pressurization method, and ethnicity. Ethnicity Asian Caucasian Mixed Pressure Peak_3DT Amb_3DT Amb_WBA Peak_3DT Amb_3DT Amb_WBA Peak_3DT Amb_3DT Amb_WBA Frequency (Hz) M SD M SD M SD M SD M SD M SD M SD M SD M SD 250 0.11 0.05 0.11 0.05 0.07 0.05 0.14 0.06 0.13 0.05 0.09 0.05 0.11 0.03 0.10 0.03 0.08 0.03 315 0.11 0.06 0.11 0.06 0.07 0.05 0.15 0.07 0.13 0.07 0.09 0.06 0.12 0.04 0.10 0.04 0.07 0.05 400 0.15 0.07 0.14 0.07 0.10 0.07 0.20 0.09 0.18 0.09 0.14 0.08 0.16 0.05 0.14 0.05 0.12 0.05 500 0.22 0.09 0.21 0.09 0.18 0.09 0.30 0.12 0.27 0.12 0.23 0.11 0.24 0.07 0.22 0.07 0.20 0.06 630 0.31 0.11 0.30 0.12 0.27 0.12 0.41 0.16 0.38 0.15 0.35 0.15 0.34 0.09 0.31 0.09 0.30 0.09 800 0.43 0.15 0.41 0.15 0.38 0.16 0.52 0.16 0.49 0.16 0.45 0.17 0.44 0.12 0.41 0.13 0.40 0.12 1000 0.53 0.15 0.52 0.15 0.48 0.15 0.59 0.14 0.57 0.14 0.52 0.15 0.54 0.13 0.51 0.13 0.50 0.15 1250 0.59 0.11 0.59 0.11 0.54 0.13 0.63 0.10 0.63 0.10 0.58 0.12 0.62 0.11 0.60 0.11 0.59 0.14 1600 0.60 0.10 0.60 0.11 0.55 0.13 0.62 0.09 0.63 0.09 0.57 0.10 0.66 0.11 0.66 0.11 0.58 0.12 2000 0.61 0.12 0.62 0.13 0.56 0.14 0.62 0.11 0.63 0.11 0.57 0.12 0.64 0.13 0.65 0.13 0.57 0.15 2500 0.69 0.14 0.69 0.14 0.66 0.15 0.69 0.11 0.70 0.11 0.67 0.12 0.71 0.15 0.72 0.15 0.68 0.15 3150 0.72 0.14 0.73 0.14 0.70 0.15 0.69 0.13 0.70 0.13 0.70 0.13 0.70 0.22 0.70 0.22 0.70 0.20 4000 0.70 0.16 0.71 0.16 0.71 0.17 0.63 0.17 0.64 0.16 0.68 0.15 0.69 0.19 0.70 0.20 0.72 0.17 5000 0.57 0.15 0.58 0.15 0.61 0.16 0.48 0.16 0.48 0.15 0.52 0.16 0.50 0.16 0.51 0.17 0.55 0.17 6300 0.40 0.14 0.40 0.14 0.43 0.15 0.32 0.13 0.32 0.14 0.34 0.14 0.26 0.13 0.26 0.12 0.31 0.12 8000 0.32 0.17 0.32 0.17 0.32 0.20 0.23 0.16 0.23 0.16 0.20 0.19 0.22 0.23 0.21 0.23 0.19 0.28 Notes: M= Mean; SD = Standard Deviation 42  Table 3.2. Descriptive summary of individual ear WBA data at each test frequency and pressurization method, organized by subject gender. Sex: Male Female Pressure: Peak_3DT Amb_3DT Amb_WBA Peak_3DT Amb_3DT Amb_WBA Frequency (Hz) M SD M SD M SD M SD M SD M SD 250 0.14 0.05 0.13 0.05 0.10 0.05 0.11 0.05 0.10 0.05 0.07 0.04 315 0.15 0.06 0.13 0.06 0.10 0.06 0.12 0.06 0.11 0.06 0.07 0.05 400 0.19 0.08 0.18 0.07 0.14 0.07 0.16 0.09 0.15 0.09 0.10 0.07 500 0.28 0.10 0.26 0.09 0.23 0.10 0.24 0.12 0.23 0.12 0.19 0.10 630 0.39 0.12 0.36 0.12 0.35 0.14 0.34 0.16 0.32 0.15 0.29 0.14 800 0.50 0.14 0.47 0.14 0.44 0.15 0.45 0.17 0.43 0.17 0.40 0.17 1000 0.59 0.13 0.57 0.13 0.52 0.15 0.54 0.15 0.52 0.15 0.48 0.16 1250 0.64 0.10 0.63 0.10 0.59 0.12 0.60 0.11 0.59 0.11 0.55 0.13 1600 0.64 0.08 0.64 0.08 0.59 0.10 0.60 0.11 0.60 0.11 0.54 0.13 2000 0.64 0.11 0.65 0.11 0.60 0.12 0.59 0.12 0.60 0.12 0.54 0.13 2500 0.71 0.12 0.72 0.12 0.69 0.13 0.68 0.12 0.68 0.13 0.65 0.14 3150 0.69 0.14 0.70 0.14 0.70 0.14 0.71 0.15 0.72 0.15 0.70 0.15 4000 0.62 0.18 0.63 0.17 0.65 0.16 0.71 0.15 0.71 0.15 0.74 0.14 5000 0.47 0.16 0.47 0.16 0.51 0.16 0.56 0.15 0.57 0.15 0.60 0.16 6300 0.34 0.13 0.34 0.14 0.35 0.14 0.37 0.15 0.37 0.15 0.39 0.16 8000 0.29 0.18 0.28 0.18 0.28 0.21 0.25 0.18 0.25 0.18 0.23 0.21 Notes: M= Mean; SD = Standard Deviation  3.1.2 ANOVA: Individual Ear Data  To evaluate the effects of pressurization method (Peak_3DT, Amb_3DT, Amb_WBA) and frequency (16 center frequencies of 1/3rd octave bands from 250-8000 Hz) on mean WBA magnitude for adults of different ethnicities and genders, a mixed model ANOVA (Table A.3 and Table A.4 in Appendix B  ) was performed with ethnicity (Caucasian, Asian, Mixed), gender (male, female), and test ear (right, left) as between-subject factors, and pressurization method and frequency as within-subject factors. Unless stated otherwise, significant results remained significant following G-G correction.  43  The main effect of pressurization method was significant [F(2, 578)=58.997, p=.0000], indicating that mean WBA varies differently between the pressure conditions when the data were collapsed across ethnicity, gender, and frequency. The main effect of frequency was significant [F(15, 3585)=482.993, p=.0000], indicating that mean absorbance varies in a frequency-dependent manner for adults of all ethnicities and genders, regardless of pressure condition. The interaction of frequency and pressure was significant [F(30, 7170)=28.291, p=.0000], indicating that mean WBA varies differently across frequency and pressure conditions, for adults of all ethnicities (Caucasian, Asian, Mixed) and genders (male, female). The factor of test ear did not show a significant main effect or interaction with any other factor (p>.05).  Ethnicity Effects   The interaction of pressurization method, frequency and ethnicity was significant [F(60, 7170)=2.501, p=.0000], indicating that mean WBA varies differently between adults of different ethnicities, across the measured frequency range, and pressure conditions. This interaction is displayed below, in ( Figure 3.1 and Figure 3.2). Post-hoc analysis with Tukey’s HSD revealed: At TPP and ambient pressure measured in the Titan 3DT tab (Peak_3DT, Ambient_3DT): The Caucasian group had significantly higher mean WBA magnitudes than the Asian group at 500-1000 Hz, whereas the Asian group had significantly higher mean WBA than the Caucasian group from 4000 – 8000 Hz. The Asian group had significantly higher mean WBA than the Mixed group at 6300-8000 Hz. There was no significant difference between the Caucasian group and the Mixed group at any frequency.   44   At ambient pressure measured in the Titan WBA tab (Amb_WBA):  The Caucasian group had significantly higher mean WBA magnitudes than the Asian group from 500-800 Hz, whereas the Asian group had significantly higher mean WBA than the Caucasian group from 5000-8000 Hz. The Asian group had significantly higher mean WBA than the Mixed group at 6300-8000 Hz. There was no significant difference between the Caucasian and Mixed groups at any frequency For the Asian Group:   Significantly higher and lower mean WBA magnitudes were obtained below 2000 Hz and above 4000 Hz, respectively, when measured with the WBT pressurization methods (Peak_3DT, Amb_3DT) compared to the purely ambient pressurization method (Amb_WBA). Significantly higher mean WBA magnitudes were obtained with the Peak_3DT pressurization method than the Amb_3DT method from 400-1000 Hz and 4000-8000 Hz.  For the Caucasian Group:   Significantly higher mean WBA magnitudes were obtained below 2000 Hz and lower values were obtained above 4000 Hz, except at 6300 Hz, when measured with the WBT pressurization methods (Peak_3DT, Amb_3DT) compared to the purely ambient pressurization method (Amb_WBA). Significantly higher mean WBA magnitudes were obtained with the Peak_3DT pressurization method than the Amb_3DT method from 400-800 Hz.  For the Mixed Ethnicity Group:   Mean WBA magnitudes measured at Peak_3DT were significantly higher from 630 to 800 Hz, compared to measurements taken at ambient pressure from either tab (Amb_3DT, Amb_WBA). 45   Figure 3.1 Mean WBA magnitudes from 250-8000 Hz for Caucasian, Asian and Mixed ethnicity normally-hearing adults, as a function of frequency and pressurization method. Current Effect: [F(60, 7170)=2.501, p=.0000]. Vertical bars denote 95% confidence intervals.  Caucasian Asian MixedPeak_3DT250400630100016002500400063000.00.10.20.30.40.50.60.70.80.91.0Absorbance Magnitude (0-1)Ambient_3DT25040063010001600250040006300Ambient_WBA2504006301000160025004000630046   Figure 3.2 Mean WBA magnitudes from 250-8000 Hz for Caucasian, Asian and Mixed ethnicity normally-hearing adults measured at Peak_3DT, Amb_3DT and Amb_WBA pressurization methods, as a function of frequency and ethnicity. Current Effect: [F(60, 7170)=2.501, p=.0000]. Vertical bars denote 95% confidence intervals.   Peak_3DT Amb_3DT Amb_WBACaucasian250400630100016002500400063000.00.10.20.30.40.50.60.70.80.91.0Absorbance Magnitude (0-1)Asian25040063010001600250040006300Mixed 2504006301000160025004000630047  Gender Effects   The interaction of pressure, frequency, and gender did not reach significance (p>.05). However, the interaction of frequency and gender was significant [F(15, 3585)=10.833, p=.0000], indicating that mean WBA varies differently between male and female groups, across frequencies. This interaction (Figure 3.3), shows that men had slightly higher mean WBA values from 250-2500 Hz and at 8000 Hz, whereas women had a slightly higher mean WBA from 3150-5000 Hz. Post-hoc analysis with Tukey’s HSD showed that mean WBA values between genders differed significantly only at 4000-5000 Hz, with women showing higher mean values at these frequencies.   Figure 3.3 Mean WBA magnitudes from 250-8000 Hz for male and female normally hearing adults as a function of frequency. Current effect: [F(15, 3585)=10.833, p=.0000]. Vertical bars denote 95% confidence intervals.  Female Male250 400 630 1000 1600 2500 4000 6300Frequency (Hz)0.00.10.20.30.40.50.60.70.80.91.0Absorbance Magnitude (0-1)48  3.2 Wideband Absorbance: Inter-Aural Difference Data  Sections 3.2.1 and 3.2.2 contain the results of descriptive statistics and the ANOVA analysis of inter-aural difference data, respectively.  3.2.1 Descriptive Statistics: Inter-Aural Difference Data  Table 3.3, below, illustrates the mean and standard deviation of inter-aural WBA difference values at each test frequency and pressurization method. The data in Table 3.3 is collapsed across the factors of gender and ethnicity, as these factors did not reach significance in the ANOVA analysis (see section 3.2.2). Overall, mean inter-aural WBA difference was smallest at the low frequencies and increased with increasing frequency.  Table 3.3. Descriptive summary of inter-aural WBA difference data at each test frequency and pressurization method.  Pressure Peak_3DT Amb_3DT Amb_WBA Frequency (Hz) M SD M SD M SD 250 0.03 0.03 0.03 0.03 0.03 0.02 315 0.04 0.03 0.04 0.03 0.03 0.03 400 0.05 0.04 0.05 0.04 0.04 0.04 500 0.06 0.05 0.06 0.06 0.06 0.05 630 0.08 0.07 0.09 0.07 0.09 0.08 800 0.09 0.08 0.10 0.08 0.11 0.09 1000 0.08 0.07 0.09 0.07 0.11 0.08 1250 0.07 0.06 0.07 0.06 0.09 0.08 1600 0.07 0.06 0.07 0.06 0.08 0.07 2000 0.09 0.07 0.09 0.07 0.10 0.07 2500 0.08 0.07 0.09 0.07 0.09 0.07 3150 0.10 0.08 0.10 0.08 0.10 0.08 4000 0.10 0.11 0.11 0.11 0.11 0.10 5000 0.09 0.09 0.10 0.09 0.10 0.09 6300 0.09 0.09 0.09 0.09 0.09 0.09 8000 0.10 0.10 0.10 0.10 0.11 0.10 Notes: M= Mean, SD = Standard Deviation   49  3.2.2 ANOVA: Inter-Aural Difference Data  To evaluate the effects of pressurization method (Peak_3DT, Amb_3DT, Amb_WBA) and frequency (16 center frequencies of 1/3rd octave bands from 250-8000 Hz) on mean inter-aural WBA difference magnitude for adults of different ethnicities and genders, a mixed model ANOVA (Table A.5 and Table A.6 in Appendix B  ) was performed with ethnicity (Caucasian, Asian, Mixed) and gender (male, female) as between-subject factors, and pressurization method and frequency as within-subject factors. Unless stated otherwise, significant results remained significant following G-G correction.   The main effect of frequency was significant [F(15, 1770)=10.2655, p=.0000], indicating that mean inter-aural WBA difference values vary in a frequency-dependent manner for adults of all ethnicities and genders, regardless of pressurization method. The interaction of pressurization method and frequency was significant [F(30, 3540)=2.2316, p=.0001], indicating that mean inter-aural WBA difference values vary differently across frequency and pressure conditions for adults of any ethnicity (Caucasian, Asian, Mixed) and gender (male, female). This interaction is displayed below (Figure 3.4). Post-hoc analysis with Tukey’s HSD revealed that mean inter-aural WBA difference values were significantly higher when measured in the Titan WBA tab (Amb_WBA) from 1000-1250 Hz than measurements obtained at TPP (Peak_3DT) or ambient pressure (Amb_3DT) from the Titan 3DT tab. Inter-aural WBA difference values did not differ significantly between Peak_3DT and Amb_WBA test pressurization conditions.  50   Figure 3.4 Mean inter-aural WBA values from 250-8000 Hz for adults with normal hearing, measured with 3 pressurization methods. Vertical bars denote 95% confidence intervals. Ethnicity and Gender  The main effects of ethnicity and gender were not significant (p>.05), and there were no significant interactions between ethnicity or gender with any other factor. The lack of significance suggests that mean inter-aural WBA difference values are essentially the same between these groups across frequencies and pressurization methods.  3.3 Receiver Operating Characteristic Curve Analysis  Receiver operating characteristic curves were generated at each frequency and pressure condition to evaluate efficacy of WBA in distinguishing results from normal and otosclerotic middle ears as a function of the normative data type (individual ear, inter-aural difference). WBA at each combination of frequency, pressurization method, and norm type was considered a ‘univariate predictor’ of otosclerosis (e.g. Absorbance at 250 Hz, measured at TPP, using  Peak_3DT Ambient_3DT Ambient_WBA250 400 630 1000 1600 2500 4000 6300Frequency (Hz)0.000.020.040.060.080.100.120.140.16Absolute Inter-Aural Absorbance Difference (0-1)51  individual ear data = one univariate predictor). There were 48 predictors of otosclerosis for each type of data (96 total univariate predictors), corresponding to the 3 pressurization methods (Peak_3DT, Amb_3DT, Amb_WBA) and 16 center frequencies of 1/3 octave bands from 250-8000 Hz. Table A.7 in Appendix B  summarizes the results of the ROC curve analyses for each univariate predictor. Figure 3.5 below displays these results graphically; numeric values displayed at the top of the chart denote the six best univariate predictors of otosclerosis in rank order, while data labels convey the AUROC value for these predictors.   Figure 3.5 Summary of AUROCs for all univariate predictors of otosclerosis (n=96). Numeric labels across the top pane denote the rank order of highest AUROC values from 1-6. Data labels provide the associated AUROC value. The three inter-aural WBA difference univariate predictors with the highest AUROCs were compared to the three individual ear data univariate predictors with the highest AUROCs ( 0.7010.6540.6880.6850.6570.7080.450.50.550.60.650.70.75100 1000 10000AUC Value (0-1)Frequency (Hz)Group_Peak_3DT Group_Ambient_3DT Group_Ambient_WBADifference_Peak_3DT Difference_Ambient_3DT Difference_Ambient_WBAMax AUC=0.7081 2, 34 6 552  Table 3.4) using DeLong’s (1988) method of pairwise comparisons for AUROC values (Figure 3.6). The results of the pair wise comparisons are displayed in Table A.8 in Appendix B  ; results indicated there was no significant difference between AUROCs of the any of the compared predictors (p>.05), implying that they functioned essentially equally at discriminating results from normal and otosclerotic middle ears.   