International Journal of Audiology 2015; 54: 182–189
Original Article
Relations between psychophysical measures of spatial hearing and self-reported spatial-hearing abilities T.E.M. Van Esch*, M.E. Lutman$, M. Vormann†, J. Lyzenga‡, M. Hällgren#, B. Larsby#, S.P. Athalye$, T. Houtgast‡, B. Kollmeier† & W.A. Dreschler* *Department
of Clinical and Experimental Audiology, Academic Medical Centre, Amsterdam, The Netherlands, †HörTech GmbH, Oldenburg, of ENT/Audiology, Free University Hospital, Amsterdam, The Netherlands, #Department of Technical Audiology, University Germany Hospital, Linköping University, Linköping, Sweden and $Institute for Sound and Vibration Research, University of Southampton, UK ‡Department
Abstract Objective: The aim of the present study was to investigate how well the virtual psychophysical measures of spatial hearing from the preliminary auditory profile predict self-reported spatial-hearing abilities. Design: Virtual spatial-hearings tests (conducted unaided, via headphones) and a questionnaire were administered in five centres in Germany, the Netherlands, Sweden, and the UK. Correlations and stepwise linear regression models were calculated among a group of hearing-impaired listeners. Study sample: Thirty normal-hearing listeners aged 19–39 years, and 72 hearing-impaired listeners aged 22–91 years with a broad range of hearing losses, including asymmetrical and mixed hearing losses. Results: Several significant correlations (between 0.24 and 0.54) were found between results of virtual psychophysical spatial-hearing tests and self-reported localization abilities. Stepwise linear regression analyses showed that the minimum audible angle (MAA) test was a significant predictor for self-reported localization abilities (5% extra explained variance), and the spatial speech reception threshold (SRT) benefit test for self-reported listening to speech in spatial situations (6% extra explained variance). Conclusions: The MAA test and spatial SRT benefit test are indicative measures of everyday binaural functioning. The binaural SRT benefit test was not found to predict self-reported spatial-hearing abilities.
Key Words: Spatial hearing; hearing impaired; clinical tests; multi-centre study; audiological diagnosis; auditory profile.
Spatial hearing has a large influence on the degree of handicap experienced by HI persons (Kramer et al, 1998; Gatehouse & Noble, 2004; Noble & Gatehouse, 2004). Spatial hearing can be estimated via several disability and handicap questionnaires, such as the speech, spatial and qualities of hearing scale (SSQ, Gatehouse & Noble, 2004), the spatial hearing questionnaire (Tyler et al, 2009), and the Gothenburg profile (Arlinger et al, 1998; Ringdahl et al, 1998). There are also a wide range of psychophysical tests that measure binaural or spatial hearing. Besides many virtual and free-field localization tests, the minimum audible angle test (MAA, Grantham et al, 2003) and spatial speech perception tests (Goverts & Houtgast, 2010; Van Esch et al, 2013) are examples of binaural tests (see also Flamme (2001) for an overview of questionnaires and tests). In order to know which psychophysical tests are indicative of disability in everyday hearing, comparison of self-reported data with psychophysical test results is essential. Nevertheless, only a
few researchers have investigated the correspondence between selfreported sound localization abilities and psychophysical measures of binaural hearing. Tyler et al (2009) found a significant correlation of 0.34 between the factor “directionality” of their spatial hearing questionnaire and results of a free-field localization test in listeners fitted with one or two cochlear implants. Gatehouse & Akeroyd (2006) found several significant partial correlations (ranging between 0.15 and 0.23) between subscales of the SSQ and results of a dynamic binaural hearing test, but not for the localization subscale. Kramer et al (1996) reported significant correlation of 0.46 between the auditory localization factor of the Amsterdam inventory for auditory disability and handicap, and a sound localization test. Recently, the preliminary auditory profile test battery has been developed (Van Esch et al, 2013). Van Esch et al (2013) described the composition and evaluation of this test battery and presented reference data from an international multi-centre study with over
Correspondence: Thamar E.M. van Esch, Clinical and Experimental Audiology, ENT Department D2-211, Amsterdam Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. E-mail: [email protected] (Received 4 July 2013; accepted 4 August 2014 ) ISSN 1499-2027 print/ISSN 1708-8186 online © 2014 British Society of Audiology, International Society of Audiology, and Nordic Audiological Society DOI: 10.3109/14992027.2014.953216
Psychophysical and self-reported measures of spatial hearing
Abbreviations Acalos bb cu GP GPloc HearCom HI hp HRTF lp MAA MCL NH OMA PTA Q6
Q7
SD SPL SRT
Adaptive categorical loudness scaling Broadband Categorical unit Gothenburg profile Localization category of the Gothenburg profile Hearing in the communication society Hearing impaired High pass Head-related transfer function Low pass Minimum audible angle Most comfortable level Normally hearing Oldenburg measurement applications Pure-tone average (0.5, 1, 2, 4 kHz) Question #6 of GP: Are there occasions when you cannot localize different sounds in the traffic? Question #7 of GP: Are there occasions when you turn your head in the wrong direction, when someone calls you? Standard deviation Sound pressure level Speech reception threshold
100 NH and HI listeners from five centres in Germany, the Netherlands, Sweden, and the UK. The test battery contains psychophysical spatial hearing tests and a questionnaire that includes a part on sound localization, as well as tests of loudness perception, spectral and temporal resolution, speech perception in quiet and in noise, cognitive abilities, and listening effort. This makes the large dataset suitable for thorough comparisons between psychophysical and selfreported localization abilities. In the preliminary auditory profile test battery, spatial hearing was measured psychophysically by means of a virtual minimum audible angle (MAA) test, and by two virtual spatial speech reception threshold (SRT) tests in noise, the spatial SRT benefit test and the binaural SRT benefit test (Van Esch et al, 2013). A virtual version of the minimum audible angle (MAA) test was chosen (Mills, 1985; Hafter et al, 1992; Grantham et al, 2003). Van Esch et al (2013) reported a small learning effect but no significant effect of test centre, and good correspondence between test and retest measurements. Some of the HI listeners (⬍ 10%) found the test too hard to complete. MAA results from the preliminary auditory profile test battery were closely comparable to previously published results (Van Esch et al, 2013). The spatial SRT benefit test measures the SRT-improvement when speech and noise are spatially separated, while the binaural SRT benefit test measures the SRT-improvement of binaural versus monaural listening. No significant learning effects were found in both tests. A small ear effect, in accordance with the well-known right-ear advantage for speech perception (Tervaniemi & Hugdahl, 2003) was found, but was considered unimportant for the interpretation of individual results. A small effect of test centre was significant in HI listeners for the binaural SRT benefit test (with German data being deviant from others), but not for the spatial SRT benefit test. Furthermore, close correspondence between test and retest measurements was found (intraclass correlation coefficients between 0.6 and 0.8), and spatial and binaural SRT benefit results from the prelimi-
183
nary auditory profile test battery were in agreement with previously published results (Van Esch et al, 2013). Self-reported hearing disability in terms of sound localization was assessed in the preliminary auditory profile test battery by means of the Gothenburg profile (GP, Arlinger et al, 1998; Ringdahl et al, 1998). The Gothenburg profile has two subscales with two categories each: subscale experienced hearing disability, with categories hearing speech and sound localization; and subscale handicap, with categories social settings and personal reactions. For the present paper only the sound localization category is relevant, so we only use results from that category. GP results from the preliminary auditory profile test battery were in good agreement with previously published results (Ringdahl et al, 1998). No significant learning effects were found and correspondence between test and retest measurements was almost perfect. Differences between test centres were significant for the localization category for NH listeners but not for HI listeners. Since the auditory profile is designed to be used for audiological assessment, all auditory tests were conducted unaided. Likewise, all listeners were instructed to fill out the questionnaire for the unaided situation. To achieve equal subjective loudness for all listeners, individual presentation levels based on a loudness-scaling experiment were applied. Presentation levels were determined for broadband, low-frequency and high-frequency separately. Regarding the representativeness for everyday hearing, testing unaided can be a limitation, since some listeners wear hearing aids in daily life. Moreover, filling out the questionnaire for the unaided situation might have been hard for some of the hearing-aid users. The aim of the present paper is to gain insight in the representativeness of the psychophysical measures of spatial hearing for everyday listening, which is yet unknown. As self-reported spatial-hearing abilities presumably reflect everyday listening, we compared psychophysical and self-reported measures of binaural hearing from the preliminary auditory profile dataset to each other. The results of each individual test from the preliminary auditory profile were described by Van Esch et al (2013), here we examine the relations between the tests. In clinical practice, where the pure-tone audiogram is usually available, measuring psychophysical spatial hearing is only meaningful if the spatial hearing tests add information about self-reported spatial-hearing abilities to the pure-tone audiogram and age. Therefore, we also examined the added value (in addition to the pure-tone audiogram and age) of the psychophysical measures of spatial hearing in predictions of self-reported spatial-hearing abilities.
Methods The materials and methods of the experiments were described in detail by Van Esch et al (2013). Here, a shortened version of the general methods, as well as brief descriptions of the tests to be examined in this paper, are presented.