Figure 3.6 ROC comparisons of univariate predictors with the three highest AUROC values as a function of test frequency, pressurization method, and normative data type (individual ear vs. inter-aural difference). 0204060801000 20 40 60 80 100100-SpecificitySensitivityDifference_Ambient_WBA_1600Difference_Ambient_WBA_630Difference_Peak_3DT_500Group_Peak_3DT_3150Group_Ambient_3DT_3150Group_Ambient_WBA_100053  Table 3.4. Summary of AUROCs of three best univariate predictors of otosclerosis from each type of normative data (individual ear data, inter-aural difference). Norm Type Univariate Predictor Rank AUROC p Criterion Sens Spec Individual Ear Ambient_WBA_1000 1 0.705 .0075 ≤0.39 60 79.51  Peak_3DT_3150 2 0.694 .0006 ≤0.55 46.67 89.34  Ambient_3DT_3150 3 0.679 .0014 ≤0.51 36.67 94.26 Difference Ambient_WBA_1600 1 0.708 <.0001 >0.08 78.57 62.3  Ambient_WBA_630 2 0.647 .0145 >0.23 28.57 96.72   Peak_3DT_500 3 0.646 .0185 >0.06 64.29 60.66 Notes:  Sens. = Sensitivity, Spec. = Specificity, AUROC= area under the receiver operating characteristic curve  The six predictors with the highest AUROC values across the two WBA measurements (individual ear, inter-aural WBA difference) were also compared using DeLong’s (1988) method. These univariate predictors are displayed with their criterion and associated sensitivity and specificity in Table 3.5 below. The results of the pair wise comparisons are displayed in Table A.9 and Figure 3.7; results indicated there was no significant difference between AUROCs of the any of the compared predictors (p>.05), implying that they functioned essentially equally at discriminating results from normal and otosclerotic middle ears.   54   Figure 3.7 ROC comparisons of univariate predictors with the six highest AUROC values as a function of test frequency, pressurization method, and normative data type (individual ear, inter-aural difference). Table 3.5. Summary of AUROCs of six best univariate predictors of otosclerosis across both types of normative data (individual ear data, inter-aural difference). Norm Type Univariate Predictor Rank AUROC p Criterion Sens Spec Difference  Ambient_WBA_1600 1 0.708 <.0001 >0.08 78.57 62.3 Individual Ear Peak_3DT_3150 2 0.701 .0006 ≤0.55 46.67 89.34  Ambient_3DT_3150 3 0.688 .0014 ≤0.51 36.67 94.26  Ambient_WBA_1000 4 0.685 .0075 ≤0.39 60.00 79.51  Ambient_WBA_5000 5 0.657 .0055 ≤0.49 66.67 63.93   Peak_3DT_4000 6 0.654 .0087 ≤0.5 43.33 85.25 Notes: Sens. = Sensitivity, Spec. = Specificity, AUROC= area under the receiver operating characteristic curve 0204060801000 20 40 60 80 100100-SpecificitySensitivityDifference_Ambient_WBA_1600Group_Peak_3DT_3150Group_Ambient_3DT_3150Group_Ambient_WBA_1000Group_Ambient_WBA_5000Group_Peak_3DT_400055  Chapter 4: Discussion Records of wideband absorbance of the middle ear were studied from a sample of 122 normally-hearing adults for determining normative values that could be used for evaluating clinical measures of WBA. Two types of normative data were analyzed: individual ear norms, and norms based on the inter-aural difference of WBA values measured within-subjects (inter-aural difference norms). A mixed model ANOVA approach was used to examine sources of variability (gender, ethnicity, test pressurization) that affect each type of WBA measurement. It was also within the scope of this study to compare the diagnostic accuracy of each type of WBA measure (individual ear, difference) for the detection of otosclerosis. An ROC curve analysis and statistical comparison of areas under the curves were conducted to meet this objective.  It was expected that difference norms would not depend on population-based factors ethnicity and gender, as this type of measurement is conducted within-subjects and would thereby omit slight intra-subject anatomical ME differences. Test pressurization method and frequency were also anticipated not to affect inter-aural WBA difference measures, as these parameters would be shared by the subject’s two ears, ultimately ‘cancelling out’ once the difference had been calculated. Consistent with this line of reasoning, it was anticipated that the more homogeneous difference norms would provide better separation of normally-hearing ears and ears with a predominantly unilateral disease, otosclerosis, as indexed by higher AUROC values.  The following discussion is organized into five sections: 4.1 Sources of Variability Affecting WBA Norms in this Study, 4.2 Test Performance of WBA with Either Style of Norm Applied, 4.3 Clinical Implications of Norms Investigated in the Present Study, 4.4 Study Limitations and, 4.5 The Direction of Future Research.  56  4.1 Sources of Variability Affecting WBA Norms in this Study  4.1.1 Individual Ear WBA Norms 4.1.1.1 Individual Ear WBA data Compared to Published Norms  The mean ±SD ambient WBA results of the present study are plotted in Figure 4.1 with norms published by Feeney and Sanford (2004) (n = 40; 18-28 yrs), Keefe et al. (1993) (n = 10; 20–50 yrs), Polat et al. (2015) (n = 110; 18-27 yrs), Shahnaz and Bork (2006) (n = 128; 18–32 yr), and Voss and Allen (1994) (n = 10; 18–24 yrs) to facilitate comparisons between studies. The studies by Feeney and Sanford (2004) and Keefe et al. (1993) employed custom research immittance equipment combined with Etymotic Research microphone and foam ear inserts to obtain WBR measures; calibration of these systems was similar, and was described by Keefe et al. (1992). Voss and Allen (1994) used the SYSid by Ariel Corp immittance system, combined with an Etymotic Research microphone system and ear tips; the calibration of this system was similar to that of Feeney and Sanford (2004) and Keefe et al., (1993). Shahnaz and Bork (2006) measured WBR with the Mimosa Acoustics (RMS system version 4.0.4.4) immittance system, which also uses Etymotic Research probe tips and a four-cavity calibration procedure. Polat et al. (2015) used the IMP440 module of the Titan by Interacoustics (version 3.1). With exception to Polat et al. (2015) and the present study, WBR was reported by all studies at 1/3 octave bands from 250-6000 Hz; values have been converted to WBA (1-WBR) and plotted on the same x-axis distribution for ease of comparison (See Table A.9 in Appendix C for original values).  Overall, the basic shape of increasing absorbance to 1000 Hz, with a second maxima at approximately 4000 Hz, and declining values in the higher frequencies was consistent across studies. Values were highly similar in the low and high frequencies, indicating that measurement error, such as an incomplete probe seal, did not contribute appreciably to differences observed 57  for WBA between studies (Vander Werff, Prieve, Georgantas, 2007). Differences in sample size and participant characteristics may then have contributed to the slight variability of reported values. Standard deviations increased with increasing frequencies, and were highest above 1000 Hz, indicating that WBA is more variable at higher frequencies (see Table A.9 in Appendix C).   Wideband absorbance values reported by Feeney and Sanford (2004) deviate the most from our values and those published by the other researchers. Their values were lower across the low and mid frequencies, with the greatest separation occurring from 630 to 2500 Hz, before approximating our values at 3150 Hz and above. It is unclear as to why these results were inconsistent. Though values within one SD at 250 Hz and 6000 Hz rule out measurement error (Vander Werff, Prieve, Georgantas, 2007). While the acoustic immittance system itself may have accounted for some of the variability between the studies, we would have expected greater differences between the present values and those reported by other researchers using similar systems to Feeney and Sanford (2004) (i.e. Keefe et al. (1993), Shahnaz & Bork (2006), Voss and Allen (1994)) if this were the cause. Other researchers have suggested that sample demographics, procedural differences, and probe assembly configuration may be responsible (Feeney & Sanford, 2004).  Although WBA values still resided within one standard deviation, and thus can be considered largely consistent, differences were apparent in the mid -frequency region between norms of Polat et al.’s (2015) study and the present investigation. Polat et al.’s (2015) WBA values were higher than our values at frequencies below 3150 Hz, with the most notable differences occurring from 800 to 2000 Hz. Their values were slightly lower at 4000 to 5000 Hz. Discrepancies between studies are likely caused by differences in sample demographics. Where Polat et al. (2015) measured WBA in a sample of 110 Turkish young adults, aged 18-26 years, 58  the current study included adults (18-35 years) from Asian, Caucasian and Mixed ethnicities. Shahnaz and Bork (2006) found that Chinese adults have a significantly lower absorbance in the low-frequency range, and higher WBA in the high-frequency range than their Caucasian counterparts. As such, the smaller values of low-frequency WBA, and slightly higher values of high-frequency WBA, obtained in the current study may reflect the WBA profile of the Asian group included in this study. Figure 4.2 summarizes the comparison of the current results with those of Polat et al. (2015).   Figure 4.1 Comparison of mean ±SD ambient individual ear WBA data to published norms.   00.10.20.30.40.50.60.70.80.9100 1000 10000Shahnaz & Bork, (2006) Keefe et al., (1993) Feeney & Sanford (2004)Voss et al., (1996) Current Study Polat et al., (2015)59   Figure 4.2 Comparison of ambient WBA data from the current study to norms published by Polat et al. (2015). The black dashed line and grey shaded region indicate the mean and 80% range of the current study, respectively. The dashed and solid blue lines indicate the mean and 80% range published by Polat et al. (2015), respectively. 4.1.1.2 Sources of Variability for Individual Ear Norms  This study found that the effects of ethnicity and gender were significant in their interactions with test frequency and pressurization method, and test frequency, respectively. Trends observed for these interactions were consistent with ethnicity and gender-related trends described in the literature (Shahnaz & Bork, 2006; Shaw, 2009). An explanation of the differences observed between these groups is found in observations of overall body size, as indexed by height and weight, ear canal volume, middle ear volume, and variability in the naturally occurring middle ear pressure of some participants (Shahnaz, Feeney & Schairer, 2013).  0.000.100.200.300.400.500.600.700.800.901.00250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300 8000Absorbance Magnitude (0-1)Frequency (Hz)10th Percentile Polat et al. (2015) 90th Percentile Polat et al (2015)Mean Current Study Mean Polat et al. (2015)60   The following sections discuss the effects of ethnicity and gender on individual ear WBA data in our sample of normally-hearing adults. The effect of pressurization method is discussed in the ethnicity subsection.  4.1.1.2.1 Ethnicity Effects  The mean WBA was compared among groups of Asian, Caucasian and Mixed participants. It was found that mean WBA changes across frequencies depending on the ethnicity of the subjects. This effect is further complicated by the pressurization method used to acquire WBA measurements.  Across pressurization methods, the WBA profile of Asian young adults was lower in the low-frequency range and higher in the high frequencies compared to their Caucasian counterparts. Differences between these ethnic groups reached significance for from 500-1000 Hz and 4000-8000 Hz for WBA measurements obtained using wideband tympanometry pressurization methods (Peak_3DT, Amb_3DT). These differences were less pronounced in the static ambient WBA recording (Amb_WBA), and were also restricted to narrower frequency ranges (500-800 Hz and 5000-8000 Hz). Alternatively, Asian young adults had significantly higher mean WBA than the Mixed ethnicity group from 6300-8000 Hz for all pressurization methods. The mean WBA of Caucasian young adults did not differ significantly from results obtained from the Mixed ethnicity group at any frequency or in any pressure condition, which may allude to a greater proportion of Caucasians in the Mixed/undeclared ethnicity group.  The findings of the present study corroborate Shahnaz and Bork’s (2006) findings in their study of ambient WBA in normally-hearing Chinese and Caucasian adults. In their sample, the mean WBA of the Chinese group was lower from 469-1500 Hz and higher from 3891-6000 Hz than the Caucasian group. Similar findings were reported by Shaw (2009), who also found 61  Chinese adults to have lower mean WBA values than a group of Caucasian subjects from 250-1250 Hz, and higher values from 4000 to 6000 Hz. These findings suggest that on average Asians have stiffer middle ears, causing a greater amount of acoustic energy to be reflected at low frequencies (recorded as lower WBA), compared to Caucasians. Conversely, Caucasians have a poorer high-frequency response of the middle ear compared to Asian populations, possibly due to mass effects (Shahnaz & Bork, 2006).  Explanations for differences of WBA between ethnicities are centered on body size differences observed between these groups. These explanations are derived from observations that in general, the body size of Caucasians is greater than that of Asians, as indexed by height and weight. Similarly, the body size of males is greater than that of females, as indexed by height and weight, across ethnicities (Bell, Adair & Popkin, 2002). Body size differences have been indirectly related to differences in equivalent ear canal volume as suggested by Shahnaz and Davies (2006). Slight differences in middle ear volume have subsequently been shown to have an impact on measures of WBA, in modeling research by Voss et al. (2008). In Shahnaz and Bork’s (2006) study, when the group of Caucasian females was compared to a group close in mean height and weight, Asian males, the ethnicity effect was no longer significant. All-together, these findings suggest that the differences observed between ethnicities are likely a product of normal variation in body size influencing the mechano-acoustic properties of the middle ear.  The interactions of the pressurization method and frequency, and pressurization method and ethnicity were previously analyzed by Shaw (2009). In their study, WBA measured at TPP was significantly higher at frequencies below 2000 Hz, compared to measurements taken at static (ambient) pressure for all ethnicities. These findings are not isolated, in that the present study, as well as Margolis, Saly, and Keefe (1999) and Liu et al. (2008), reported similar pressurization 62  effects. It was of note that Caucasian participants in Shaw’s (2009) study showed a greater increase in mean WBA compared to Chinese participants between pressurization methods. These findings are repeated in the current study between the Caucasian and Asian groups (Figure 3.2).  The interaction between ethnicity and pressurization method may be a product of greater variability in naturally occurring middle ear pressure observed in Caucasian adults compared to other ethnicities. Shaw (2009) included an analysis of middle ear pressure in their study, and found that Caucasian middle ears had higher variability of naturally occurring middle ear pressure. Therefore, greater differences between WBA measurements taken at ambient and TPP would be expected for these subjects, compared to other ethnicities. The reason for the shift is that an inherent assumption of recordings taken in a purely ambient setting is that maximum energy transfer of the middle ear occurs at 0 daPa; however, if natural middle ear pressure deviates from this value, the ME would absorb energy more efficiently at the pressure corresponding to TPP (Feeney et al., 2014).  A final point pertaining to the interaction of pressurization, frequency and ethnicity observed in this study is of note. WBA measured at static ambient pressure differed significantly from the ambient measure extracted at 0 daPa from the 3DT tab for all ethnicity groups. The trend of lower low-frequency (250-2500 Hz) WBA and higher high-frequency (4000-6300 Hz) WBA obtained through measurements at static ambient pressure implied more ME stiffness was observed in this setting. Among ethnicity groups, the frequencies affected and magnitude of change differed, however the trend was consistent.  The observed differences to WBA between the two apparently ambient pressurization methods could possibly be explained by two factors. Firstly, as suggested by Liu et al. (2008), slight positive pressure induced in the ear canal via compression of air with probe tip insertion 63  could cause a stiffening effect, simulating negative middle ear pressure (Sun et al., 2012). This minor pressure deviation would be uncompensated for in static WBA measures, as this pressurization method makes no accommodation to equalize ear canal pressure with the surrounding medium once the probe is in place (Interacoustics, 2017). By contrast, ambient WBA obtained with WBT is retrieved from the point in the pressure sweep corresponding to 0 daPa relative to the surrounding medium, which according to the Titan Instructions for Use document (2017), “result[s] in a measurement at zero pressure.”   A difference in the pressure environment present in the ear canal between these two measures would be an indication that separate normative values are essential for WBA obtained under these two conditions. Alternatively, if a difference in pressure is responsible, equalizing ear canal pressure with the surrounding medium prior to starting the measurement in the wideband absorbance tab measures could eliminate the differences observed between the ambient conditions (Liu et al., 2008), precluding the requirement for separate norms. Future WAI research should investigate the underlying cause of differences to WBA observed between ambient measurements obtained with WBT and static pressure methods.  The second explanation is that there could be hysteresis, or lagging, of the pressurization effects on the TM through the tympanic pressure sweep. A WBA measure extracted at 0 daPa, as referenced to the surrounding medium, then encompasses the effects of the ear canal pressure environment immediately preceding the ambient value. Moreover, continuous motion of the TM through the process of changing ear canal pressure implies that WBA retrieved from the pressure sweep at 0 daPa shows the absorbance of the ME as the TM is motile. Consequently, WBA extracted at 0 daPa in a dynamic WBT sweep may be expected to exhibit less stiffness than static ambient measures of WBA obtained through the wideband absorbance tab of the Titan IMP440 64  module. In a study that investigated the effects of WBT trials on WBA measures of the ME, Burdiek and Sun (2014) reported that repeated pressurization of the ear canal yielded a decrease in TM stiffness. Changes observed as an increase in mean WBA below 1500 Hz and a decrease in mean WBA around 2000 and 5000-6000 kHz were largely consistent with the differences observed between WBA obtained at static ambient pressure and with WBT in this study.   Taken together with findings reported in the literature, the significant interaction of frequency, ethnicity and pressurization methods in the present study inform us of two points. Firstly, the use of separate norms for different ethnic populations is advisable, given the frequency-specific changes to WBA for participants of different ethnic descent. Secondly, differences between ethnicities are more salient when WBA is measured using WBT (pressurized) measurements; this emphasizes the importance of using pressure-specific normative data, especially for some ethnicities (i.e. Caucasians). These findings may also be taken as support for recommendations from Feeney et al. (2014) and Sanford et al. (2013) that measurements of WBA should be taken at both ambient pressure and TPP.  4.1.1.2.2 Gender Effects  The effect of gender on individual ear WBA values was examined by comparing mean WBA obtained from male and female subjects. Our findings indicated that there is a significant interaction between test frequency and gender, wherein males had a trend of higher WBA values than females at low frequencies, and females had higher mean WBA in the high frequencies. These differences reached significance at 4000-5000 Hz, where the WBA of females was significantly higher (Figure 3.3).  The effects of gender on WBA have been previously been studied by Feeney and Sanford (2004), Shahnaz and Bork (2006), Mazlan et al., (2015), and Polat et al. (2015). The trends 65  observed in the present study are consistent with results reported by these researchers. For example, in their sample of 40 normally-hearing young adults, Feeney and Sanford (2004) found that females had WBA values that were 10% lower than males at 794 and 1000 Hz, and 18% higher than males at 5040 Hz. Mazlan et al. (2015) similarly found that females had a trend of slightly, but significantly, lower WBA than males at frequencies below 1000 Hz. This pattern was reversed from 2830 to 4490 Hz, where females had significantly higher mean WBA values than males. Polat et al. (2015) found that Turkish males have significantly lower WBA from 3100 to 6900 Hz than females, though no differences were observed between groups at lower frequencies. Researchers have suggested that gender-related WBA findings corroborate earlier acoustic impedance findings, which described males as having middle ear systems that are less stiffness-dominated than females below 1000 Hz (Margolis, Saly & Keefe, 1999; Mazlan et al., 2015). More specifically, the trend of lower absorbance values in the low frequencies, and higher absorbance in the high frequencies is consistent with females having relatively stiffer middle ears than males (Mazlan et al., 2015).  As discussed in the previous section, Shahnaz and Bork (2006) and Shahnaz and Davies (2006) posited that gender- and ethnicity-related differences in acoustic immittance profiles may reflect differences in body size between these groups, as indexed by equivalent ear canal volume, middle ear cavity volume, height, and weight. In their (2006) study Shahnaz and Bork compared mean WBA of Chinese males and Caucasian females, as these groups shared more similar body proportions (height and weight), than males and females within ethnicity groups. Though not eliminated, the gap between the low-frequency WBA of males and females was reduced considerably with this comparison. As such, gender-related differences may also be attributed at least in part, to differences in body size. Other contributing factors may include ear the cross-66  sectional area of the ear canal, head circumference, or middle ear volume (Jaffer, 2016); though this is a consideration for future WAI research.  4.1.2 Inter-Aural Difference WBA Norms  4.1.2.1 Comparison of Inter-Aural Difference Data to Published Norms Mean inter-aural difference values in this study were consistent with values reported by other researchers, though this data should be interpreted with caution. Without converting the measured between-ear differences to absolute values prior to calculating the mean, our results showed that values measured at ambient pressure (Amb_WBA) ranged from -0.028 to +0.023 (Figure 4.3). The SDs about these differences ranged from 0.034 to 0.152, with larger SDs at the higher frequencies. Our values were well within the -0.041 to +0.044 range (±SDs of 0.04 to 0.24) reported by Rosowski et al. (2012) for a sample of 58 ears, and was consistent with Werner et al.’s (2010) reported values of ±0.02 for a sample of 210 adults. The relatively lower SDs reported in this study indicate that our sample was less variable than that of Rosowski et al. (2012). The consistent trend of increasing SD with increasing frequency between studies implies that inter-aural difference measures of WBA are more variable at higher frequencies.  The data may be misrepresented by taking the mean of non-absolute difference values. The mathematical operations may confound the specific stiffness relationship between the ears observed for some individuals in the sample with the actual difference of the WBA values measured between ears. To calculate the inter-aural difference, the WBA data from a participant’s one ear is subtracted from the contralateral ear across frequencies (right minus left in this study). The right ear may have a lower WBA value than the left, resulting in a negative difference. Conversely, the opposite situation can occur resulting in a positive difference. This relationship can vary across frequencies for one participant’s two ears – positive difference at 67  some frequencies, and negative at others. As was observed in our raw data, this relationship also varies for difference values measured from different participants, so that the trend of positive versus negative differences is not the same at each frequency across all participants. This finding has also been alluded to by the contrasting trends observed for frequency by test ear interactions reported in the literature (Feeney et al., 2014; Feeney & Sanford, 2004; Rosowski et al., 2012; Werner et al., 2010).  If the trend was consistent within frequency bins, averaging the differences without first observing the absolute values would not be problematic, and may be informative about the relationship of mechano-acoustic properties between the ears. However, because it is not, the positive or negative value of the average can imply a stiffness relationship between the ears that is misrepresentative of what was observed for most individuals in the population. This is especially relevant if outliers have not been eliminated, as was the case in the Rosowski et al. (2012) paper and the current investigation. The current study therefore elected to analyze absolute values of the inter-aural differences, as they more reliably represent the separation (i.e. actual difference) of WBA values obtained from each ear of a participant. Note the decision not to remove the outlying values for inter-aural difference norms is acknowledged as a limitation of the current study, and is justified in section 4.4. 68   Figure 4.3 Mean ± SD of inter-aural difference data in this study (not absolute values). 4.1.2.2 Sources of Variability for Inter-Aural Difference Norms  Our results showed that overall, absolute between-ear differences of WBA were very small (0.02-0.11). Differences were smallest at the low frequencies (about 0.02 at 250 Hz) and increased steeply with rising frequency to 800 Hz (values of about 0.09). In the mid- to high-frequencies, the general upward trend for difference values continued, though with a greater fluctuation and a gentler slope (Figure 3.4). Inter-aural difference values of WBA obtained with the Amb_WBA pressurization method had significantly higher values from 1000-1250 Hz, compared to results obtained using WBT pressurization methods (Peak_3DT, Amb_3DT). Although these differences were significant, they were small (0.02 on average), and likely do not hold any clinical relevance. Inter-aural difference values of WBA did not differ between gender or ethnic groups, indicating that the same norms can be used across these sub-populations. -0.20-0.15-0.10-0.050.000.050.100.150.20200 2000Peak_3DT Amb_3DT Amb_WBA69   The frequency-related changes to between-ear difference values of WBA were unexpected. Given our individual ear WBA findings, which indicated no significant effects or interactions of test ear (Figure 4.4), a relatively flat contour across frequencies for difference values was anticipated.   Figure 4.4 Mean individual ear WBA data as a function of the pressurization method and frequency (Hz). Vertical bars denote 95% confidence intervals.  