Test set-up The tests were implemented on the Oldenburg measurement application (OMA), which is a combined software and hardware test platform. Tests ran on a PC and sounds were played via an RME soundcard (type Fireface 800, DIGI96/8 PAD or HDSP 9632) and fed through an amplifier to Sennheiser HDA 200 headphones. Experiments took place in sound-insulated booths. Written instructions were translated in the four languages (Dutch, English, German, and Swedish) and used in all centres, complemented with oral explanations when needed.
184
T.E.M. Van Esch et al.
Centres and listeners The five participating centres were audiological centres from the Academic Medical Centre, Amsterdam, The Netherlands (NLAMC); Hörzentrum Oldenburg GmbH, Oldenburg, Germany (DEHZO); Free University Hospital, Amsterdam, The Netherlands (NL-VUMC); Linköping University, Linköping, Sweden, (SELINK); and the Institute for Sound and Vibration Research, Southampton, UK (UK-ISVR) (Van Esch et al, 2013). Seventy-two HI listeners participated in the multi-centre study: 12 from UK-ISVR participated, and 15 from each of the other centres (NL-AMC, DEHZO, SE-LINK and NL-VUMC). All centres were approved by their local research ethics committees for the conduct of the study1, in accordance with the Declaration of Helsinki, and all listeners gave written informed consent to participate in the study. HI listeners were aged 22 to 91 years (mean: 63). Pure-tone audiometry was conducted prior to the test session using a clinical audiometer calibrated according to ISO 389-1 (1998). The mean air-conduction audiograms of left and right ears are shown in Figure 1 for HI listeners with purely sensorineural losses (n ⫽ 58), and HI listeners with conductive components (n ⫽ 14). The majority of listeners had symmetric hearing losses, but there were 13 listeners in the HI group with an asymmetry of 10 dB or more (averaged over 0.5, 1, 2, and 4 kHz). Further details about the listeners and about conductive and asymmetric hearing losses were described by Van Esch et al (2013). In the present paper all the HI listeners were combined into one group.
Protocol All auditory tests were conducted unaided and binaurally via headphones. After the pure-tone audiogram, a loudness scaling test (Acalos test, developed by Brand and Hohmann, 2002) was conducted. In the Acalos test, listeners judged loudness on a scale, based on which the stimulus level was adaptively varied. For three stimulus types (broadband noise and narrow-band noises at 0.5 and 3 kHz), individual loudness growth curves were fit. From these curves most-comfortable levels (MCL) were calculated as the level corresponding to the perceived loudness of 20 categorical units (cu), on a scale from 0 to 50. These MCLs were used as individual presen-
tation levels in subsequent tests; the presentation levels were limited to maxima of 95 and 85 dB SPL for narrowband and broadband signals, respectively. This way, all auditory tests were conducted at similar subjective loudness levels to obtain similar overall audibility for all listeners. This approach has both advantages and disadvantages as discussed by Van Esch et al (2013). The tests were conducted in test and retest in two sessions on separate days (one to three weeks apart). Since learning effects in the spatial and binaural SRT benefit tests and the GP are minimal (Van Esch et al, 2013), means of test and retest values are used as the pooled measures of these tests in the present analyses. Because of the known learning effect in the MAA data (Van Esch et al, 2013), retest data from the MAA test are used.
Test procedures All auditory profile tests were described in detail by Van Esch et al (2013). Here, brief descriptions of the tests involved in the present analyses are provided.
MAA TEST A virtual version of the MAA test (Grantham et al, 2003) was used, which measures the just-noticeable difference in horizontal sound azimuth. The stimuli were noise bursts filtered with generic head-related transfer functions (HRTFs, see Silzle, 2007) for different directions to simulate the different spatial locations. The generic HRTFs were measured in human ears with 15° resolution. To interpolate for intermediate locations, delays were removed before interpolating magnitudes and adding interpolated delays (Silzle, 2007). In each trial, two stimuli were presented consecutively from different directions, symmetrically spaced on different sides of the straight-ahead direction. The order of the sounds was randomized. The two sounds were perceived as a sound that was either first left, then right, or vice versa. The task for the listener was to indicate the order (direction) of the two sounds. Three different stimulus sets were used and were presented at MCL: broadband white noise (0.02 kHz to 20 kHz), low-pass noise (cut-off frequency 0.5 kHz, lower limit 0.02 kHz), and high-pass noise (cut-off frequency 3 kHz, upper limit 20 kHz). Van Esch et al (2013) presented MAA results for the three stimuli types in their Figure 7.
Figure 1. Average air-conduction hearing thresholds of left and right ears (means ⫾ 1 SD) of NH listeners (circles, dotted lines); and HI listeners with purely perceptive hearing loss (asterisks, solid lines) or conductive components (diamonds, dashed lines). Figure adapted from Van Esch et al (2013).
Psychophysical and self-reported measures of spatial hearing
SPATIAL SRT BENEFIT
TEST AND BINAURAL
SRT BENEFIT TEST
The spatial SRT benefit test measures the improvement in SRT for speech perception in noise when speech and noise are virtually spatially separated (see also Wagener et al, 2006). SRTs for this test were measured using closed-set sentence tests: NL-matrix for Dutch (Koopman et al, 2006), OlSa for German (Wagener et al, 1999a,b,c), UK-matrix for English (Athalye, 2010, Appendix IV), and Hagerman for Swedish (Hagerman, 1982). The noises were speech-shaped stationary noises, fixed at MCL, while the speech level was varied adaptively (Brand & Kollmeier, 2002). Virtual free-field conditions were created by filtering with generic HRTFs from different directions. The speech signal was always filtered with the HRTF of straight ahead (0°) and the noise was filtered either with the HRTF of 0° or with the HRTF of ⫾ 90°. The filtered speech and noise were added and presented dichotically. For measuring both left and right spatial SRT benefit, three SRT measurements were obtained: • • •
S0N0: speech and noise both coming from the front (0°) S0N90: speech coming from the front (0°) and noise coming from the right side (90o) S0N-90: speech coming from the front (0°) and noise coming from the left side (⫺ 90°)
The spatial SRT benefit was calculated as the SRT difference between the S0N90 or S0N⫺ 90 and the S0N0 measurement for the noise-right and noise-left measurements respectively. The binaural SRT benefit gives the SRT benefit for binaural hearing as opposed to monaural hearing with speech and noise from virtually separated locations (see also Wagener et al, 2006a). To calculate the binaural SRT benefit, two monaural measurements were conducted in addition to the spatial SRT benefit measurements, using the same speech material, noises, and procedures: • •
S0LN90L: speech coming from the front (0°) and noise coming from the right side (90o) with the right ear virtually blocked: both signals were presented monaurally to the left ear. S0RN-90R: speech coming from the front (0°) and noise coming from the left side (⫺ 90o) with the left ear virtually blocked: both signals were presented monaurally to the right ear.
The binaural SRT benefit was calculated the SRT difference between S0LN90L and S0N90 and between S0RN⫺ 90R and S0N⫺ 90, for right and left respectively. Uncorrected and corrected spatial and binaural SRT benefit results are presented by van Esch et al (2013, Figure 8).
GOTHENBURG PROFILE To measure self-reported disability and handicap, subjects filled in an online implementation of the Gothenburg profile (Arlinger et al, 1998; Ringdahl et al, 1998). The Gothenburg profile consists of twenty questions, divided over two subscales: experienced hearing disability (categories: hearing speech and sound localization), and handicap (categories: social settings and personal reactions). Listeners were instructed to fill in the questionnaire for the unaided situation. They responded on a 10-point scale. Graphs of corrected GP results (i.e. results after subtraction of NH reference scores) are presented by Van Esch et al, 2013 (Figure 11). Where needed, results were transformed such that lower scores always referred to better performance. In this paper, results from the five questions of the sound localization category were used. This category contains, besides questions about directional hearing, also questions about distance hearing and detection. As a further specification on directional
185
hearing we evaluated separately results from the two questions from the sound localization category on directional hearing: ‘Are there occasions when you cannot localize different sounds in the traffic?’ (question #6, referred to as Q6) and ‘Are there occasions when you turn your head in the wrong direction, when someone calls you?’ (question #7, referred to as Q7).
Statistical methods
CORRECTION FOR
TEST-MATERIAL EFFECTS BETWEEN DIFFERENT
LANGUAGES
Language-validation studies with NH listeners were conducted in separate independent experiments for the language-dependent tests (see Van Esch et al, 2013). The results were used to correct for test-material effects by presenting all outcome measures relative to reference values, based on the average scores of NH listeners for each language. The language-dependent tests comprise the spatial and binaural SRT benefit tests (see also Wagener et al, 2006) and the Gothenburg profile. Results of these three tests showed significant differences between centres, even though the spatial and binaural SRT benefit results are difference scores (see van Esch et al, 2013 for uncorrected results and more information on the differences between centres). In the present paper, corrected data (i.e. data after subtraction of reference values) are presented for the spatial SRT tests and the Gothenburg profile.