Differences in naturally occurring middle ear pressure between the two ears of a subject, and increasing variability of WBA measures with rising frequency are probable contributors to the significant frequency effects in our results. Natural fluctuations in ME pressure occur in subjects with normal Eustachian tube function. The actions of yawning and swallowing coincide with the momentary opening of the Eustachian tube; this allows for equalization of pressure of the ME with the environment, and consequently temporary fluctuations in ME pressure (Robinson, Thompson & Allen, 2016). Few studies have quantified the degree of between-ear  Right LeftPeak_3DT250400630100016002500400063000.00.10.20.30.40.50.60.70.80.91.0Absorbance Magnitude (0-1)Amb_3DT25040063010001600250040006300Amb_WBA2504006301000160025004000630070  variation of ME pressure of normally-hearing adults, however, the incidence of unilateral non-pathological negative ME pressure in children suggests ME pressure may fluctuate in each ear independently (Lindholt, 1980).  Margolis, Saly, and Keefe (1999) reported that WBA values are systematically influenced by non-ambient middle ear pressures that occur in normally-hearing adults. Robinson, Thompson, and Allen (2016) found that even minor deviations in middle ear pressure (65 daPa) around ambient resulted in obvious changes to WBA recorded from normally-hearing subjects. In their study, negative pressure related changes were most pronounced from 1.0-1.9 kHz, though were significant from 0.6-2 kHz, when subjects induced ME pressure shifts with the Toynbee manoeuvre. Published work from Shaver and Sun (2013) has also demonstrated the effect of ME pressure on WBA recordings from normally-hearing adults. In their study, Toynbee maneuver induced negative ME pressure exerted its greatest effects on WBA at 1.0-1.5 kHz and 4.5-5.5 kHz. In the mid-frequencies mean WBA was significantly decreased, whereas significantly higher values were obtained in the high frequencies. The changes are consistent with a stiffening of the middle ear system, as described by Margolis, Saly, and Keefe (1999). A pertinent finding of Shaver and Sun’s (2013) study was that WBA of negative-pressure induced conditions recorded using the pressurization methods of wideband tympanometry closely resembled baseline (non-manipulation) conditions. As such, measurements made with WBT mitigated the effect of ME pressure variation on WBA.  The ME pressure effects reported by Robinson, Thompson, and Allen (2016) and Shaver and Sun (2013) are of relevance to the present findings. The frequency-dependent effect of ME pressure on WBA reported in these studies appears to corroborate the greater between-ear WBA difference at mid- and high-frequencies observed in our results. Further, greater differences from 71  1000-1250 Hz obtained in the Amb_WBA recording not only fall into the most affected regions reported by Shaver and Sun (2013), but were also similarly diminished with WBT pressurization methods. Liu et al. (2008) have noted that compression of the air in the ear canal caused by probe tip insertion may induce a slight amount of positive pressure, which is uncompensated for in ambient WBA recordings. Sun et al. equate induced positive ear canal pressure to naturally occurring negative ME pressure results in their (2012) study. As such, slight stiffening of the ME coinciding with probe tip insertion in ambient recordings is a probable cause of the interaction of frequency and pressurization method observed in our results.  As seen in Figure 3.4, between-ear differences were less pronounced for measures of WBA obtained with WBT. These conditions provide compensation of probe-tip induced pressure, as well as ME pressures varying from environmental values (Interacoustics, 2017). Given our results, this action appears to minimize the effects of slight deviations of ME pressure between ears. Therefore, obtaining WBA under pressurized conditions may be important for reducing the confound of a slight variation of middle ear pressure, even for within-subject measures.  The greater high-frequency variability observed for WBA measures may also have contributed to our results. Vander Werff, Prieve, and Georgantas (2007) analyzed the test-retest reliability of WBA measures in 10 normally-hearing adults with and without probe-tip reinsertion between trails. Although frequency effects did not reach significance, a trend of increasing variability with frequency was evident above 1000 Hz for reinsertion trails, as indexed by the mean and 90% range. Vander Werff, Prieve, and Georgantas (2007) reported that differences between trials were 0.05 or less; visual inspection of their results shows that at 250 Hz differences between trials were closer to 0.02 and increased to 0.05 across test frequencies. 72  These findings are in line with published normative data, which show greater variability of WBA across subject ears at frequencies above 1000 Hz, as shown by larger standard deviations (Feeney et al., 2004; Keefe et al., 1993; Polat et al., 2015; Shahnaz & Bork, 2006; Voss & Allen, 1994). Therefore, the trend of increasing inter-aural difference of WBA with frequency in our results may be attributed, in part, to greater variability of WBA measures at these frequencies in general.  The relatively higher inter-aural difference values of WBA obtained in ambient recordings (Amb_WBA) do not appear to have a clinical impact, at least for the detection of otosclerosis. Confirmation of this was given by the statistical comparison of the AUROC values of WBA obtained with each pressurization method (Amb_WBA, Amb_3DT, Peak_3DT) at 1000 and 1250 Hz using DeLong’s (1988) method (Figure 4.5). Pairwise comparisons yielded no significant differences (p>.05) between the AUROCs of any of the variables, showing that they functioned equally at discriminating between normally-hearing and otosclerotic middle ears. Therefore, we conclude that although inter-aural differences of WBA were significantly higher from 1000-1250 Hz when measured with the Amb_WBA pressurization method, the use of separate norms for each pressurization method is unnecessary, as the differences observed do not influence the clinical outcomes of these measures.  73   Figure 4.5 ROC comparison of WBA measured with the Amb_WBA, Amb_3DT, and Peak_3DT pressurization methods from 1000-1250 Hz.  0204060801000 20 40 60 80 100100-SpecificitySensitivityDifference_Ambient_WBA_1000Difference_Ambient_WBA_1250Difference_Ambient_3DT_1000Difference_Ambient_3DT_1250Difference_Peak_3DT_1000Difference_Peak_3DT_125074  Table 4.1. Pairwise comparisons of AUROCs of WBA measured with the Amb_WBA, Amb_3DT, and Peak_3DT pressurization methods from 1000-1250 Hz, applying difference norms. AUROC Plot AUROC SE 95% CI   Difference_Ambient_WBA_1000 0.537 0.0645 0.454 to 0.619   Difference_Ambient_WBA_1250 0.572 0.0616 0.488 to 0.652   Difference_Ambient_3DT_1000 0.581 0.0676 0.497 to 0.660   Difference_Ambient_3DT_1250 0.592 0.0654 0.509 to 0.672   Difference_Peak_3DT_1000 0.63 0.0604 0.547 to 0.707   Difference_Peak_3DT_1250 0.613 0.0597 0.530 to 0.691   Pairwise Comparison of AUROC AUROC Difference SE 95% CI Z-statistic p Difference_Ambient_WBA_1000 ~ Difference_Ambient_WBA_1250 0.0345 0.0653 -0.0934 to 0.163 0.529  .597 Difference_Ambient_WBA_1000 ~ Difference_Ambient_3DT_1000 0.0433 0.066 -0.0861 to 0.173 0.656 .512 Difference_Ambient_WBA_1000 ~ Difference_Ambient_3DT_1250 0.0549 0.0605 -0.0637 to 0.173 0.907 .364 Difference_Ambient_WBA_1000 ~ Difference_Peak_3DT_1000 0.0927 0.0623 -0.0295 to 0.215 1.487 .137 Difference_Ambient_WBA_1000 ~ Difference_Peak_3DT_1250 0.0757 0.0576 -0.0372 to 0.189 1.314 .189 Difference_Ambient_WBA_1250 ~ Difference_Ambient_3DT_1000 0.00878 0.072 -0.132 to 0.150 0.122 .903 Difference_Ambient_WBA_1250 ~ Difference_Ambient_3DT_1250 0.0203 0.0551 -0.0876 to 0.128 0.369 .712 Difference_Ambient_WBA_1250 ~ Difference_Peak_3DT_1000 0.0581 0.0739 -0.0868 to 0.203 0.786 .432 Difference_Ambient_WBA_1250 ~ Difference_Peak_3DT_1250 0.0411 0.0569 -0.0704 to 0.153 0.723 .470 Difference_Ambient_3DT_1000 ~ Difference_Ambient_3DT_1250 0.0116 0.0577 -0.102 to 0.125 0.200 .841 Difference_Ambient_3DT_1000 ~ Difference_Peak_3DT_1000 0.0493 0.0498 -0.0484 to 0.147 0.990 .322 Difference_Ambient_3DT_1000 ~ Difference_Peak_3DT_1250 0.0323 0.0689 -0.103 to 0.167 0.469 .639 Difference_Ambient_3DT_1250 ~ Difference_Peak_3DT_1000 0.0378 0.057 -0.0740 to 0.150 0.662 .508 Difference_Ambient_3DT_1250 ~ Difference_Peak_3DT_1250 0.0208 0.0322 -0.0423 to 0.0839 0.646 .519 Difference_Peak_3DT_1000 ~ Difference_Peak_3DT_1250 0.017 0.0549 -0.0906 to 0.125 0.309 .757 75  4.2  Test Performance of WBA with Either Style of Norm Applied Novel contributions of this WBA study included the presentation of a new style of normative data, which is more resilient to natural sources of intra-subject variability. Inter-aural differences describe the absolute difference of WBA obtained from the two ears of each subject; these measures omit sources of variability introduced to aggregate data by intra-subject measurements. We hypothesized that the use of an inter-aural difference measure of WBA could lead to improved diagnostic accuracy of this tool for the detection of otosclerosis for two reasons. Firstly, greater homogeneity of the difference norms (as shown by fewer sources of variability) could provide greater separation between distributions of normally hearing and otosclerotic ears, in a similar manner to reports by Shahnaz and Bork (2006). Secondly, otosclerosis has a predominantly unilateral onset (Foster & Backous, 2018); for this reason, even if both ears resided within the wide normative range for individual ear data, they may exceed normal values in an absolute difference measure.  Analysis of ROC plots showed that WBA with either style of normative data applied predicted the presence of otosclerosis above chance levels. In contrast to our hypothesis, WBA test performance for the detection of otosclerotic ears was equal for either style of WBA measure, as indexed by insignificant results of AUROC comparisons. It was therefore concluded that the increased homogeneity of inter-aural difference norms did not ultimately provide better separation between distributions of normally hearing and otosclerotic middle ears.  Visual inspection of the plot of otosclerotic and normally-hearing WBA (individual ear data) showed that mean WBA in the otosclerotic ears trended toward lower values from 800-5000 Hz for all pressurization methods (Figure 4.6). This differs from previous research which found subnormal WBA for otosclerotic ears predominantly below 1000 Hz (Nakajima et al., 76  2012; Shahnaz et al., 2009); however, is in line with reports from other researchers which showed these groups differ at frequencies below 2000 Hz (Merchant et al., 2016; Voss et al., 2012), and up to 4000 Hz (Keefe et al., 2017). By contrast, mean absolute inter-aural WBA difference values for otosclerotic ears were greater than values obtained from the normal group across all frequencies (Figure 4.7). WBA results from our sample of otosclerotic ears was comparatively more diverse than other samples reported in the literature (Shahnaz et al., 2009), which may have contributed to relatively lower AUROC values obtained in this study.   For individual ear data, WBA of otosclerotic ears fell outside of the 80% normative range in 28 of 30 ears (26 participants), although the pattern at which abnormal values were observed was not consistent (Figure 4.6). In most otosclerotic ears (21 of 30; 70%) WBA was lower than the 10th percentile of normally-hearing ears across frequencies; 14 of the 30 (47%) demonstrated subnormal values below 2000 Hz, whereas for the remaining 7 (23%) differences were observed at higher frequencies. This finding differs from those reported by Shahnaz et al. (2009), who analyzed a comparably sized sample (n=28) of otosclerotic ears; their results showed that 71% of otosclerotic ears had subnormal values below 1000 Hz, and only 10% of their sample had subnormal values at frequencies above 1500 Hz.  Modeling research and cadaveric studies have demonstrated that subnormal WBA values below 2000 Hz obtained from ears with stapes fixation is related to an increase in middle ear stiffness (Feeney & Keefe, 1999; Merchant et al., 2016; Voss et al., 2012), wherein the contribution of the annular ligament has been specifically implicated (Allen et al., 2005). Experimentally derived explanations for differences observed above 2000 Hz are lacking, despite this finding being relatively common (Keefe et al., 2017; Voss et al., 2012). However, Keefe et 77  al. (2017) suggested these differences “may be due to functional differences in ossicular- chain transmission at frequencies above the dominant resonance of the TM.”  Seven of 30 otosclerotic ears (23%) were characterized values that exceeded the 90th percentile of normal. This finding violates the expected changes to WBA caused by a disease that has been demonstrated to increase stiffness, as it implies that more incident energy has flowed into the ME system than is typically observed for normally-hearing ears. It is however, consistent with findings of Shahnaz et al. (2009), who found that 18% of their otosclerotic sample had elevated WBA. In their study, the group of otosclerotic ears with elevated WBA also had abnormally sharp tympanometric peaks as indexed by narrow tympanic width, implying decreased stiffness. The findings of the present study and those of Shahnaz et al. (2009) corroborate those of Zhao et al. (2002), who described stiffness-related sub-classifications of otosclerosis in their sample of 36 ears – those with elevated, reduced and normal middle ear stiffness, which may correspond to stages of the disease progression. Altogether comparison of our clinical sample with those of other researchers has shown otosclerosis presents in a more diverse way that has been implied by the existing WAI literature; further research into the physiological changes induced by otosclerosis, and their corresponding impact on WAI parameters is needed.  To objectively assess the test performance of WBA with either type of norm applied (individual ear, inter-aural difference), ROC curves were generated at center frequencies of 1/3 octave bands from 250-8000 Hz. AUROC values ranged from 0.501 to 0.708 and 0.501 to 0.701 for WBA univariate predictors of otosclerosis based on inter-aural difference norms and individual-ear norms, respectively. The AUROC of several univariate predictors of otosclerosis from each type of WBA measure were significantly greater than 0.5 (Table A.7 in Appendix B  . 78  This finding indicates that both inter-aural difference and individual ear measures of WBA are effective at distinguishing the results of normally-hearing and otosclerotic ears. In contrast to our research hypothesis, neither individual ear nor inter-aural difference measures of WBA out-performed the other. Pairwise comparisons of AUROCs of the three most successful univariate predictors of otosclerosis (as indexed by highest ranked AUROC values) based on either type of normative data yielded insignificant differences (p>.05) (Table A.8.). The same result was obtained for pairwise comparisons of the 6 most successful univariate predictors of otosclerosis across normative data type (Table A.9.). Based on this finding alone, there would appear to be no clinical motivation for selecting either style of WBA measurement over the other. However, results from 26 of the total 28 otosclerosis patients fell outside of the 80% normative range for each measure, at least at one frequency. As seen in Table 4.2, inter-aural difference measures identified a patient that would otherwise have been missed with individual ear measures and vice versa. This finding suggests the two types of WBA measure may be combined to achieve higher diagnostic performance for the detection of otosclerosis than either measure in isolation.  Turning to other comparable literature that has investigated the diagnostic performance of WAI for otosclerosis; Shahnaz et al. (2009) reported relatively higher AUROC values than the present study. Shahnaz et al. (2009) generated ROC plots at 1/3 octave band frequencies from 315 Hz to 1000 Hz to evaluate diagnostic performance of ambient WBR in its ability to distinguish results from 62 normally-hearing ears and 28 ears with otosclerosis. AUROC values in their study ranged from 0.67 to 0.86, wherein WBR at 400 Hz and 500 Hz shared the highest AUROC value, WBR at 1000 Hz had the lowest, and the other frequencies varied in between. 79  These values exceeded the AUROCs of 0.501 to 0.701 obtained for individual ear WBA data in the present study considerably.  Methodological differences between Shahnaz et al.’s (2009) study and the present investigation may have contributed to the relatively poorer test performance observed for WBA in this study. Shahnaz et al. (2009) curtailed their normally-hearing subjects to include records from Caucasian subjects only, the justification being that their otosclerotic sample was predominantly Caucasian. Upon examination, however, 29% of the otosclerotic ear data in their study was provided by a combination of Chinese, Hispanic, East Indian and Filipino participants. The use of exclusively Caucasian records in the control group would increase the apparent test performance of WBR in their study, for the same reasons used by Shahnaz and Bork (2006) to justify the use of ethnically-homogeneous norms for WAI. Namely, the trend of lower low-frequency WBR (higher WBA) observed for Caucasian subjects would provide a greater degree of separation from results of otosclerotic ears, which were shown in their study to have abnormally elevated WBR (reduced WBA) from 400 to 1000 Hz, than would likely be observed if the ethnic representation in the normal group had been more diverse. By contrast, the present study included results of Asian, Caucasian and Mixed/undeclared ethnicity participants in the normative sample for the ROC analyses, as 40% of our otosclerotic group reported their ethnicity as Asian, Mixed, or did not declare their ethnicity. The Asian and Mixed WBA data in trended toward lower values of WBA in the low and mid frequencies, and may thereby have simulated the WBA results of most the otosclerotic ears, ultimately lowering diagnostic performance of WBA.  WBA test performance in the present study more closely resembles values reported by Keefe et al. (2017), who also calculated the diagnostic performance of univariate predictors of 80  otosclerosis across a wide frequency range. Contributing to their analysis were 13 ears with otosclerosis, and 23 normally-hearing ears of undisclosed ethnicity. These researchers investigated several WBA variables, including ambient WBA, group delay, WBA at TPP, as well as the difference between WBA measured at TPP and the positive (+200 daPa) and negative (-300 daPa) tails of the tympanometric pressure sweep. ROC analyses yielded AUROCs that ranged from 0.5 to 0.87, wherein comparable variables of ambient WBA at 4000 and 2800 Hz were among the most successful predictors of otosclerosis. The AUROC values of these variables were approximately 0.76 and 0.75, respectively (note: AUROCs were reported graphically). The highest univariate AUROCs in their study were, however, obtained by ambient group delay at 1400 Hz (0.80), and the difference between WBA measured at TPP and the positive tail of the pressure sweep at 2.8 kHz (0.87), suggesting that the alternative variables may be more informative for the detection of otosclerosis. It is unclear as to why test performance of this study was more similar with those of Keefe et al.’s (2017) study, although the similar methodological practice is a consideration.  The test frequencies at which the best WBA test performance was obtained were inconsistent across studies. Shahnaz et al. (2009) described optimal ambient WBA test function at 400 and 500 Hz, though they did not test frequencies above 1000 Hz. Keefe et al. (2017) reported that ambient WBA performed optimally at 2800 and 4000 Hz. The present study found that WBA test performance, when considering individual ear data, was best at 3150 Hz (measured at ambient and TPP). Each study tested a different population of otosclerotic ears, which were unanimously described as having highly variable WBA results. It is possible, therefore, that the ‘best univariate predictor of otosclerosis’ pertaining to WBA test frequency, does not generalize across samples of otosclerotic ears due to the diverse nature of the disease 81  presentation. However, WBA as a wideband measure is effective at distinguishing the results of otosclerotic and normally functioning middle ears, as indexed by near perfect sensitivity and specificity (Shahnaz et al., 2009). Keefe et al. (2017) addressed this issue in their paper by analyzing the use of multivariate predictors of otosclerosis, with promising results; these findings are discussed in section 4.5.  Table 4.2 Otosclerosis patient results which were identified as abnormal by ambient WBA results that were outside the 80% normative range for either individual ear measures or inter-aural difference measures, and a battery combining both measures. Participant Individual Ear Inter-Aural Difference Combination Oto_1 + + + Oto_2 + + + Oto_3 - - - Oto_4 + + + Oto_5 + - + Oto_6 + + + Oto_7 + + + Oto_8 + + + Oto_9 + + + Oto_10 + + + Oto_11 - + + Oto_12 + + + . . . . . . . . . . . . Oto_28 + + + Total Identified 26 26 27  82   Figure 4.6 Mean and individual ear ambient WBA results for the otosclerosis group (n ears=30) plotted with the normative data. The grey shaded region shows the 80% normative range.    Figure 4.7 Mean and individual absolute inter-aural difference ambient WBA results for the otosclerosis group (n=28) plotted with the normative data. The grey shaded region shows the 80% normative range.   -0.200.000.200.400.600.801.00250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300 8000Mean_Normals Mean_Oto0.000.100.200.300.400.500.60250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300 8000Mean_Normals Mean_Oto83  4.3 Clinical implications of the Current Findings  The present study’s findings have several implications for the clinical use of WBA. Starting with individual ear data, our finding that WBA depends on the ethnicity and gender of participants, as well as the pressurization method used to obtain the measurement, suggest there is cause for parameterization of norms based on these characteristics. As ME status of participants was determined to be normal by other means of assessment, the observed differences are likely a result of natural anatomical/physiological variation between these groups. Accordingly, the use of separate norms is advised to avoid falsely identifying an ear as abnormal, or misidentifying an abnormal result as falling within the normal range, when the result simply relates to one of these parameters.  Moreover, our data has shown that the magnitude of pressurization method related changes to WBA are greater for Caucasians than other ethnicities. It may then be more important to observe normative data specific to pressurization method for Caucasian patients. As discussed by Liu et al. (2008) and Sanford, Hunter, Feeney and Nakajima (2013), measuring WBA at both TPP and ambient pressure may increase the specificity of WBA by clearly showing the contributions of ME pressure to observed patterns. Abnormal WBA patterns at ambient pressure that are within normal range at TPP may attributed to ME pressure that deviates from the surrounding medium. Though if the abnormality persists at TPP, it may be indicative of an underlying pathology, where the unique WBA contour gives insight as to the etiology (Nakajima et al., 2013). Considering this, the greater disparity between WBA measured at ambient and TPP observed for Caucasians could make this group more susceptible to test errors if non-pressure specific norms are used.  Turning to the inter-aural difference data presented in this study, the resilience of this style of measurement to population-based factors ethnicity and gender suggest it could be a more 84  constant supplement to traditional individual ear norms. Having norms for the inter-aural difference of WBA may be particularly helpful for the interpretation of results where pathology/abnormality is indicated in one ear only. Situations may arise in which inconclusive findings have been obtained for individual ear WBA; results of both ears are within normal range, though there is clear separation between values obtained from the right and left ears, and the patient states the ears feel ‘different’. In referring to inter-aural difference norms, it can be determined whether the differences are within or outside of normal range for this measure. If abnormal, a diagnostic approach that considers the possible etiology, referring to the results of other tests in the audiological battery, is necessary. For example, this finding in the absence of ABGs, and absent acoustic reflexes, may indicate a poor probe seal/slit leak. However, if conductive hearing loss (as indexed by ABGs) is present, an abnormal inter-aural difference could be consistent with the onset of a unilateral pathology, or bilateral pathology that is more severe on one side.  The use of inter-aural difference measures of WBA may also be helpful for interpreting data obtained from patients for which population-based norms are not yet available. It has been well established that WBA norms based on individual ear data depend on the ethnicity of the participants (Jaffer, 2016; Shahnaz & Bork, 2006; Shaw, 2009). However, most studies of adult populations have looked at the WBA patterns of Chinese and Caucasian participants, or Asian and Caucasian participants. Normative studies evaluating the expected WBA patterns for other ethnicities are lacking. Consequently, norms that use the patient’s contralateral ear as a reference may help to differentiate results due to natural ethnicity-based anatomical variations, and those caused by the onset of middle ear pathology, which have not been identified by other elements of the audiological test battery.  85  This brings about another important consideration for the discussion of clinical usage of WBA: immittance testing is only part of an audiological test battery, and the results must be interpreted in conjunction with those of the other tests. Nakajima et al. (2013) pointed out the rarity of normally-hearing ears that present with ABGs, absent MEMRs, and abnormal WBA findings. Thus, the test performance of WBA alone does not determine clinical outcomes. The more realistic clinical pursuit is to differentiate the etiology of conductive hearing losses, such so that a pre-surgical diagnosis may be obtained (Nakajima et al., 2013).  4.4 Study Limitations This section acknowledges the limitations of this research.  Limited Ethnic Representation   Retrospective analysis of previously collected WAI data limited the ethnic groups to those included in the studies that served as data sources for the present investigation (Asian, Caucasian, Mixed). Studies in pediatric populations have shown significant variability between WBA obtained from Caucasian and First Nations and Metis neonates (Abbott, 2018), Caucasian and Australian Aboriginal neonates (Aithal et al., 2014), and Caucasian and Chinese school-aged children (Beers et al., 2010). Given the diverse nature of North American society, and findings of improved test performance of WBA with ethnicity-specific norms, it would be of value to include a greater variety of ethnicities in normative WBA studies.  Limited Access to Information Regarding Variables that Differ between Ethnic and Gender Groups A further consideration is to identify the underlying cause of WBA differences observed between ethnicity and gender groups. As discussed by Manly (2005), basing distinctions between experimental groups on the factors such as ethnicity and gender is not a truly scientific practice. 86  Grouping individuals based on the variables that underlie differences between groups (e.g. sex hormones, height, weight, melanin production) is more informative, and avoids the social connotations of drawing these distinctions. Current WAI research has not yet identified the all the relevant parameters which underlie gender and ethnicity related effects on WBA, however, body size indices have been implicated (Shahnaz & Bork, 2006; Shahnaz & Davies, 2006).  Given the retrospective nature of this study, access to body size indices for the records analyzed was restricted to those collected by the original researchers. Only one of the data sources included a measure of body mass index (BMI), which would have served this purpose. However, BMI was analyzed as a covariate in the individual ear WBA data analysis in that study, and was found not to have an impact on the observed ethnicity and gender effects (Jaffer, 2016). As such, body size indices were not analyzed as part of the present study, though the researcher acknowledges that this is an important consideration for future WAI research.  