NORMALITY The normality of the outcome measures was tested in Van Esch et al (2013) by visual inspection and the Shapiro-Wilk and Kolmogorov-Smirnov tests. In the HI group, all outcome measures except the MAA results were distributed (approximately) normally. Log-transformed MAA data were distributed normally and are used in the present analyses.
LINEAR
REGRESSION AND INCLUSION OF AUDIOGRAM THRESHOLDS
Multiple linear regressions were performed in SPSS on the data from the group of HI listeners. Measures of localization behaviour were used as dependent variables. The models involved stepwise inclusion of possible predictors (inclusion: p ⬍ 0.05, and exclusion: p ⬎ 0.10). As we wanted to investigate the relevance of the preliminary auditory profile tests in addition to the pure-tone audiogram and age, we first included the audiogram measures and age in one block, before including all other measures in a second block. For every analysis, the distribution of residuals was checked for approximate normality, a plot of residuals versus predicted values was checked for linearity and homoscedasticity, and autocorrelation of the residuals was tested using the Durbin-Watson statistic (Durbin & Watson, 1971).
Results The results of each individual test are presented in Van Esch et al (2013). Here we report the correspondence between the psychophysical spatial hearing tests and self-reported spatial-hearing abilities. Table 1 reports the Pearson correlation coefficients and significances between the scores related to spatial hearing from the Gothenburg Profile and the MAA and binaural SRT results. Scatter plots of the same data are presented in Appendix A. As a reference, correlations with listener age and audiogram results are also displayed. Because of the binaural nature of the tests, we chose to use binaural data from the audiogram also: the pure-tone average (0.5, 1, 2, 4 kHz, PTA) and the slope of the audiogram (difference between
186
T.E.M. Van Esch et al. Table 1. Pearson correlation coefficients in the HI group between the corrected Gothenburg profile scores (average localization category score, GPloc, and scores for questions 6 and 7, Q6 and Q7), log-transformed MAA results for broadband (bb), low pass (lp), and high pass (hp) stimuli, left and right spatial and binaural SRT benefit, age, and hearing loss (PTA, slope and asymmetry). Significant correlations at the p ⬍ 0.01 and p ⬍ 0.05 level are marked ** and * respectively. Spatial SRT benefit
MAA
GPloc Q6 Q7
Binaural SRT benefit
Audiogram
bb
lp
hp
left
right
left
right
Age
PTA
slope
asym
0.31* 0.38** 0.29*
0.45** 0.46** 0.21
0.07 0.14 ⫺0.01
0.34** 0.24* 0.48**
0.43** 0.28* 0.54**
0.30* 0.29* 0.39**
⫺0.01 ⫺0.08 0.09
⫺0.19 ⫺0.14 ⫺0.10
0.60** 0.45** 0.59**
⫺0.12 ⫺0.18 ⫺0.16
0.09 0.01 0.18
thresholds at 0.5 and 4 kHz), both averaged over both ears. Moreover, since asymmetry of hearing loss is expected to be relevant for localization abilities, the absolute difference between left and right ear PTAs was also included in the analyses. It can be seen from Table 1 that the correlations between the PTAs and GP results were highly significant (r ⱖ 0.45, p ⬍ 0.01, see also Supplementary Appendix Figure 1 available online at http.// informahealthcare.com/doi/abs/10.3109/14992027.2014.953216). The asymmetry in hearing thresholds, slope of the audiogram, and age did not show significant correlations with self-reported localization abilities. For the MAA test, most correlations with the GP results were significant (see also Supplementary Appendix Figure 2 available online at http.//informahealthcare.com/doi/abs/10.3109 /14992027.2014.953216). The strongest correlations were found between MAAs with broadband and low-pass noises and question #6 (about localizing traffic sounds) and the localization category from the GP (r ⫽ 0.38; r ⫽ 0.45). A significant correlation was found between the MAA measured with broadband noise and question #7 (about turning your head in the direction of someone calling you; r ⫽ 0.29; p ⬍ 0.05). For the high-pass noise condition, no significant correlations with the GP results were found. Significant correlations between the spatial SRT benefit test and the GP results were also found (see also Supplementary Appendix Figure 3 available online at http.//informahealthcare.com/doi/abs/10.3109/14992027.2014. 953216): both noise-left and noise-right measurements correlated significantly with the results from the localization category and question #7 (r ⫽ 0.48; r ⫽ 0.54; both p ⬍ 0.01) and with the results of question #6 (r ⫽ 0.24; r ⫽ 0.28; p ⬍ 0.05). Nevertheless, these correlations were lower than those between PTA and GP. For the binaural SRT benefit test, significant correlations with GP results were found for noise-left measurements (r ⱖ 0.29; p ⬍ 0.05), while the correlations were negligible for noise-right measurements2 (see also Supplementary Appendix Figure 4 available online at http.// informahealthcare.com/doi/abs/10.3109/14992027.2014.953216). Three linear regression analyses were performed in two blocks to find the best predicting variables for the localization category and for questions #6 and #7 of the Gothenburg profile (GPloc, Q6, and Q7). In the first block, the audiogram variables (PTA, asymmetry, and slope) and age were included by a stepwise procedure to find the significant predictors from these measures. Next, in a second block, the other possible predictors (MAA and SRT benefit measures) were entered, also by a stepwise procedure. The results of these analyses are reported in Table 2. For the prediction of GPloc, only results from the models with age and audiogram variables (block 1) are shown, since none of the variables from
the second block proved to contribute significantly to the model. For Q6 and Q7 on the other hand, both results from the model with block 1, and from the model with blocks 1 and 2 are presented. For each model the included variables, values of R and adjusted R2 (corrected for the available degrees of freedom) are shown. The regression formulas for the three outcome measures were: Q6 ⫽ 0.208 ⫹ (0.049 ⫻ PTA) ⫹ (1.894 ⫻ low-pass MAA) Q7 ⫽ 2.724 ⫹ (0.069 ⫻ PTA) ⫹ (0.334 ⫻ spatial SRT benefit (right)) GPloc ⫽ 0.136 ⫹ (0.091 ⫻ PTA) ⫺ (0.033 ⫻ age) These models explained 24%, 40%, and 39% of the variance in Q6, Q7, and GPloc data respectively (adjusted R2 values). All models have a significance of p ⬍ 0.001. Note that the regression equations predict the corrected GP scores in HI listeners (i.e. the calculated scores for HI listeners are relative to the reference scores, so negative scores refer to better than reference), to account for differences between results from different languages. Scatter plots of predicted versus measured GP scores are shown in Figure 2. For all three outcome measures, higher average hearing losses (PTA) were associated with an increase in perceived problems, at a rate of 0.5–0.9 points increment (on a 10-point scale) per 10 dB PTA. Adding the low-pass MAA to the equation significantly improved the prediction of question #6 (about localizing traffic sounds) by 5% (increment of adjusted R2, p ⬍ 0.05). The prediction of question #7 (about turning your head in the direction of the speaker) improved significantly by 6% (increment of adjusted R2, p ⬍ 0.05) by adding spatial SRT benefit (right side) to the equation. Both for questions Table 2. Results of stepwise linear regression analyses of GP data. For three dependents (Q6, Q7, and GPloc) significant predictors of the stepwise linear regression models are shown, as well as R and adjusted R2 values of these models. All shown models have a significance of p ⬍ 0.001. For Q6 and GPloc, none of the significant predictors from the second block proved to be significant. Dependent Q6 Q7
GPloc
Model
Predictors
R
Adjusted R2
block 1 block 1& 2 block 1 block 1 & 2
PTA PTA, MAA lp PTA PTA, spatial SRT benefit (right) PTA, age
0.45 0.52 0.59 0.65
0.19 0.24 0.34 0.40
0.65
0.39
blocks 1
Psychophysical and self-reported measures of spatial hearing GP localization
Question #6
8
4 2 0 –2
6 4 2 0 –2
–2
0
2
4
6
Measured scores
8
Question #7
8 Predicted scores
6
Predicted scores
Predicted scores
8
187
6 4 2 0 –2
–2
0
2
4
6
Measured scores
8
–2
0
2
4
6
8
Measured scores
Figure 2. Predicted (vertical axes) versus measured (horizontal axes) corrected Gothenburg profile scores for the sound localization category (left panel), question #6 (middle panel), and question #7 (right panel).
#6 and #7, poorer psychophysical scores correspond to more selfreported problems. For the localization category, age proved to be a significant predictor besides PTA, accounting for a 5% increase in adjusted R2 (p ⬍ 0.05). On average, after having accounted for their hearing loss, older listeners reported less problems. Note that average hearing loss proved to be the only significant audiogram predictor for all evaluated GP measures; slope and asymmetry did not add any predictive power. Like in the correlation analysis, asymmetry of hearing loss did not play a significant role. Moreover, the binaural SRT benefit results did not make a significant contribution to the regression models.
Discussion The aim of the present study was to investigate to what extent the psychophysical measures of spatial hearing of the preliminary auditory profile (Van Esch et al, 2013) are indicative of self-reported functioning as measured by the Gothenburg profile (Arlinger et al, 1998; Ringdahl et al, 1998). To that end, we examined the correspondence between the Gothenburg profile localization scores and the MAA and the spatial and binaural SRT benefit results, both by correlation and linear regression analyses. When interpreting these analyses, it has to be recognized that the correspondence between objective and self-reported localization abilities depends on both validity of the psychophysical tests and validity of the questionnaire. The Swedish version of the GP has been validated in a large group of HI listeners (see Ringdahl et al, 1998). A potential drawback of the present GP data is that listeners were instructed to fill out the questionnaire for the unaided situation (as they were also tested without hearing aids), which might have been hard for some of the hearing-aid users. Additionally, although no significant test centre effect was found in the localization category of the GP (Van Esch et al, 2013), non-significant centre effects might have contaminated the data. Regarding the psychophysical tests, the use of generic HRTFs might have decreased the reliability as they do not guarantee an equal applicability for all listeners (see Middlebrooks, 1999a and 1999b, for a comparison of virtual localization with own-ear versus other-ear HRTFs). For the MAA measurements, several significant correlations, ranging from 0.29 to 0.45 (p ⬍ 0.01 and p ⬍ 0.05, see Table 1) were found for the broadband and (mostly) for the low-pass noise conditions, but not for the high-pass noise condition. This correspondence between MAA results and GP localization measures has not been demonstrated before and was not anticipated, as the MAA test is a
spatial discrimination measure rather than a measure of location perception. Moreover, the stepwise linear regression analysis showed that adding the low-pass MAA results to the regression model, after inclusion of PTA, increased the explained variance of the prediction of question #6 by 5% (p ⬍ 0.05). This indicates that the MAA test has potential added value in predicting the self-reported ability to localize different sounds in traffic. In other words: the results of the stepwise linear regression analysis show that the correspondence between MAA and GP results is not solely a consequence of mutual correlations with the audiogram. Although MAA has been measured in many previous studies (see e.g. Grantham et al, 2003; Hausler et al, 1983), its relationship to self-reported performance has not been described. So, despite the limitations regarding clinical applicability3 of the MAA test reported by Van Esch et al (2013), the MAA test does have some predictive power for self-reported localization performance as measured by the GP. Many spatial SRT experiments have been described (e.g. Goverts & Houtgast, 2010; Johansson & Arlinger, 2002; Wilson et al, 1982), but to our knowledge, the relationship with self-reported spatial hearing abilities has not been investigated before. Our spatial SRT benefit results showed significant correlations, ranging from 0.24 to 0.54, for both noise-left and noise-right measurements, for question #6 (p ⬍ 0.05), and more strongly for question #7 and the GP localization category (p ⬍ 0.01, see Table 1). It is interesting to see that the highest correlations (0.48 and 0.54 for left and right, both p ⬍ 0.01) were found between the spatial SRT benefit test and question #7, which asked about head turning in the direction of a speaker. Compared to correlations between GP and PTA however, correlations with spatial SRT benefit were of the same order of magnitude and significance, or lower. This could indicate that at least a substantial part of these correlations was caused by underlying correlations with PTA. Nonetheless, the stepwise linear regression analysis showed that adding spatial SRT benefit to the regression model (after inclusion of PTA) improved the prediction of question #7 results by a 6% increment of adjusted R2 (p ⬍ 0.05). So, the spatial SRT benefit test has some added value in predicting self-reported binaural functioning (as measured by the GP), especially for listening to speech in spatial situations. To our best knowledge, the relevance of this test for self-reported spatial hearing performance has not been evaluated before. Considering the binaural SRT benefit test, the correlations between GP and the noise-left measurement (i.e. the measurement with noise from the left side) were between 0.29 and 0.39 (p ⬍ 0.05). For the noise-right measurement on the other hand the correlations were
188
T.E.M. Van Esch et al.
very low and even negative (ranging from ⫺ 0.08 to 0.09)4. This is confirmed by the scatter plots in Supplementary Figure 4 (Appendix A available online at http.//informahealthcare.com/doi/ abs/10.3109/14992027.2014.953216). Linear regression analyses nevertheless showed that results from the binaural SRT benefit test did not contribute significantly to the prediction of GP data. As a result, the binaural SRT benefit test does not appear to be an indicative measure of spatial hearing, when related to self-report on the GP questionnaire. Another finding from the linear regression analyses is that age proved to be a significant predictor for the GP localization category, accounting for 5% of explained variance (adjusted R2, p ⬍ 0.05). We found that older listeners reported (on average, after having accounted for their hearing loss) less problems. A hypothetical reason for this result is that it is more common among older listeners to have a hearing loss and that older listeners therefore report less problems due to a better acceptance of their disabilities. Lutman (1991) and Gatehouse (1991) described similar effects of the expectations of older listeners on their self-reported disability. Regarding the pure-tone audiogram, correlation and regression analyses showed that in the present data set only the pure-tone average was predictive of self-reported localization performance, accounting for 34%, 19%, and 34% of explained variance for GPloc, Q6, and Q7 respectively (adjusted R2 values). We repeated the regression analyses with a binaural weighted average of the audiogram (4:1 weighting of better:worse, see e.g. Lutman et al, 1987) and the results were almost identical. Slope of the audiogram, and asymmetry of hearing loss did not show any significant correlations with, or predictive power for, self-reported localization performance. However, the present study was not designed to examine the influence of asymmetry on spatial hearing, and due to the small number of listeners with asymmetric hearing loss (13 listeners), a significant correlation between asymmetry and self-reported localization in the full HI group was not expected in the present data. In general, the percentages of variance in GP results that could be explained by the psychophysical measures including the pure-tone audiogram were below 50% (see Table 2). These numbers have to be interpreted relative to the test-retest reliability of the GP. As the correlation between test and retest results of the localization category was 0.92, the percentage of explained variance from the regression models could not meaningfully exceed 84% (0.922) if the predicting variables were exactly accurate. Due to uncertainty in the predicting variables, the percentage of explained variance will be even lower. Nevertheless, the values we found may still seem low. There are important differences between virtual laboratory free-field measurements of spatial hearing compared with real life experiences that probably have played a role. Clearly, real-life localization involves much more than the detection of sounds (as measured with the pure-tone audiogram) and pure binaural hearing (as measured with the psychophysical tests). In the outside world, spatial hearing is influenced by level differences, reflections, background noise, and dynamic situations. The importance of the dynamic aspects of real life listening situations was argued to be important by Gatehouse & Noble (2004) and Noble & Gatehouse (2004). Gatehouse & Akeroyd (2006) suggested that a more dynamic localization test would have shown stronger correspondence to self-reported functioning. None of the above-mentioned aspects of real-life localization are captured by the auditory profile spatial hearing tests. Moreover, in the auditory profile spatial hearing tests averaged HRTFs were used, which do not exactly match the real world situation for individual listeners. Therefore, and also because of the measurement inaccuracy of each
measure, a considerable proportion of unexplained variance in the GP data was to be expected.
Conclusions We conclude that the MAA test and the spatial SRT benefit test are indicative measures for self-reported binaural functioning. The MAA test best represents self-reported localization, while the SRT benefit test was found to be the most important for self-reported listening to speech in spatial situations. The binaural SRT benefit test was not found to be relevant for self-reported spatial hearing performance.
Acknowledgements Data from the present study have been published before by Van Esch et al (2013). Materials of several of the tests from the preliminary auditory profile are available from the HearCom website (www. hearcom.eu): Matrix sentence frameworks in the four languages, and the Gothenburg profile questionnaire in the four languages. We thank Daniel Berg for technical support and implementation of the tests on the platform (OMA), and Kirsten Wagener for her work on consolidation of the speech tests. Finally we thank the test subjects for their participation. The study was supported by grants from the European Union FP6, Project 0004171 HEARCOM. The information in this document is provided as is, and no guarantee or warranty is given that the information is fit for any particular purpose. The user thereof uses the information at his/her sole risk and liability. The authors alone are responsible for the content and writing of the paper. Parts of this work have been presented at the International Hearing Aid Research Conference, August 16, 2008, Lake Tahoe, California: Evaluation of the ‘Auditory Profile’ test battery in an international multi-centre study.
Notes 1. NL-AMC: no. 05/127 # 05.17.0934, dated August 3, 2005; HZO-DE: “Klinische Tests zur Bestimmung individueller Hördefizite und Kommuniationsfähigkeiten”, dated November 15, 2006; ISVR-UK: 791, dated February 13, 2007; SE-LINK: M83-06; VUMC-NL: MEC05/12 - 2006/171, dated November 2, 2006. 2. Spearman’s rank correlations (which are less sensitive to outliers) were calculated to investigate whether the significant correlation for noise-left measurements were influenced by outliers. We found that Spearman’s correlations between GP and BILD were very similar to, and of the same significance as, Pearson’s rank correlations: 0,314 (p ⬍ 0.01), 0,284 (p ⬍ 0.05) and 0,419 (p ⬍ 0.01) for the localization subscale and questions #6 and #7 respectively. This indicates that the influence of outliers on the correlation values was negligible. Also for the other variables (PTA, MAA, and spatial SRT benefit) Pearson’s correlations and Spearman’s rank correlations were nearly identical (data not shown). 3. Van Esch et al (2013) found that the low-pass noise condition of the MAA test was too hard for a considerable number of listeners. Additionally, they reported a significant learning effect for the MAA test. 4. Van Esch et al (2013) found in the same data set that listeners perceived significantly more binaural benefit with noise from the left side (after correction for their ear-specific hearing thresholds),
Psychophysical and self-reported measures of spatial hearing which agrees with the generally known right-ear advantage for speech perception (see Tervaniemi & Hugdahl, 2003, for a review). The present analyses suggest that the binaural benefit with noise from the left side was also more closely related to subjective binaural functioning. If both findings can be replicated in other studies, it implies that the right-ear advantage is related to everyday binaural functioning.