Calibration/Data Collection Procedures    While all data analyzed in the present study was collected using the Titan by Interacoustics, there were modifications to the system that occurred in the time span that the original studies were conducted. The most notable change was the update to the calibration procedure. The update, researched by Nørgaard et al. (2017), involved placing the titan probe assembly in the waveguide unit aperture without the ear tip attached, as opposed to including the ear tip. The change to the calibration procedure likely accounted for some of the variability observed in our results.  Age of the Otosclerosis Sample Compared to the Normal Sample   The age range observed for the otosclerosis group (30-77 yrs) overlapped minimally with that of the normally-hearing group (18-35 yrs). It could then be argued that changes observed 87  between WBA patterns for each group (not statistically compared) could have been, at least partially, age-related. Maturational anatomical and physiological changes have been documented for the structures of the middle ear through the adult lifespan; Ruah et al. (1991) reported that TM becomes thinner, with a progressive decrease in vascularity, cellularity, and elasticity over time. The resulting increase in rigidity would be expected to correspond with immittance findings indicative of increased ME stiffness, such as decreased static admittance on 226-Hz tympanograms and/or decreased low-frequency WBA. However, the few studies that have measured WBA longitudinally have not shown an age effect (Feeney et al., 2014). Moreover, cross-sectional WBA studies examining the effects of age have reported higher low-frequency WBA for older groups, suggesting ME stiffness is lesser in this group (Feeney & Sanford, 2004; Mazlan et al., 2015). This is opposite to what was observed for most of the otosclerotic group in this study (70%). For this reason, it is unlikely that our results would have changed had we compared the otosclerotic group to a normally-hearing group closer in age.  Inclusion of Outliers in Inter-Aural Difference Norm Dataset  The current study based its inclusion criteria on the individual ear data collected by other researchers. If the participant was indicated to have met inclusion criteria for the normal group binaurally, their records were included in both the individual ear and inter-aural difference analyses of this study. Upon observing the raw data, records of three participants appeared to have values that were much greater than the rest; their values fell outside of the interquartile range for normals, labelling them as outliers. The records were not eliminated, as preliminary analyses indicated that removal of these subjects would not impact measures of central tendency, or outcomes of the ROC. Inclusion of this data may, however, have contributed to the width of 88  the 80% normative range reported for inter-aural difference norms in Figure 4.7, which could potentially impact results of future retrospective studies using this data.  4.5 The Direction of Future Research  Other WBA Parameters   The current study focused on normative values for wideband absorbance at ambient and TPP, for individual ear and inter-aural difference measures. The AUROC values for the detection of otosclerosis, while primarily reported to compare the efficacy of the two WBA measures presented herein, were lower than those of recent otosclerosis studies which have reported alternative WBA variables. For example, Keefe et al. (2017) showed AUROC values of 0.80 and 0.87 for ambient group delay and the difference of absorbance obtained at TPP and positive tail pressure, respectively. These values were considerably higher than the maximum AUROC (0.708) obtained in the present investigation for ambient WBA obtained at 1600 Hz with difference norms. Moreover, Keefe et al. (2017) found that the alternative variables, group delay and absorbance difference between TPP and tail pressure measures, consistently provided higher AUROC values than WBA at ambient and TPP in their study. Investigations into normative values for alternative WBA variables, such as those described by Keefe et al. (2017), may then be important for future research.  Multivariate vs. Univariate Tests for ME Pathology   The present study evaluated the test performance of WBA using univariate predictors of otosclerosis. The decision to do so was based on the recommendation of previous WBA researchers, who identified a need for quantitative methods of distinguishing normally-hearing from pathological middle ears at specific frequency points (Keefe & Simmons, 2003; Sanford & Brocket, 2014). However, given that the clinical value of WAI techniques is closely related to 89  their ability to show distinct frequency-specific patterns for normally-functioning and abnormal populations (Nakajima et al., 2013), this seems like an inefficient approach. Indeed, recent studies appear to have stepped away from the use of univariate predictors, favoring a multivariate approach.   Keefe et al. successfully demonstrated a multivariate WBA approach for the detection of otosclerosis in their 2017 paper. By combining univariate WBA variables with the highest AUROCs in a ROC-supervised principal component analysis and a subsequent logistic regression model, a higher AUROC value was obtained than for any one variable alone. Furthermore, the AUROCs increased substantially with the inclusion of each additional univariate predictor; AUROCs increased from 0.87 to 0.95 when the number of tympanometric WBA variables included in the test increased from one to three. Researchers suggested that the increase was likely related to the relatively wider frequency range observed with this method, as well as the combination of different WBA variables which may reveal more about the disease-related changes to the mechano-acoustic properties of the middle ear.   Comparison of our ROC analysis results to other research showed that there is little agreement upon which frequencies WBA variables best distinguish normally hearing and otosclerotic middle ears. This may be attributed to the highly variable nature of WBA obtained from otosclerotic ears between samples, and between otosclerotic ears in general. As such, the ROC supervised PCA approach to creating a multivariate test for otosclerosis reported by Keefe et al. (2017) may not generalize to other samples of otosclerotic ears. For this reason, an alternative means of analysis, not dependent on ROC/AUROC values, may be beneficial.  Considering this, future studies investigating the test performance of WBA could aim to evaluate multivariate predictors of the disease studied without regard to specific ‘candidate 90  variables’. Such studies may consider other means of analysis, which would more readily represent the complex interaction of factors (e.g. ethnicity, gender, age) that lead to a classification of ‘normal’ or ‘abnormal’ ME status. Discriminant analyses, which consider the between and within-subject factors which contribute to a binary designation of normal or abnormal at a given time point, could be a viable method of achieving this objective. For example, mixed model generalized estimating equations with a logit link are a consideration. Analogous research designs have been used to evaluate the significant predictors of at haul-back mortality of sharks with promising results, and in studies with an epidemiological focus (Coelho, Infante & Santos, 2012). Investigations into how these models may be integrated with pattern recognition features in immittance software to support automation of test interpretation may also be of value.   91  Conclusions  Wideband acoustic absorbance testing provides useful information regarding the status of the middle ear. This information can be used to generate a suspicion index for various middle ear pathologies, as each pathology has been shown to be associated with a unique, frequency-specific WBA profile (Nakajima et al., 2013). The present study took a novel approach to normative data for WBA, by reporting values that represent the expected between-ear difference of WBA for normally-hearing adults. Inter-aural WBA difference norms show fewer sources of variability than individual ear norms, and may be used reliably across gender and ethnicity related subpopulations, as well as pressurization methods. The ability of WBA to classify results as belonging to normally-hearing or otosclerosis groups was compared using conventional individual ear norms, and norms based on the inter-aural difference of WBA. WBA tests functioned equally with the application of either type of norm.   92  Bibliography Abbott, L. (2018). Wideband acoustic immittance measures as part of a newborn hearing screening program in Canadian first nations and metis, Caucasian and other ethnicity neonates. (Unpublished Master’s Thesis). University of British Columbia, Vancouver. Retrieved Dec 12, 2018, from https://open.library.ubc.ca/collections/ubctheses/24/items/1.0366301  Aithal, S., Kei, J., & Driscoll, C. (2014). Wideband absorbance in Australian aboriginal and Caucasian neonates. Journal of the American Academy of Audiology, 25(5), 482-494. doi:10.3766/jaaa.25.5.7  Allen, J., Jeng, P., & Levitt, H. (2005). Evaluation of human middle ear function via an acoustic power assessment. Journal of Rehabilitation Research and Development, 42(4), 63-77. doi:10.1682/JRRD.2005.04.0064  American National Standards Institute, (ANSI). (1987). Specifications for instruments to measure aural acoustic impedance and admittance (aural acoustic immittance). ANSI S3.391. New York: American National Standards Institute.  American Speech-Language-Hearing Association, (1990). Guidelines for screening for hearing impairment and middle-ear disorders. Working group on acoustic immittance measurements and the committee on audiologic evaluation. ASHA Supplement, (2), 17.  93  Beers, A. N., Shahnaz, N., Westerberg, B. D., & Kozak, F. K. (2010). Wideband reflectance in normal Caucasian and Chinese school-aged children and in children with otitis media with effusion. Ear and Hearing, 31(2), 221-233. doi:10.1097/AUD.0b013e3181c00eae  Bell, A., Adair, L., & Popkin, B. (2002). Ethnic differences in the association between body mass index and hypertension. American Journal of Epidemiology, 155(4), 346-353. doi:10.1093/aje/155.4.346  Bender, R., & Lange, S. (2001). Adjusting for multiple testing—when and how? Journal of Clinical Epidemiology, 54(4), 343-349. doi:10.1016/S0895-4356(00)00314-0  Burdiek, L. M., & Sun, X. (2014). Effects of consecutive wideband tympanometry trials on energy absorbance measures of the middle ear. Journal of Speech, Language, and Hearing Research, 57(5), 1997-2004. doi:10.1044/2014_JSLHR-H-13-0344  Coelho, R., Infante, P., & Santos, M. N. (2013). Application of generalized linear models and generalized estimation equations to model at-haulback mortality of blue sharks captured in a pelagic longline fishery in the Atlantic Ocean. Fisheries Research, 145, 66-75. doi:10.1016/j.fishres.2013.02.010  Colletti, V. (1974). Indagine timpanometrica a piu frequenze: Nota preliminare. Valsalva, 50, 152-173.  Danesh, A. A., Shahnaz, N., & Hall, J. W. (2018). The audiology of otosclerosis. In A. A. Eshraghi, & F. F. Telischi (Eds.), Otosclerosis and stapes surgery (Vol 51, Number 2, pp. 94  327-342). Philadelphia, Pennsylvania: Elsevier. doi:https://doi-org.ezproxy.library.ubc.ca/10.1016/j.otc.2017.11.007 de Jonge, R. R., & Valente, M. (1979). Interpreting ear differences in static compliance measurements. The Journal of Speech and Hearing Disorders, 44(2), 209.  DeLong, E. R., DeLong, D. M., & Clarke-Pearson, D. L. (1988). Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics, 44(3), 837-845. doi:10.2307/2531595  Ellison, J. C., Gorga, M., Cohn, E., Fitzpatrick, D., Sanford, C. A., & Keefe, D. H. (2012). Wideband acoustic transfer functions predict middle-ear effusion. The Laryngoscope, 122(4), 887-894. doi:10.1002/lary.23182  Emanuel, D., Henson, O., & Knapp, R. (2012). Survey of audiological immittance practices. American Journal of Audiology, 21(1), 60-75. doi:10.1044/1059-0889(2012/11-0037)  Feeney, M. P., Hunter, L. L., Kei, J., Lilly, D. J., Margolis, R. H., Nakajima, H. H., Neely, S. T., Prieve, B. A., Rosowski, J. J., Sanford, C. A., Schairer, K. S., Shahnaz, N., Stenfelt, S., Voss, S. E., Linkopings Universitet, Institutionen for Klinisk och Experimentell Medicin, Halsouniversitetet, Teknisk audiologi. (2013). Consensus statement: Eriksholm workshop on wideband absorbance measures of the middle ear. Ear and Hearing, 34(7 Suppl 1), 78s-79s. doi:10.1097/AUD.0b013e31829c726b  Feeney, M. P., Keefe, D. H., Hunter, L. L., Fitzpatrick, D. F., Garinis, A. C., Putterman, D. B., & McMillan, G. P. (2016). Normative wideband reflectance, equivalent admittance at the 95  tympanic membrane, and acoustic stapedius reflex threshold in adults. Ear and Hearing, 38(3), 142-160. doi:10.1097/AUD.0000000000000399  Feeney, M. P., Stover, B., Keefe, D. H., Garinis, A. C., Day, J. E., & Seixas, N. (2014). Sources of variability in wideband energy reflectance measurements in adults. Journal of the American Academy of Audiology, 25(5), 449. doi:10.3766/jaaa.25.5.4 Feeney, M. P., Grant, I. L., & Marryott, L. P. (2003). Wideband energy reflectance measurements in adults with middle-ear disorders. Journal of Speech, Language, and Hearing Research, 46(4), 901-911. doi:10.1044/1092-4388(2003/070)  Feeney, M. P., & Keefe, D. H. (1999). Acoustic reflex detection using wide-band acoustic reflectance, admittance, and power measurements. Journal of Speech, Language, and Hearing Research, 42(5), 1029-1041. doi:10.1044/jslhr.4205.1029  Feeney, M., & Sanford, C. (2004). Age effects in the human middle ear: Wideband acoustical measures. Journal of the Acoustical Society of America, 116(6), 3546-3558. doi:10.1121/1.1808221  Feldman, A. S. (1974). Eardrum abnormality and the measurement of middle ear function. Archives of Otolaryngology, 99(3), 211-217. doi:10.1001/archotol.1974.00780030219013  Foster, M. F., & Backous, D. D. (2018). Clinical evaluation of the patient with otosclerosis. In A. A. Eshraghi, & F. F. Telischi (Eds.), Otosclerosis and stapes surgery (Vol 51, Number 2, pp. 319-326). Philedelphia, Pennsylvania: Elsevier. doi:https://doi-org.ezproxy.library.ubc.ca/10.1016/j.otc.2017.11.004 96  Hunter, L. L., Feeney, M. P., Lapsley Miller, J. A., Jeng, P. S., & Bohning, S. (2010). Wideband reflectance in newborns: Normative regions and relationship to hearing-screening results. Ear and Hearing, 31(5), 599-610. doi:10.1097/AUD.0b013e3181e40ca7  Hunter, L. L., Keefe, D. H., Feeney, M. P., Fitzpatrick, D. F., & Lin, L. (2016; 2015). Longitudinal development of wideband reflectance tympanometry in normal and at-risk infants. Hearing Research, 340, 3-14. doi:10.1016/j.heares.2015.12.014  Hunter, L. L., & Shanaz, N. (2014). Acoustic immittance measures: Basic and advanced practice. San Diego, CA: Plural Publishing, Inc.  