Declaration of interest: The authors report no conflicts of interest.
References Arlinger S., Billermark E., Oberg M., Lunner T. & Hellgren J. 1998. Clinical trial of a digital hearing aid. Scand Audiol, 27, 51–61. Athalye S.P. 2010. Factors affecting speech recognition in noise and hearing loss in adults with a wide variety of auditory capabilities. PhD thesis, Institute of Sound and Vibration Research, Southampton, UK. Brand T. & Hohmann H. 2002. An adaptive procedure for categorical loudness scaling. J Acoust Soc Am, 112, 1597–1604. Brand T. & Kollmeier B. 2002. Efficient adaptive procedures for threshold and concurrent slope estimates for psychophysics and speech intelligibility tests. J Acoust Soc Am, 111, 2801–2810. Durbin J. & Watson G.S. 1971. Testing for serial correlation in least squares regression. III. Biometrika, 58, 1–19. Flamme G.A. 2001. Examination of the validity of auditory traits and tests. Trends Amplif, 5, 111–138. Gatehouse S. 1991. The role of non-auditory factors in measured and selfreported disability. Acta Oto-Laryngol, 249–256. Gatehouse S. & Akeroyd M. 2006. Two-eared listening in dynamic situations. Int J Audiol, 45, S120–S124. Gatehouse S. & Noble W. 2004. The speech, spatial and qualities of hearing scale (SSQ). Int J Audiol, 43, 85–99. Goverts S.T. & Houtgast T. 2010. The binaural intelligibility level difference in hearing-impaired listeners: The role of supra-threshold deficits. J Acoust Soc Am, 127, 3073–3084. Grantham D.W., Hornsby B.W. & Erpenbeck E.A. 2003. Auditory spatial resolution in horizontal, vertical, and diagonal planes. J Acoust Soc Am, 114, 1009–1022. Hafter E.R., Saberi K., Jensen E.R. & Briolle F. 1992. Localization in an echoic environment. Adv Biosci, 83, 555–561. Hagerman B. 1982. Sentences for testing speech intelligibility in noise. Scand Audiol, 11, 79–87. Hausler R., Colburn S. & Marr E. 1983. Sound localization in subjects with impaired hearing. Spatial-discrimination and interaural-discrimination tests. Acta Otolaryngol, Suppl. 400, 1–62. Johansson M.S. & Arlinger S.D. 2002. Binaural masking level difference for speech signals in noise. Int J Audiol, 41, 279–284. Koopman J., Houben R., Dreschler W. & Verschuure J. 2006. Development of a speech in noise test (Matrix). 8th EFAS Congress; 10th DGA Congress; 6–9 June 2006, Heidelberg, Germany.
Supplementary material available online Supplementary Appendix Figures 1–4.
189
Kramer S.E., Kapteyn T.S., Festen J.M. & Tobi H. 1996. The relationships between self-reported hearing disability and measures of auditory disability. Audiology 35, 277–287. Kramer S.E., Kapteyn T.S. & Festen J.M. 1998. The self-reported handicapping effect of hearing disabilities. Audiology, 37, 302–312. Lutman M.E. 1991. Hearing disability in the elderly. Acta Oto-Laryngol, 239–248. Lutman M.E., Brown E.J. & Coles R.R.A. 1987. Self-reported disability and handicap in the population in relation to pure-tone threshold, age, sex, and type of hearing loss. Brit J Audiol, 21, 45–58. Middlebrooks J.C. 1999a. Individual differences in external-ear transfer functions reduced by scaling in frequency. J Acoust Soc Am, 106, 1480–1492. Middlebrooks J.C. 1999b. Virtual localization improved by scaling nonindividualized external-ear transfer functions in frequency. J Acoust Soc Am, 106, 1493–1510. Mills A.W. 1958. On the minimum audible angle. J Acoust Soc Am, 30, 237–246. Noble W., Ter-Horst K. & Byrne D. 1995. Disabilities and handicaps associated with impaired auditory localization. J Am Acad Audiol, 6, 129–140. Noble W. & Gatehouse S. 2004. Interaural asymmetry of hearing loss, Speech, Spatial and Qualities of Hearing Scale (SSQ) disabilities, and handicap. Int J Audiol, 43, 100–114. Ringdahl A., Eriksson-Mangold M. & Andersson G. 1998. Psychometric evaluation of the Gothenburg Profile for measurement of experienced hearing disability and handicap: applications with new hearing-aid candidates and experienced hearing-aid users. Br J Audiol, 32, 375–385. Silzle A. 2007. Generation of quality taxonomies for auditory virtual environments by means of systematic expert survey. PhD thesis, Ruhr-Universität Bochum, Germany. Tervaniemi M. & Hugdahl K. 2003. Lateralization of auditory-cortex functions. Brain Res Rev, 43, 231–246. Tyler R.S., Perreau A.E. & Ji H. 2009. Validation of the spatial hearing questionnaire. Ear Hear, 30, 466–474. Van Esch T.E., Kollmeier B., Vormann M., Lyzenga J., Houtgast T. et al. 2013. Evaluation of the preliminary auditory profile test battery in an international multi-centre study. Int J Audiol, 52, 305–321. Villchur E. 1974. Simulation of the effect of recruitment on loudness relationships in speech. J Acoust Soc Am, 56, 1601–1611. Wagener K., Brand T. & Kollmeier B. 1999a. Development and evaluation of a German sentence test part I: Design of the Oldenburg sentence test. Z Audiol, 38, 4–15. Wagener K., Brand T. & Kollmeier B. 1999b. Development and evaluation of a German sentence test part II: Optimization of the Oldenburg sentence test. Z Audiol, 38, 44–56. Wagener K., Brand T. & Kollmeier B. 1999c. Development and evaluation of a German sentence test part III: Evaluation of the Oldenburg sentence test. Z Audiol, 38, 86–95. Wagener K., Brand T. & Kollmeier B. 2006. International cross-validation of sentence intelligibility tests. 8th EFAS Congress; 10th DGA Congress, 6–9 June 2006, Heidelberg, Germany. Wilson R.H., Hopkins J.L., Mance C.M. & Novak R.E. 1982. Detection and recognition masking-level differences for the individual CID W-1 spondaic words. J Speech Hear Res, 25, 235–242.
Copyright of International Journal of Audiology is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
Original Article
Relations between psychophysical measures of spatial hearing and self-reported spatial-hearing abilities T.E.M. Van Esch*, M.E. Lutman$, M. Vormann†, J. Lyzenga‡, M. Hällgren#, B. Larsby#, S.P. Athalye$, T. Houtgast‡, B. Kollmeier† & W.A. Dreschler* *Department
of Clinical and Experimental Audiology, Academic Medical Centre, Amsterdam, The Netherlands, †HörTech GmbH, Oldenburg, of ENT/Audiology, Free University Hospital, Amsterdam, The Netherlands, #Department of Technical Audiology, University Germany Hospital, Linköping University, Linköping, Sweden and $Institute for Sound and Vibration Research, University of Southampton, UK ‡Department
Abstract Objective: The aim of the present study was to investigate how well the virtual psychophysical measures of spatial hearing from the preliminary auditory profile predict self-reported spatial-hearing abilities. Design: Virtual spatial-hearings tests (conducted unaided, via headphones) and a questionnaire were administered in five centres in Germany, the Netherlands, Sweden, and the UK. Correlations and stepwise linear regression models were calculated among a group of hearing-impaired listeners. Study sample: Thirty normal-hearing listeners aged 19–39 years, and 72 hearing-impaired listeners aged 22–91 years with a broad range of hearing losses, including asymmetrical and mixed hearing losses. Results: Several significant correlations (between 0.24 and 0.54) were found between results of virtual psychophysical spatial-hearing tests and self-reported localization abilities. Stepwise linear regression analyses showed that the minimum audible angle (MAA) test was a significant predictor for self-reported localization abilities (5% extra explained variance), and the spatial speech reception threshold (SRT) benefit test for self-reported listening to speech in spatial situations (6% extra explained variance). Conclusions: The MAA test and spatial SRT benefit test are indicative measures of everyday binaural functioning. The binaural SRT benefit test was not found to predict self-reported spatial-hearing abilities.
Key Words: Spatial hearing; hearing impaired; clinical tests; multi-centre study; audiological diagnosis; auditory profile.
Spatial hearing has a large influence on the degree of handicap experienced by HI persons (Kramer et al, 1998; Gatehouse & Noble, 2004; Noble & Gatehouse, 2004). Spatial hearing can be estimated via several disability and handicap questionnaires, such as the speech, spatial and qualities of hearing scale (SSQ, Gatehouse & Noble, 2004), the spatial hearing questionnaire (Tyler et al, 2009), and the Gothenburg profile (Arlinger et al, 1998; Ringdahl et al, 1998). There are also a wide range of psychophysical tests that measure binaural or spatial hearing. Besides many virtual and free-field localization tests, the minimum audible angle test (MAA, Grantham et al, 2003) and spatial speech perception tests (Goverts & Houtgast, 2010; Van Esch et al, 2013) are examples of binaural tests (see also Flamme (2001) for an overview of questionnaires and tests). In order to know which psychophysical tests are indicative of disability in everyday hearing, comparison of self-reported data with psychophysical test results is essential. Nevertheless, only a
few researchers have investigated the correspondence between selfreported sound localization abilities and psychophysical measures of binaural hearing. Tyler et al (2009) found a significant correlation of 0.34 between the factor “directionality” of their spatial hearing questionnaire and results of a free-field localization test in listeners fitted with one or two cochlear implants. Gatehouse & Akeroyd (2006) found several significant partial correlations (ranging between 0.15 and 0.23) between subscales of the SSQ and results of a dynamic binaural hearing test, but not for the localization subscale. Kramer et al (1996) reported significant correlation of 0.46 between the auditory localization factor of the Amsterdam inventory for auditory disability and handicap, and a sound localization test. Recently, the preliminary auditory profile test battery has been developed (Van Esch et al, 2013). Van Esch et al (2013) described the composition and evaluation of this test battery and presented reference data from an international multi-centre study with over
Correspondence: Thamar E.M. van Esch, Clinical and Experimental Audiology, ENT Department D2-211, Amsterdam Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. E-mail: [email protected] (Received 4 July 2013; accepted 4 August 2014 ) ISSN 1499-2027 print/ISSN 1708-8186 online © 2014 British Society of Audiology, International Society of Audiology, and Nordic Audiological Society DOI: 10.3109/14992027.2014.953216
Psychophysical and self-reported measures of spatial hearing
Abbreviations Acalos bb cu GP GPloc HearCom HI hp HRTF lp MAA MCL NH OMA PTA Q6
Q7
SD SPL SRT
Adaptive categorical loudness scaling Broadband Categorical unit Gothenburg profile Localization category of the Gothenburg profile Hearing in the communication society Hearing impaired High pass Head-related transfer function Low pass Minimum audible angle Most comfortable level Normally hearing Oldenburg measurement applications Pure-tone average (0.5, 1, 2, 4 kHz) Question #6 of GP: Are there occasions when you cannot localize different sounds in the traffic? Question #7 of GP: Are there occasions when you turn your head in the wrong direction, when someone calls you? Standard deviation Sound pressure level Speech reception threshold
100 NH and HI listeners from five centres in Germany, the Netherlands, Sweden, and the UK. The test battery contains psychophysical spatial hearing tests and a questionnaire that includes a part on sound localization, as well as tests of loudness perception, spectral and temporal resolution, speech perception in quiet and in noise, cognitive abilities, and listening effort. This makes the large dataset suitable for thorough comparisons between psychophysical and selfreported localization abilities. In the preliminary auditory profile test battery, spatial hearing was measured psychophysically by means of a virtual minimum audible angle (MAA) test, and by two virtual spatial speech reception threshold (SRT) tests in noise, the spatial SRT benefit test and the binaural SRT benefit test (Van Esch et al, 2013). A virtual version of the minimum audible angle (MAA) test was chosen (Mills, 1985; Hafter et al, 1992; Grantham et al, 2003). Van Esch et al (2013) reported a small learning effect but no significant effect of test centre, and good correspondence between test and retest measurements. Some of the HI listeners (⬍ 10%) found the test too hard to complete. MAA results from the preliminary auditory profile test battery were closely comparable to previously published results (Van Esch et al, 2013). The spatial SRT benefit test measures the SRT-improvement when speech and noise are spatially separated, while the binaural SRT benefit test measures the SRT-improvement of binaural versus monaural listening. No significant learning effects were found in both tests. A small ear effect, in accordance with the well-known right-ear advantage for speech perception (Tervaniemi & Hugdahl, 2003) was found, but was considered unimportant for the interpretation of individual results. A small effect of test centre was significant in HI listeners for the binaural SRT benefit test (with German data being deviant from others), but not for the spatial SRT benefit test. Furthermore, close correspondence between test and retest measurements was found (intraclass correlation coefficients between 0.6 and 0.8), and spatial and binaural SRT benefit results from the prelimi-
183
nary auditory profile test battery were in agreement with previously published results (Van Esch et al, 2013). Self-reported hearing disability in terms of sound localization was assessed in the preliminary auditory profile test battery by means of the Gothenburg profile (GP, Arlinger et al, 1998; Ringdahl et al, 1998). The Gothenburg profile has two subscales with two categories each: subscale experienced hearing disability, with categories hearing speech and sound localization; and subscale handicap, with categories social settings and personal reactions. For the present paper only the sound localization category is relevant, so we only use results from that category. GP results from the preliminary auditory profile test battery were in good agreement with previously published results (Ringdahl et al, 1998). No significant learning effects were found and correspondence between test and retest measurements was almost perfect. Differences between test centres were significant for the localization category for NH listeners but not for HI listeners. Since the auditory profile is designed to be used for audiological assessment, all auditory tests were conducted unaided. Likewise, all listeners were instructed to fill out the questionnaire for the unaided situation. To achieve equal subjective loudness for all listeners, individual presentation levels based on a loudness-scaling experiment were applied. Presentation levels were determined for broadband, low-frequency and high-frequency separately. Regarding the representativeness for everyday hearing, testing unaided can be a limitation, since some listeners wear hearing aids in daily life. Moreover, filling out the questionnaire for the unaided situation might have been hard for some of the hearing-aid users. The aim of the present paper is to gain insight in the representativeness of the psychophysical measures of spatial hearing for everyday listening, which is yet unknown. As self-reported spatial-hearing abilities presumably reflect everyday listening, we compared psychophysical and self-reported measures of binaural hearing from the preliminary auditory profile dataset to each other. The results of each individual test from the preliminary auditory profile were described by Van Esch et al (2013), here we examine the relations between the tests. In clinical practice, where the pure-tone audiogram is usually available, measuring psychophysical spatial hearing is only meaningful if the spatial hearing tests add information about self-reported spatial-hearing abilities to the pure-tone audiogram and age. Therefore, we also examined the added value (in addition to the pure-tone audiogram and age) of the psychophysical measures of spatial hearing in predictions of self-reported spatial-hearing abilities.
Methods The materials and methods of the experiments were described in detail by Van Esch et al (2013). Here, a shortened version of the general methods, as well as brief descriptions of the tests to be examined in this paper, are presented.
Test set-up The tests were implemented on the Oldenburg measurement application (OMA), which is a combined software and hardware test platform. Tests ran on a PC and sounds were played via an RME soundcard (type Fireface 800, DIGI96/8 PAD or HDSP 9632) and fed through an amplifier to Sennheiser HDA 200 headphones. Experiments took place in sound-insulated booths. Written instructions were translated in the four languages (Dutch, English, German, and Swedish) and used in all centres, complemented with oral explanations when needed.
184
T.E.M. Van Esch et al.
Centres and listeners The five participating centres were audiological centres from the Academic Medical Centre, Amsterdam, The Netherlands (NLAMC); Hörzentrum Oldenburg GmbH, Oldenburg, Germany (DEHZO); Free University Hospital, Amsterdam, The Netherlands (NL-VUMC); Linköping University, Linköping, Sweden, (SELINK); and the Institute for Sound and Vibration Research, Southampton, UK (UK-ISVR) (Van Esch et al, 2013). Seventy-two HI listeners participated in the multi-centre study: 12 from UK-ISVR participated, and 15 from each of the other centres (NL-AMC, DEHZO, SE-LINK and NL-VUMC). All centres were approved by their local research ethics committees for the conduct of the study1, in accordance with the Declaration of Helsinki, and all listeners gave written informed consent to participate in the study. HI listeners were aged 22 to 91 years (mean: 63). Pure-tone audiometry was conducted prior to the test session using a clinical audiometer calibrated according to ISO 389-1 (1998). The mean air-conduction audiograms of left and right ears are shown in Figure 1 for HI listeners with purely sensorineural losses (n ⫽ 58), and HI listeners with conductive components (n ⫽ 14). The majority of listeners had symmetric hearing losses, but there were 13 listeners in the HI group with an asymmetry of 10 dB or more (averaged over 0.5, 1, 2, and 4 kHz). Further details about the listeners and about conductive and asymmetric hearing losses were described by Van Esch et al (2013). In the present paper all the HI listeners were combined into one group.