Interacoustics. (2017). Instructions for use - US titan. Retrieved Dec 12, 2018, from http://www.interacoustics.com/impedance/clinical-diagnostic/titan  Jaffer, S. (2016). Wideband acoustic immittance: Instrument, ethnicity, and gender specific normative data. (Unpublished Master’s Thesis). University of British Columbia, Vancouver. Retrieved Dec 12, 2018, from https://open.library.ubc.ca/collections/ubctheses/24/items/1.0300134  Keefe, D. H., Bulen, J. C., Arehart, K. H., & Burns, E. M. (1993). Ear-canal impedance and reflection coefficient in human infants and adults. The Journal of the Acoustical Society of America, 94(5), 2617-2638. doi:10.1121/1.407347 Keefe, D. H., Ling, R., & Bulen, J. C. (1992). Method to measure acoustic impedance and reflection coefficient. Journal of the Acoustical Society of America, 91(1), 470-485. doi:10.1121/1.402733 97  Keefe, D., Archer, K., Schmid, K., Fitzpatrick, D., Feeney, M., & Hunter, L. (2017). Identifying otosclerosis with aural acoustical tests of absorbance, group delay, acoustic reflex threshold, and otoacoustic emissions. Journal of the American Academy of Audiology, 28(9), 838-860. doi:10.3766/jaaa.16172  Keefe, D., & Simmons, J. (2003). Energy transmittance predicts conductive hearing loss in older children and adults. Journal of the Acoustical Society of America, 114(6), 3217-3238. doi:10.1121/1.1625931  Kenny, S. P. V. N. (2011). Clinical application of the Interacoustics REFLWIN system wideband reflectance machine in the assessment of the Eustachian tube. (Unpublished Master’s Thesis). University of British Columbia, Vancouver. Retrieved Dec 12, 2018, from https://open.library.ubc.ca/cIRcle/collections/ubctheses/24/items/1.0072303   Lidén, G., Peterson, J. L., & Björkman, G. (1970). Tympanometry. Archives of Otolaryngology, 92(3), 248-257. doi:10.1001/archotol.1970.04310030038009  Lildholdt, T. (1980). Negative middle ear pressure: Variations by season and sex. Ann Otol Rhinol Laryngol, 89(3), 67-70. doi:10.1177/00034894800890S319  Liu, Y., Sanford, C. A., Ellison, J. C., Fitzpatrick, D. F., Gorga, M. P., & Keefe, D. H. (2008). Wideband absorbance tympanometry using pressure sweeps: System development and results on adults with normal hearing. The Journal of the Acoustical Society of America, 124(6), 3708. doi:10.1121/1.3001712  98  Manly, J. J. (2005). Advantages and disadvantages of separate norms for African Americans. The Clinical Neuropsychologist, 19(2), 270-275. doi:10.1080/13854040590945346  Margolis, R. H., Saly, G. L., & Keefe, D. H. (1999). Wideband reflectance tympanometry in normal adults. The Journal of the Acoustical Society of America, 106(1), 265-280. doi:10.1121/1.427055 Martin, F. N., & Clark, J. G. (2019). Introduction to audiology (Thirteenth ed.). NY: Pearson.  Mazlan, R., Kei, J., Ya, C. L., Yusof, W. N. H. M., Saim, L., & Zhao, F. (2015). Age and gender effects on wideband absorbance in adults with normal outer and middle ear function. Journal of Speech, Language, and Hearing Research: JSLHR, 58(4), 1377-1386. doi:10.1044/2015_JSLHR-H-14-0199  Merchant, S. N., Merchant, G. R., Rosowski, J. J., & Nakajima, H. H. (2016). Controlled exploration of the effects of conductive hearing loss on wideband acoustic immittance in human cadaveric preparations. Hearing Research, 341, 19-30. doi:10.1016/j.heares.2016.07.018  Minor, L. B. (2005). Clinical manifestations of superior semicircular canal dehiscence. The Laryngoscope, 115(10), 1717-1727. doi:10.1097/01.mlg.0000178324.55729.b7 Mukerji, S., Windsor, A. M., & Lee, D. J. (2010). Auditory brainstem circuits that mediate the middle ear muscle reflex. Trends in Amplification, 14(3), 170-191. doi:10.1177/1084713810381771 99  Nakajima, H. H., Pisano, D. V., Roosli, C., Hamade, M. A., Merchant, G. R., Mahfoud, L., Halpin, C.F., Rosowski, J. J., Merchant, S. N. (2012). Comparison of ear-canal reflectance and umbo velocity in patients with conductive hearing loss: A preliminary study. Ear and Hearing, 33(1), 35-43. doi:10.1097/AUD.0b013e31822ccba0  Nakajima, H. H., Rosowski, J. J., Shahnaz, N., & Voss, S. E. (2013). Assessment of ear disorders using power reflectance. Ear and Hearing, 34(7 Suppl 1), 48s-53s. doi:10.1097/AUD.0b013e31829c964d  Naumann, I. C., Porcellini, B., & Fisch, U. (2005). Otosclerosis: Incidence of positive findings on high-resolution computed tomography and their correlation to audiological test data. Annals of Otology, Rhinology & Laryngology, 114(9), 709-716. doi:10.1177/000348940511400910  Nørgaard, K. R., Fernandez-Grande, E., & Laugesen, S. (2017). Incorporating evanescent modes and flow losses into reference impedances in acoustic thévenin calibration. The Journal of the Acoustical Society of America, 142(5), 3013-3024. doi:10.1121/1.5010891  Polat, Z., Baş, B., Hayır, D., Bulut, E., & Ataş, A. (2015). Wideband tympanometry normative data for Turkish young adult population. The Journal of International Advanced Otology, 11(2), 157. doi:10.5152/iao.2015.809 Probst, R. (2007). Audiological evaluation of patients with otosclerosis. Advances in Oto-Rhino-Laryngology, 65, 119.  100  Rappaport, J. M. & Provencal, C. (2002). Neurotology for audiologists. In J. Katz, R. F. Burkard, & L. Medwetsky (Eds). Handbook of Clinical Audiology (5th ed., pp. 9–32). Philadelphia, PA: Lippincott, Williams and Wilkins. Riddler, R. J. (2017). Effects of middle ear pressure compensation on evoked otoacoustic emissions and power absorbance in adults. (Unpublished Master’s Thesis). University of British Columbia, Vancouver. Retrieved Dec 12, 2018, from https://open.library.ubc.ca/collections/ubctheses/24/items/1.0343220  Robinson, S. R., Thompson, S., & Allen, J. B. (2016). Effects of negative middle ear pressure on wideband acoustic immittance in normal-hearing adults. Ear and Hearing, 37(4), 452-464. doi:10.1097/AUD.0000000000000280  Rosowski, J. J., Nakajima, H. H., Hamade, M. A., Mahfoud, L., Merchant, G. R., Halpin, C. F., & Merchant, S. N. (2012). Ear-canal reflectance, umbo velocity, and tympanometry in normal-hearing adults. Ear and Hearing, 33(1), 19-34. doi:10.1097/AUD.0b013e31822ccb76  Rosowski, J. J., Stenfelt, S., Lilly, D., Linköpings universitet, Institutionen för klinisk och experimentell medicin, Hälsouniversitetet, & Teknisk audiologi. (2013). An overview of wideband immittance measurements techniques and terminology: You say absorbance, I say reflectance. Ear and Hearing, 34(7 Suppl 1), 9s-16s. doi:10.1097/AUD.0b013e31829d5a14  Ruah, C. B., Schachern, P. A., Zelterman, D., Paparella, M. M., & Yoon, T. H. (1991). Age-related morphologic changes in the human tympanic membrane: A light and electron 101  microscopic study. Archives of Otolaryngology–Head & Neck Surgery, 117(6), 627-634. doi:10.1001/archotol.1991.01870180063013  Rudic, M., Keogh, I., Wagner, R., Wilkinson, E., Kiros, N., Ferrary, E., Sterkers, O., Bozorg Grayeli, A., Zarkovic, K., Zarkovic, N. (2015). The pathophysiology of otosclerosis: Review of current research. Hearing Research, 330, 51-56. doi:https://doi-org.ezproxy.library.ubc.ca/10.1016/j.heares.2015.07.014 Sanford, C. A., & Brockett, J. E. (2014). Characteristics of wideband acoustic immittance in patients with middle-ear dysfunction. Journal of the American Academy of Audiology, 25(5), 425. doi:10.3766/jaaa.25.5.2 Sanford, C. A., Hunter, L. L., Feeney, M. P., & Nakajima, H. H. (2013). Wideband acoustic immittance: Tympanometric measures. Ear and Hearing, 34(7 Suppl 1), 65s-71s. doi:10.1097/AUD.0b013e31829c7250  Shahnaz, N. (2010). Clinical application of wideband reflectance (WBR) in infants, children and adults. Canadian Hearing Report, 5, 23-29.  Shahnaz, N., & Bork, K. (2006). Wideband reflectance norms for Caucasian and Chinese young adults. Ear and Hearing, 27(6), 774-788. doi:10.1097/01.aud.0000240568.00816.4a  Shahnaz, N., & Bork, K. (2008). Comparison of standard and multi-frequency tympanometric measures obtained with the virtual 310 system and the grason-stadler tympstar. Canadian Journal of Speech-Language Pathology and Audiology/Revue Canadienne d 'Orthophonie Et d'Audiologie, 32(4), 146-157.  102  Shahnaz, N., Bork, K., Polka, L., Longridge, N., Bell, D., & Westerberg, B. D. (2009). Energy reflectance and tympanometry in normal and otosclerotic ears. Ear and Hearing, 30(2), 219-233. doi:10.1097/AUD.0b013e3181976a14  Shahnaz, N., & Davies, D. (2006). Standard and multifrequency tympanometric norms for Caucasian and Chinese young adults. Ear and Hearing, 27(1), 75-90. doi:10.1097/01.aud.0000194516.18632.d2  Shahnaz, N., Feeney, M. P., & Schairer, K. S. (2013). Wideband acoustic immittance normative data: Ethnicity, gender, aging, and instrumentation. Ear and Hearing, 34(7 Suppl 1), 27s-35s. doi:10.1097/AUD.0b013e31829d5328  Shahnaz, N., & Polka, L. (1997). Standard and multifrequency tympanometry in normal and otosclerotic ears. Ear & Hearing, 18(4), 326-341. doi:10.1097/00003446-199708000-00007  Shaw, J. (2009). Comparison of wideband energy reflectance and tympanometric measures obtained with reflwin Interacoustics, mimosa acoustics and GSI tympstar systems. (Unpublished Master's Thesis). University of British Columbia, Vancouver. doi:10.14288/1.0070857 Sun, X. (2012). Ear canal pressure variations versus negative middle ear pressure: Comparison using distortion product otoacoustic emission measurement in humans. Ear and Hearing, 33(1), 69-78. doi:10.1097/AUD.0b013e3182280326  103  Sun, X. (2016). Wideband acoustic immittance: Normative study and test-retest reliability of tympanometric measurements in adults. Journal of Speech, Language, and Hearing Research, 59(4), 819. doi:10.1044/2016_JSLHR-H-14-0322  Sziklai, I., SpringerLINK ebooks - Medicine, & SpringerLink (Online service). (2016). Surgery of stapes fixations (1st 2016.; 1st 2016 ed.). Cham: Springer International Publishing. doi:10.1007/978-3-319-28576-4  TIBCO Software Inc. (2017). Statistica [data analysis software system], version 13. Retrieved May 1, 2018, from http://statistica.io.  Turner, R., Robinette, M., Bauch, C., Musiek, F., & Rintelmann, W. (1999). Clinical decisions. In F. Musiek & W. Rintelmann (Eds.), Contemporary perspectives in hearing assessment. (pp. 437-463). Boston, MA: Allyn and Bacon.  Vander Werff, K. R., Prieve, B. A., & Georgantas, L. M. (2007). Test-retest reliability of wideband reflectance measures in infants under screening and diagnostic test conditions. Ear and Hearing, 28(5), 669-681. doi:10.1097/AUD.0b013e31812f71b1  Voss, S. E., & Allen, J. B. (1994). Measurement of acoustic-impedance and reflectance in the human ear canal. Journal of the Acoustical Society of America, 95(1), 372-384. doi:10.1121/1.408329 Voss, S. E. (2014, June 5). Development of a normative database for wideband acoustic immittance measures. Retrieved Dec 12, 2018, from http://grantome.com/grant/NIH/R15-DC014129-01  104  Voss, S. E., Horton, N. J., Woodbury, R. R., & Sheffield, K. N. (2008). Sources of variability in reflectance measurements on normal cadaver ears. Ear and Hearing, 29(4), 651-665. doi:10.1097/AUD.0b013e318174f07c  Wada, H., Koike, T., & Kobayashi, T. (1998). Clinical applicability of the sweep frequency measuring apparatus for diagnosis of middle ear diseases. Ear & Hearing, 19(3), 240-249. doi:10.1097/00003446-199806000-00007  Wan, I. K. K., & Wong, L. L. N. (2002). Tympanometric norms for Chinese young adults. Ear and Hearing, 23(5), 416-421. doi:10.1097/00003446-200210000-00003  Werner, L. A., Levi, E. C., & Keefe, D. H. (2010). Ear-canal wideband acoustic transfer functions of adults and two- to nine-month-old infants. Ear and Hearing, 31(5), 587-598. doi:10.1097/AUD.0b013e3181e0381d  Youden, W. J. (1950). Index for rating diagnostic tests. Cancer, 3(1), 32-35. doi:10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3  Zhao, F., Wada, H., Koike, T., Ohyama, K., Kawase, T., & Stephens, D. (2002). Middle ear dynamic characteristics in patients with otosclerosis. Ear and Hearing, 23(2), 150-158. doi:10.1097/00003446-200204000-00007   105  Appendices Appendix A  Supporting Documents for Methods  Table A.1. Bandwidth of each 1/3 octave band, organized per center frequency. Center Frequency (Hz) Bandwidth (Hz) 250 224-282 315 282-355 400 355-447 500 447-562 630 562-708 800 708-891 1000 891-1122 1250 1122-1413 1600 1413-1778 2000 1778-2239 2500 2239-2818 3150 2818-3548 4000 3548-4467 5000 4467-5623 6300 5623-7079 8000 7079-8913  Table A.2. Records for which Mean Substitution was used to Accommodate Missing Data. Study Group Subjects with Mean Substitution (n=) Effects of race, Caucasian, East Asian, on middle ear function and hearing sensitivity norms  Normal 0 Effects of Middle Ear Pressure Compensation on Evoked Otoacoustic Emissions and Power Absorbance in Adults   Normal 0 Wideband Energy Reflectance in Surgically Confirmed Middle Ear Ossicular Chain Abnormalities  Normal 0 Otosclerosis 4 Notes: Mean substitution was used to accommodate missing data at several frequency points for 4 subjects. As stated in the Methods section, missing data was permitted, and mean substitution was used to fill these data-points, only if missing values occurred sporadically within the dataset for the subject, and did not preclude inclusion in an entire pressure condition.    106  Appendix B  Supporting Documents for Results  This section provides supporting documents for the results section, including detailed outputs of statistical analyses.  