Protocol All auditory tests were conducted unaided and binaurally via headphones. After the pure-tone audiogram, a loudness scaling test (Acalos test, developed by Brand and Hohmann, 2002) was conducted. In the Acalos test, listeners judged loudness on a scale, based on which the stimulus level was adaptively varied. For three stimulus types (broadband noise and narrow-band noises at 0.5 and 3 kHz), individual loudness growth curves were fit. From these curves most-comfortable levels (MCL) were calculated as the level corresponding to the perceived loudness of 20 categorical units (cu), on a scale from 0 to 50. These MCLs were used as individual presen-
tation levels in subsequent tests; the presentation levels were limited to maxima of 95 and 85 dB SPL for narrowband and broadband signals, respectively. This way, all auditory tests were conducted at similar subjective loudness levels to obtain similar overall audibility for all listeners. This approach has both advantages and disadvantages as discussed by Van Esch et al (2013). The tests were conducted in test and retest in two sessions on separate days (one to three weeks apart). Since learning effects in the spatial and binaural SRT benefit tests and the GP are minimal (Van Esch et al, 2013), means of test and retest values are used as the pooled measures of these tests in the present analyses. Because of the known learning effect in the MAA data (Van Esch et al, 2013), retest data from the MAA test are used.
Test procedures All auditory profile tests were described in detail by Van Esch et al (2013). Here, brief descriptions of the tests involved in the present analyses are provided.
MAA TEST A virtual version of the MAA test (Grantham et al, 2003) was used, which measures the just-noticeable difference in horizontal sound azimuth. The stimuli were noise bursts filtered with generic head-related transfer functions (HRTFs, see Silzle, 2007) for different directions to simulate the different spatial locations. The generic HRTFs were measured in human ears with 15° resolution. To interpolate for intermediate locations, delays were removed before interpolating magnitudes and adding interpolated delays (Silzle, 2007). In each trial, two stimuli were presented consecutively from different directions, symmetrically spaced on different sides of the straight-ahead direction. The order of the sounds was randomized. The two sounds were perceived as a sound that was either first left, then right, or vice versa. The task for the listener was to indicate the order (direction) of the two sounds. Three different stimulus sets were used and were presented at MCL: broadband white noise (0.02 kHz to 20 kHz), low-pass noise (cut-off frequency 0.5 kHz, lower limit 0.02 kHz), and high-pass noise (cut-off frequency 3 kHz, upper limit 20 kHz). Van Esch et al (2013) presented MAA results for the three stimuli types in their Figure 7.
Figure 1. Average air-conduction hearing thresholds of left and right ears (means ⫾ 1 SD) of NH listeners (circles, dotted lines); and HI listeners with purely perceptive hearing loss (asterisks, solid lines) or conductive components (diamonds, dashed lines). Figure adapted from Van Esch et al (2013).
Psychophysical and self-reported measures of spatial hearing
SPATIAL SRT BENEFIT
TEST AND BINAURAL
SRT BENEFIT TEST
The spatial SRT benefit test measures the improvement in SRT for speech perception in noise when speech and noise are virtually spatially separated (see also Wagener et al, 2006). SRTs for this test were measured using closed-set sentence tests: NL-matrix for Dutch (Koopman et al, 2006), OlSa for German (Wagener et al, 1999a,b,c), UK-matrix for English (Athalye, 2010, Appendix IV), and Hagerman for Swedish (Hagerman, 1982). The noises were speech-shaped stationary noises, fixed at MCL, while the speech level was varied adaptively (Brand & Kollmeier, 2002). Virtual free-field conditions were created by filtering with generic HRTFs from different directions. The speech signal was always filtered with the HRTF of straight ahead (0°) and the noise was filtered either with the HRTF of 0° or with the HRTF of ⫾ 90°. The filtered speech and noise were added and presented dichotically. For measuring both left and right spatial SRT benefit, three SRT measurements were obtained: • • •
S0N0: speech and noise both coming from the front (0°) S0N90: speech coming from the front (0°) and noise coming from the right side (90o) S0N-90: speech coming from the front (0°) and noise coming from the left side (⫺ 90°)
The spatial SRT benefit was calculated as the SRT difference between the S0N90 or S0N⫺ 90 and the S0N0 measurement for the noise-right and noise-left measurements respectively. The binaural SRT benefit gives the SRT benefit for binaural hearing as opposed to monaural hearing with speech and noise from virtually separated locations (see also Wagener et al, 2006a). To calculate the binaural SRT benefit, two monaural measurements were conducted in addition to the spatial SRT benefit measurements, using the same speech material, noises, and procedures: • •
S0LN90L: speech coming from the front (0°) and noise coming from the right side (90o) with the right ear virtually blocked: both signals were presented monaurally to the left ear. S0RN-90R: speech coming from the front (0°) and noise coming from the left side (⫺ 90o) with the left ear virtually blocked: both signals were presented monaurally to the right ear.
The binaural SRT benefit was calculated the SRT difference between S0LN90L and S0N90 and between S0RN⫺ 90R and S0N⫺ 90, for right and left respectively. Uncorrected and corrected spatial and binaural SRT benefit results are presented by van Esch et al (2013, Figure 8).
GOTHENBURG PROFILE To measure self-reported disability and handicap, subjects filled in an online implementation of the Gothenburg profile (Arlinger et al, 1998; Ringdahl et al, 1998). The Gothenburg profile consists of twenty questions, divided over two subscales: experienced hearing disability (categories: hearing speech and sound localization), and handicap (categories: social settings and personal reactions). Listeners were instructed to fill in the questionnaire for the unaided situation. They responded on a 10-point scale. Graphs of corrected GP results (i.e. results after subtraction of NH reference scores) are presented by Van Esch et al, 2013 (Figure 11). Where needed, results were transformed such that lower scores always referred to better performance. In this paper, results from the five questions of the sound localization category were used. This category contains, besides questions about directional hearing, also questions about distance hearing and detection. As a further specification on directional
185
hearing we evaluated separately results from the two questions from the sound localization category on directional hearing: ‘Are there occasions when you cannot localize different sounds in the traffic?’ (question #6, referred to as Q6) and ‘Are there occasions when you turn your head in the wrong direction, when someone calls you?’ (question #7, referred to as Q7).
Statistical methods
CORRECTION FOR
TEST-MATERIAL EFFECTS BETWEEN DIFFERENT
LANGUAGES
Language-validation studies with NH listeners were conducted in separate independent experiments for the language-dependent tests (see Van Esch et al, 2013). The results were used to correct for test-material effects by presenting all outcome measures relative to reference values, based on the average scores of NH listeners for each language. The language-dependent tests comprise the spatial and binaural SRT benefit tests (see also Wagener et al, 2006) and the Gothenburg profile. Results of these three tests showed significant differences between centres, even though the spatial and binaural SRT benefit results are difference scores (see van Esch et al, 2013 for uncorrected results and more information on the differences between centres). In the present paper, corrected data (i.e. data after subtraction of reference values) are presented for the spatial SRT tests and the Gothenburg profile.
NORMALITY The normality of the outcome measures was tested in Van Esch et al (2013) by visual inspection and the Shapiro-Wilk and Kolmogorov-Smirnov tests. In the HI group, all outcome measures except the MAA results were distributed (approximately) normally. Log-transformed MAA data were distributed normally and are used in the present analyses.
LINEAR
REGRESSION AND INCLUSION OF AUDIOGRAM THRESHOLDS
Multiple linear regressions were performed in SPSS on the data from the group of HI listeners. Measures of localization behaviour were used as dependent variables. The models involved stepwise inclusion of possible predictors (inclusion: p ⬍ 0.05, and exclusion: p ⬎ 0.10). As we wanted to investigate the relevance of the preliminary auditory profile tests in addition to the pure-tone audiogram and age, we first included the audiogram measures and age in one block, before including all other measures in a second block. For every analysis, the distribution of residuals was checked for approximate normality, a plot of residuals versus predicted values was checked for linearity and homoscedasticity, and autocorrelation of the residuals was tested using the Durbin-Watson statistic (Durbin & Watson, 1971).