Table A.3. Results: Mixed model ANOVA for individual ear WBA data prior to GG correction. Effect Repeated Measures Analysis of Variance (Refined Database ready for analysis_Aug 29). Sigma-Restricted Parameterization. Effective hypothesis decomposition; Std. Error of Estimate: 0.4362 SS Deg. Of F MS F p Intercept  1094.720 1 1094.720 5753.593 0.000000 Ethnicity 0.221 2 0.111 0.582 0.559713 Gender 0.683 1 0.683 3.589 0.059364 Ear  0.020 1 0.020 0.107 0.743974 Error  45.474 239 0.190     Pressure 0.822 2 0.411 58.997 0.000000 Pressure*Ethnicity 0.014 4 0.004 0.515 0.724470 Pressure*Gender 0.010 2 0.005 0.692 0.501024 Pressure*Ear 0.002 2 0.001 0.151 0.860219 Error  3.330 478 0.007     Frequency 256.578 15 17.105 482.993 0.000000 Frequency*Ethnicity 9.279 30 0.309 8.734 0.000000 Frequency*Gender 5.755 15 0.384 10.833 0.000000 Frequency*Ear 0.329 15 0.022 0.619 0.862019 Error  126.963 3585 0.035     Pressure*Frequency 1.342 30 0.045 28.291 0.000000 Pressure*Frequency*Ethnicity 0.237 60 0.004 2.501 0.000000 Pressure*Frequency*Gender 0.038 30 0.001 0.796 0.776449 Pressure*Frequency*Ear 0.012 30 0.000 0.257 0.999986 Error  11.338 7170 0.002     Notes: Deg. of F = Degrees of Freedom  107  Table A.4. Results: Mixed Model ANOVA for individual ear data adjusted with GG-Correction. Effect Adjusted Univariate Tests for Repeated Measure: DV_1(Refined database ready for analysis Aug 29); Sigma-restricted parameterization; Effective hypothesis decomposition.  Deg. Of F F p G-G Epsilon G-G Adj. df 1 G-G Adj. df2 G-G Adj. p Pressure 2 58.9974 0.000000 0.536876 1.07375 256.627 0.000000 Pressure*Ethnicity  4 0.5154 0.724470 0.536876 2.14750 256.627 0.610870 Pressure*Gender 2 0.6921 0.501024 0.536876 1.07375 256.627 0.415980 Pressure*Ear 2 0.1506 0.860219 0.536876 1.07375 256.627 0.716481 Error 478             Frequency 15 482.9931 0.000000 0.275751 4.13626 988.566 0.000000 Frequency*Ethnicity 30 8.7336 0.000000 0.275751 8.27252 988.566 0.000000 Frequency*Gender 15 10.8331 0.000000 0.275751 4.13626 988.566 0.000000 Frequency*Ear 15 0.6188 0.862019 0.275751 4.13626 988.566 0.654593 Error 3585             Pressure*Frequency 30 28.2910 0.000000 0.179334 5.38001 1285.822 0.000000 Pressure*Frequency*Ethnicity 60 2.5014 0.000000 0.179334 10.76001 1285.822 0.004432 Pressure*Frequency*Gender 30 0.7965 0.776449 0.179334 5.38001 1285.822 0.560351 Pressure*Frequency*Ear 30 0.2573 0.999986 0.179334 5.38001 1285.822 0.944880 Error 7170             Notes:  Deg. of F = Degrees of Freedom; GG Adj. df = Greenhouse-Geisser adjusted degrees of freedom  108  Table A.5. Results: Mixed model ANOVA for inter-aural difference WBA data prior to GG correction. Effect Repeated Measures Analysis of Variance (Refined Database ready for analysis_Aug 29). Sigma-Restricted Parameterization. Effective hypothesis decomposition; Std. Error of Estimate: 0.4362 SS Deg. Of Freedom MS F P Inrcept  18.438 1 18.438 328.139 0.000000 Ethnicity  0.165 2 0.082 1.465 0.235332 Gender  0.004 1 0.004 0.064 0.800329 Error 6.630 118 0.056   Pressure 0.013 2 0.006 1.250 0.288348 Pressure*Ethnicity  0.003 4 0.001 0.165 0.956025 Pressure*Gender 0.007 2 0.003 0.656 0.519718 Error 1.195 236 0.005   Frequency 1.641 15 0.109 10.266 0.000000 Frequency*Ethnicity  0.314 30 0.010 0.983 0.492061 Frequency*Gender 0.162 15 0.011 1.013 0.438453 Error 18.864 1770 0.011   Pressure*Frequency 0.084 30 0.003 2.232 0.000137 Pressure*Frequency*Ethnicity 0.098 60 0.002 1.297 0.062977 Pressure*Frequency*Gender 0.010 30 0.000 0.259 0.999985 Error 4.454 3540 0.001     Notes:  Deg. of F = Degrees of Freedom; GG Adj. df = Greenhouse-Geisser adjusted degrees of freedom 109  Table A.6 Results: Mixed model ANOVA for inter-aural difference WBA data after GG correction. Effect Adjusted Univariate Tests for Repeated Measure: DV_1(Refined database ready for analysis Aug 29); Sigma-restricted parameterization; Effective hypothesis decomposition.  Deg. Of Freedom F p G-G Epsilon G-G Adj. df 1 G-G Adj. df2 G-G Adj. p Pressure  2 1.2502 0.288348 0.627197 1.25439 148.019 0.275441 Pressure*Ethnicity 4 0.1648 0.956025 0.627197 2.50879 148.019 0.891127 Pressure*Gender 2 0.6563 0.519718 0.627197 1.25439 148.019 0.452068 Error 236             Frequency 15 10.2655 0.00000 0.396509 5.94764 701.822 0.00000 Frequency*Ethnicity 30 0.9834 0.492061 0.396509 11.89528 701.822 0.462754 Frequency*Gender 15 1.0127 0.438453 0.396509 5.94764 701.822 0.415398 Error 1770             Pressure*Frequency 30 2.2316 0.000137 0.283394 8.50183 1003.215 0.020525 Pressure*Frequency*Ethnicity 60 1.2966 0.062977 0.283394 17.00365 1003.215 0.18599 Pressure*Frequency*Gender 30 0.2588 0.999985 0.283394 8.50183 1003.215 0.982139 Error 3540             Notes:  Deg. of F = Degrees of Freedom; GG Adj. df = Greenhouse-Geisser adjusted degrees of freedom  110  Table A.7. AUROC and p-values for each univariate predictor of otosclerosis, organized by pressure condition and frequency. Highlighted cells indicate significance (p<.05). Individual Ear WBA Data (Group Norms) Inter-Aural WBA Difference Data (Difference Norms) Variable AUROC p-value Variable AUROC p-value Indiv.Ear_Peak_3DT_250 0.561 0.3584 Difference_Peak_3DT_250 0.642 0.0285 Indiv.Ear _Peak__3DT_315 0.586 0.2168 Difference_Peak_3DT_315 0.642 0.026 Indiv.Ear_Peak_3DT_400 0.560 0.3932 Difference_Peak_3DT_400 0.627 0.0558 Indiv.Ear_Peak_3DT_500 0.538 0.594 Difference_Peak_3DT_500 0.646 0.0185 Indiv.Ear_Peak_3DT_630 0.544 0.5263 Difference_Peak_3DT_630 0.613 0.0659 Indiv.Ear_Peak_3DT_800 0.515 0.8333 Difference_Peak_3DT_800 0.579 0.2153 Indiv.Ear_Peak_3DT_1000 0.535 0.6117 Difference_Peak_3DT_1000 0.630 0.0317 Indiv.Ear_Peak_3DT_1250 0.506 0.9362 Difference_Peak_3DT_1250 0.613 0.0586 Indiv.Ear_Peak_3DT_1600 0.563 0.3756 Difference_Peak_3DT_1600 0.623 0.0317 Indiv.Ear_Peak_3DT_2000 0.576 0.2512 Difference_Peak_3DT_2000 0.527 0.6874 Indiv.Ear_Peak_3DT_2500 0.622 0.0486 Difference_Peak_3DT_2500 0.570 0.2573 Indiv.Ear__Peak_3DT_3150 0.701 0.0006 Difference_Peak_3DT_3150 0.518 0.7787 Indiv.Ear_Peak_3DT_4000 0.654 0.0087 Difference_Peak_3DT_4000 0.587 0.1408 Indiv.Ear_Peak_3DT_5000 0.623 0.031 Difference_Peak_3DT_5000 0.541 0.5087 Indiv.Ear__Peak_3DT_6300 0.524 0.7108 Difference_Peak_3DT_6300 0.522 0.7216 Indiv.Ear__Peak_3DT_8000 0.503 0.9661 Difference_Peak_3DT_8000 0.535 0.5682 Indiv.Ear__Ambient_3DT_250 0.561 0.3584 Difference_Ambient_3DT_250 0.642 0.0285 Indiv.Ear__Ambient_3DT_315 0.545 0.4981 Difference_Ambient_3DT_315 0.640 0.03 Indiv.Ear__Ambient_3DT_400 0.515 0.8292 Difference_Ambient_3DT_400 0.619 0.0714 Indiv.Ear_Ambient_3DT_500 0.512 0.8604 Difference_Ambient_3DT_500 0.604 0.1343 Indiv.Ear_Ambient_3DT_630 0.520 0.7705 Difference_Ambient_3DT_630 0.598 0.137 Indiv.Ear_Ambient_3DT_800 0.509 0.8981 Difference_Ambient_3DT_800 0.551 0.4382 Indiv.Ear_Ambient_3DT_1000 0.566 0.3354 Difference_Ambient_3DT_1000 0.581 0.2336 Indiv.Ear__Ambient_3DT_1250 0.534 0.6264 Difference_Ambient_3DT_1250 0.592 0.1592 Indiv.Ear__Ambient_3DT_1600 0.594 0.1806 Difference_Ambient_3DT_1600 0.605 0.1194 Indiv.Ear__Ambient_3DT_2000 0.609 0.0994 Difference_Ambient_3DT_2000 0.504 0.9558 111  Individual Ear WBA Data (Group Norms) Inter-Aural WBA Difference Data (Difference Norms) Variable AUROC p-value Variable AUROC p-value Indiv.Ear__Ambient_3DT_2500 0.615 0.0609 Difference_Ambient_3DT_2500 0.613 0.0667 Indiv.Ear__Ambient_3DT_3150 0.688 0.0014 Difference_Ambient_3DT_3150 0.540 0.5735 Indiv.Ear__Ambient_3DT_4000 0.646 0.0125 Difference_Ambient_3DT_4000 0.617 0.0404 Indiv.Ear__Ambient_3DT_5000 0.629 0.0245 Difference_Ambient_3DT_5000 0.563 0.3098 Indiv.Ear__Ambient_3DT_6300 0.526 0.6889 Difference_Ambient_3DT_6300 0.510 0.875 Indiv.Ear__Ambient_3DT_8000 0.504 0.955 Difference_Ambient_3DT_8000 0.523 0.7209 Indiv.Ear__Ambient_WBA_250 0.58 0.2094 Difference_Ambient_WBA_250 0.613 0.0681 Indiv.Ear__Ambient_WBA_315 0.551 0.4301 Difference_Ambient_WBA_315 0.590 0.1606 Indiv.Ear__Ambient_WBA_400 0.573 0.2681 Difference_Ambient_WBA_400 0.614 0.0702 Indiv.Ear__Ambient_WBA_500 0.557 0.4053 Difference_Ambient_WBA_500 0.629 0.0445 Indiv.Ear__Ambient_WBA_630 0.555 0.4476 Difference_Ambient_WBA_630 0.647 0.0145 Indiv.Ear__Ambient_WBA_800 0.606 0.1432 Difference_Ambient_WBA_800 0.521 0.7516 Indiv.Ear__Ambient_WBA_1000 0.685 0.0075 Difference_Ambient_WBA_1000 0.537 0.5641 Indiv.Ear__Ambient_WBA_1250 0.612 0.1216 Difference_Ambient_WBA_1250 0.572 0.2443 Indiv.Ear__Ambient_WBA_1600 0.599 0.1412 Difference_Ambient_WBA_1600 0.708 <0.0001 Indiv.Ear__Ambient_WBA_2000 0.535 0.5724 Difference_Ambient_WBA_2000 0.510 0.8663 Indiv.Ear__Ambientt_WBA_2500 0.501 0.9915 Difference_Ambient_WBA_2500 0.501 0.9797 Indiv.Ear__Ambient_WBA_3150 0.590 0.1534 Difference_Ambient_WBA_3150 0.527 0.677 Indiv.Ear__Ambient_WBA_4000 0.600 0.1045 Difference_Ambient_WBA_4000 0.508 0.9059 Indiv.Ear__Ambient_WBA_5000 0.657 0.0055 Difference_Ambient_WBA_5000 0.539 0.5004 Indiv.Ear__Ambient_WBA_6300 0.541 0.5193 Difference_Ambient_WBA_6300 0.547 0.4072 Indiv.Ear__Ambient_WBA_8000 0.539 0.531 Difference_Ambient_WBA_8000 0.523 0.7347 Notes:  Format of variables: Norm Type_Pressurization Method_Test Frequency (Hz)  Highlighted rows indicate that the univariate predictor performed significantly above chance levels (i.e. p<.05).   112  Table A.8. Pair-wise comparisons of univariate predictors with three highest AUROC values from individual ear and inter-aural difference norms. AUROC plot AUROC SE 95% CI   Difference_Ambient_WBA_1600 0.708 0.0464 0.629 to 0.780   Difference_Ambient_WBA_630 0.647 0.0601 0.565 to 0.723   Difference_Peak_3DT_500 0.646 0.0621 0.564 to 0.722   Group_Peak_3DT_3150 0.694 0.0611 0.614 to 0.767   Group_Ambient_3DT_3150 0.679 0.0614 0.598 to 0.753   Group_Ambient_WBA_1000 0.705 0.0692 0.625 to 0.777   Pairwise comparison of AUROC AUROC Difference SE 95% CI Z-statistic p Difference_Ambient_WBA_1600 ~ Difference_Ambient_WBA_630 0.0613 0.0674 -0.0709 to 0.194 0.909 .3632 Difference_Ambient_WBA_1600 ~ Difference_Peak_3DT_500 0.0621 0.073 -0.0811 to 0.205 0.85 .3954 Difference_Ambient_WBA_1600 ~ Group_Peak_3DT_3150 0.0141 0.0871 -0.157 to 0.185 0.161 .8718 Difference_Ambient_WBA_1600 ~ Group_Ambient_3DT_3150 0.0288 0.0875 -0.143 to 0.200 0.33 .7417 Difference_Ambient_WBA_1600 ~ Group_Ambient_WBA_1000 0.00337 0.0822 -0.158 to 0.164 0.041 .9673 Difference_Ambient_WBA_630 ~ Difference_Peak_3DT_500 0.000732 0.0655 -0.128 to 0.129 0.0112 .9911 Difference_Ambient_WBA_630 ~ Group_Peak_3DT_3150 0.0473 0.0893 -0.128 to 0.222 0.53 .5963 Difference_Ambient_WBA_630 ~ Group_Ambient_3DT_3150 0.0325 0.0891 -0.142 to 0.207 0.365 .7153 Difference_Ambient_WBA_630 ~ Group_Ambient_WBA_1000 0.058 0.11 -0.157 to 0.273 0.528 .5975 Difference_Peak_3DT_500 ~ Group_Peak_3DT_3150 0.048 0.0879 -0.124 to 0.220 0.546 .5848 Difference_Peak_3DT_500 ~ Group_Ambient_3DT_3150 0.0332 0.0856 -0.135 to 0.201 0.388  .698 Difference_Peak_3DT_500 ~ Group_Ambient_WBA_1000 0.0587 0.106 -0.149 to 0.266 0.555 .579 Group_Peak_3DT_3150 ~ Group_Ambient_3DT_3150 0.0148 0.0154 -0.0155 to 0.0450 0.958 .3379 Group_Peak_3DT_3150 ~ Group_Ambient_WBA_1000 0.0107 0.0871 -0.160 to 0.181 0.123 .9023 Group_Ambient_3DT_3150 ~ Group_Ambient_WBA_1000 0.0255 0.0868 -0.145 to 0.196 0.293 .7691 Notes: 113  Table A.9. Pair-wise comparisons of univariate predictors with the six highest AUROC values across norm type. AUROC plot AUROC SE 95% CI 	 	Difference_Ambient_WBA_1600 0.708 0.0464 0.629 to 0.780   Group_Peak_3DT_3150 0.694 0.0611 0.614 to 0.767   Group_Ambient_3DT_3150 0.679 0.0614 0.598 to 0.753   Group_Ambient_WBA_1000 0.705 0.0692 0.625 to 0.777   Group_Ambient_WBA_5000 0.644 0.059 0.562 to 0.720   Group_Peak_3DT_4000 0.664 0.0614 0.582 to 0.739   Pairwise comparison of AUROC AUROC Difference SE 95% CI Z-statistic p Difference_Ambient_WBA_1600 ~ Group_Peak_3DT_3150 0.0141 0.0871 -0.157 to 0.185 0.161 .8718 Difference_Ambient_WBA_1600 ~ Group_Ambient_3DT_3150 0.0288 0.0875 -0.143 to 0.200 0.33 .7417 Difference_Ambient_WBA_1600 ~ Group_Ambient_WBA_1000 0.00337 0.0822 -0.158 to 0.164 0.041 .9673 Difference_Ambient_WBA_1600 ~ Group_Ambient_WBA_5000 0.0644 0.0764 -0.0852 to 0.214 0.844 .3989 Difference_Ambient_WBA_1600 ~ Group_Peak_3DT_4000 0.0442 0.082 -0.117 to 0.205 0.539 .5899 Group_Peak_3DT_3150 ~ Group_Ambient_3DT_3150 0.0148 0.0154 -0.0155 to 0.0450 0.958 .3379 Group_Peak_3DT_3150 ~ Group_Ambient_WBA_1000 0.0107 0.0871 -0.160 to 0.181 0.123 .9023 Group_Peak_3DT_3150 ~ Group_Ambient_WBA_5000 0.0504 0.0657 -0.0785 to 0.179 0.766 .4436 Group_Peak_3DT_3150 ~ Group_Peak_3DT_4000 0.0302 0.0339 -0.0363 to 0.0966 0.889 .3741 Group_Ambient_3DT_3150 ~ Group_Ambient_WBA_1000 0.0255 0.0868 -0.145 to 0.196 0.293 .7691 Group_Ambient_3DT_3150 ~ Group_Ambient_WBA_5000 0.0356 0.0679 -0.0976 to 0.169 0.524 .6005 Group_Ambient_3DT_3150 ~ Group_Peak_3DT_4000 0.0154 0.0379 -0.0590 to 0.0897 0.405 .6853 Group_Ambient_WBA_1000 ~ Group_Ambient_WBA_5000 0.061 0.0886 -0.113 to 0.235 0.689 .4908 Group_Ambient_WBA_1000 ~ Group_Peak_3DT_4000 0.0408 0.0868 -0.129 to 0.211 0.47 .6382 Group_Ambient_WBA_5000 ~ Group_Peak_3DT_4000 0.0202 0.0523 -0.0822 to 0.123 0.387 .6991 Notes:  114  Appendix C  Supporting data for the Discussion Section Table A.10. Comparison of normative values of the current study with other published norms.   Shahnaz & Bork, (2006) Keefe et al. (1993) Feeney et al.  (2004) Voss & Allen, (1994) Polat et al. (2015) Current Study  Frequency (Hz) Mean SD Mean SD Mean SD Mean SD Mean  SD  Mean  SD 250 0.09 0.08 0.08 0.04 0.04 0.03 0.08 0.05 0.12 0.05 0.11 0.05 315 0.12 0.08 0.08 0.04 0.05 0.04 0.1 0.05 0.15 0.07 0.11 0.07 400 0.16 0.09 0.12 0.05 0.07 0.05 0.13 0.06 0.19 0.01 0.15 0.08 500 0.24 0.12 0.17 0.07 0.12 0.07 0.18 0.07 0.28 0.11 0.24 0.11 630 0.32 0.14 0.24 0.08 0.18 0.1 0.25 0.08 0.37 0.13 0.34 0.14 800 0.42 0.16 0.33 0.11 0.24 0.13 0.36 0.11 0.52 0.15 0.45 0.16 1000 0.53 0.16 0.42 0.14 0.31 0.15 0.52 0.13 0.66 0.14 0.54 0.15 1250 0.6 0.15 0.49 0.13 0.37 0.17 0.63 0.16 0.68 0.14 0.60 0.12 1600 0.63 0.13 0.58 0.16 0.4 0.18 0.61 0.13 0.67 0.14 0.60 0.11 2000 0.66 0.12 0.63 0.17 0.42 0.17 0.59 0.08 0.68 0.15 0.60 0.13 2500 0.71 0.13 0.63 0.18 0.51 0.16 0.64 0.1 0.72 0.17 0.69 0.13 3150 0.74 0.14 0.65 0.2 0.65 0.16 0.7 0.12 0.71 0.18 0.70 0.14 4000 0.7 0.17 0.63 0.28 0.76 0.15 0.69 0.16 0.64 0.18 0.68 0.16 5000 0.53 0.22 0.49 0.27 0.67 0.2 0.62 0.19 0.49 0.17 0.53 0.16 6300 0.38 0.19 0.34 0.19 0.35 0.2 0.46 0.19 0.44 0.13 0.36 0.15 8000         0.33 0.24 0.26 0.19 Notes:  Center frequencies for Shahnaz & Bork were (Hz): 250, 315, 397, 500, 630, 794, 1000, 1260, 1587, 2000, 2520, 3175, 4000, 5040, 6000 Center frequencies for Keefe et al. (1993), Feeney et al. (2004), and Voss & Allen (1994) were the same as Shahnaz & Bork (2006), with the exception that their highest frequency was measured at 6350 Hz.  Center frequencies for Polat et al. (2015) were: 224, 324, 385, 500, 629, 793, 1000, 1259, 1587, 2000, 2519, 3174, 4000, 5039, 6349, 8000  

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.24.1-0375802/manifest

Comment

Related Items