Results The results of each individual test are presented in Van Esch et al (2013). Here we report the correspondence between the psychophysical spatial hearing tests and self-reported spatial-hearing abilities. Table 1 reports the Pearson correlation coefficients and significances between the scores related to spatial hearing from the Gothenburg Profile and the MAA and binaural SRT results. Scatter plots of the same data are presented in Appendix A. As a reference, correlations with listener age and audiogram results are also displayed. Because of the binaural nature of the tests, we chose to use binaural data from the audiogram also: the pure-tone average (0.5, 1, 2, 4 kHz, PTA) and the slope of the audiogram (difference between
186
T.E.M. Van Esch et al. Table 1. Pearson correlation coefficients in the HI group between the corrected Gothenburg profile scores (average localization category score, GPloc, and scores for questions 6 and 7, Q6 and Q7), log-transformed MAA results for broadband (bb), low pass (lp), and high pass (hp) stimuli, left and right spatial and binaural SRT benefit, age, and hearing loss (PTA, slope and asymmetry). Significant correlations at the p ⬍ 0.01 and p ⬍ 0.05 level are marked ** and * respectively. Spatial SRT benefit
MAA
GPloc Q6 Q7
Binaural SRT benefit
Audiogram
bb
lp
hp
left
right
left
right
Age
PTA
slope
asym
0.31* 0.38** 0.29*
0.45** 0.46** 0.21
0.07 0.14 ⫺0.01
0.34** 0.24* 0.48**
0.43** 0.28* 0.54**
0.30* 0.29* 0.39**
⫺0.01 ⫺0.08 0.09
⫺0.19 ⫺0.14 ⫺0.10
0.60** 0.45** 0.59**
⫺0.12 ⫺0.18 ⫺0.16
0.09 0.01 0.18
thresholds at 0.5 and 4 kHz), both averaged over both ears. Moreover, since asymmetry of hearing loss is expected to be relevant for localization abilities, the absolute difference between left and right ear PTAs was also included in the analyses. It can be seen from Table 1 that the correlations between the PTAs and GP results were highly significant (r ⱖ 0.45, p ⬍ 0.01, see also Supplementary Appendix Figure 1 available online at http.// informahealthcare.com/doi/abs/10.3109/14992027.2014.953216). The asymmetry in hearing thresholds, slope of the audiogram, and age did not show significant correlations with self-reported localization abilities. For the MAA test, most correlations with the GP results were significant (see also Supplementary Appendix Figure 2 available online at http.//informahealthcare.com/doi/abs/10.3109 /14992027.2014.953216). The strongest correlations were found between MAAs with broadband and low-pass noises and question #6 (about localizing traffic sounds) and the localization category from the GP (r ⫽ 0.38; r ⫽ 0.45). A significant correlation was found between the MAA measured with broadband noise and question #7 (about turning your head in the direction of someone calling you; r ⫽ 0.29; p ⬍ 0.05). For the high-pass noise condition, no significant correlations with the GP results were found. Significant correlations between the spatial SRT benefit test and the GP results were also found (see also Supplementary Appendix Figure 3 available online at http.//informahealthcare.com/doi/abs/10.3109/14992027.2014. 953216): both noise-left and noise-right measurements correlated significantly with the results from the localization category and question #7 (r ⫽ 0.48; r ⫽ 0.54; both p ⬍ 0.01) and with the results of question #6 (r ⫽ 0.24; r ⫽ 0.28; p ⬍ 0.05). Nevertheless, these correlations were lower than those between PTA and GP. For the binaural SRT benefit test, significant correlations with GP results were found for noise-left measurements (r ⱖ 0.29; p ⬍ 0.05), while the correlations were negligible for noise-right measurements2 (see also Supplementary Appendix Figure 4 available online at http.// informahealthcare.com/doi/abs/10.3109/14992027.2014.953216). Three linear regression analyses were performed in two blocks to find the best predicting variables for the localization category and for questions #6 and #7 of the Gothenburg profile (GPloc, Q6, and Q7). In the first block, the audiogram variables (PTA, asymmetry, and slope) and age were included by a stepwise procedure to find the significant predictors from these measures. Next, in a second block, the other possible predictors (MAA and SRT benefit measures) were entered, also by a stepwise procedure. The results of these analyses are reported in Table 2. For the prediction of GPloc, only results from the models with age and audiogram variables (block 1) are shown, since none of the variables from
the second block proved to contribute significantly to the model. For Q6 and Q7 on the other hand, both results from the model with block 1, and from the model with blocks 1 and 2 are presented. For each model the included variables, values of R and adjusted R2 (corrected for the available degrees of freedom) are shown. The regression formulas for the three outcome measures were: Q6 ⫽ 0.208 ⫹ (0.049 ⫻ PTA) ⫹ (1.894 ⫻ low-pass MAA) Q7 ⫽ 2.724 ⫹ (0.069 ⫻ PTA) ⫹ (0.334 ⫻ spatial SRT benefit (right)) GPloc ⫽ 0.136 ⫹ (0.091 ⫻ PTA) ⫺ (0.033 ⫻ age) These models explained 24%, 40%, and 39% of the variance in Q6, Q7, and GPloc data respectively (adjusted R2 values). All models have a significance of p ⬍ 0.001. Note that the regression equations predict the corrected GP scores in HI listeners (i.e. the calculated scores for HI listeners are relative to the reference scores, so negative scores refer to better than reference), to account for differences between results from different languages. Scatter plots of predicted versus measured GP scores are shown in Figure 2. For all three outcome measures, higher average hearing losses (PTA) were associated with an increase in perceived problems, at a rate of 0.5–0.9 points increment (on a 10-point scale) per 10 dB PTA. Adding the low-pass MAA to the equation significantly improved the prediction of question #6 (about localizing traffic sounds) by 5% (increment of adjusted R2, p ⬍ 0.05). The prediction of question #7 (about turning your head in the direction of the speaker) improved significantly by 6% (increment of adjusted R2, p ⬍ 0.05) by adding spatial SRT benefit (right side) to the equation. Both for questions Table 2. Results of stepwise linear regression analyses of GP data. For three dependents (Q6, Q7, and GPloc) significant predictors of the stepwise linear regression models are shown, as well as R and adjusted R2 values of these models. All shown models have a significance of p ⬍ 0.001. For Q6 and GPloc, none of the significant predictors from the second block proved to be significant. Dependent Q6 Q7
GPloc
Model
Predictors
R
Adjusted R2
block 1 block 1& 2 block 1 block 1 & 2
PTA PTA, MAA lp PTA PTA, spatial SRT benefit (right) PTA, age
0.45 0.52 0.59 0.65
0.19 0.24 0.34 0.40
0.65
0.39
blocks 1
Psychophysical and self-reported measures of spatial hearing GP localization
Question #6
8
4 2 0 –2
6 4 2 0 –2
–2
0
2
4
6
Measured scores
8
Question #7
8 Predicted scores
6
Predicted scores
Predicted scores
8
187
6 4 2 0 –2
–2
0
2
4
6
Measured scores
8
–2
0
2
4
6
8
Measured scores
Figure 2. Predicted (vertical axes) versus measured (horizontal axes) corrected Gothenburg profile scores for the sound localization category (left panel), question #6 (middle panel), and question #7 (right panel).
#6 and #7, poorer psychophysical scores correspond to more selfreported problems. For the localization category, age proved to be a significant predictor besides PTA, accounting for a 5% increase in adjusted R2 (p ⬍ 0.05). On average, after having accounted for their hearing loss, older listeners reported less problems. Note that average hearing loss proved to be the only significant audiogram predictor for all evaluated GP measures; slope and asymmetry did not add any predictive power. Like in the correlation analysis, asymmetry of hearing loss did not play a significant role. Moreover, the binaural SRT benefit results did not make a significant contribution to the regression models.
Discussion The aim of the present study was to investigate to what extent the psychophysical measures of spatial hearing of the preliminary auditory profile (Van Esch et al, 2013) are indicative of self-reported functioning as measured by the Gothenburg profile (Arlinger et al, 1998; Ringdahl et al, 1998). To that end, we examined the correspondence between the Gothenburg profile localization scores and the MAA and the spatial and binaural SRT benefit results, both by correlation and linear regression analyses. When interpreting these analyses, it has to be recognized that the correspondence between objective and self-reported localization abilities depends on both validity of the psychophysical tests and validity of the questionnaire. The Swedish version of the GP has been validated in a large group of HI listeners (see Ringdahl et al, 1998). A potential drawback of the present GP data is that listeners were instructed to fill out the questionnaire for the unaided situation (as they were also tested without hearing aids), which might have been hard for some of the hearing-aid users. Additionally, although no significant test centre effect was found in the localization category of the GP (Van Esch et al, 2013), non-significant centre effects might have contaminated the data. Regarding the psychophysical tests, the use of generic HRTFs might have decreased the reliability as they do not guarantee an equal applicability for all listeners (see Middlebrooks, 1999a and 1999b, for a comparison of virtual localization with own-ear versus other-ear HRTFs). For the MAA measurements, several significant correlations, ranging from 0.29 to 0.45 (p ⬍ 0.01 and p ⬍ 0.05, see Table 1) were found for the broadband and (mostly) for the low-pass noise conditions, but not for the high-pass noise condition. This correspondence between MAA results and GP localization measures has not been demonstrated before and was not anticipated, as the MAA test is a
spatial discrimination measure rather than a measure of location perception. Moreover, the stepwise linear regression analysis showed that adding the low-pass MAA results to the regression model, after inclusion of PTA, increased the explained variance of the prediction of question #6 by 5% (p ⬍ 0.05). This indicates that the MAA test has potential added value in predicting the self-reported ability to localize different sounds in traffic. In other words: the results of the stepwise linear regression analysis show that the correspondence between MAA and GP results is not solely a consequence of mutual correlations with the audiogram. Although MAA has been measured in many previous studies (see e.g. Grantham et al, 2003; Hausler et al, 1983), its relationship to self-reported performance has not been described. So, despite the limitations regarding clinical applicability3 of the MAA test reported by Van Esch et al (2013), the MAA test does have some predictive power for self-reported localization performance as measured by the GP. Many spatial SRT experiments have been described (e.g. Goverts & Houtgast, 2010; Johansson & Arlinger, 2002; Wilson et al, 1982), but to our knowledge, the relationship with self-reported spatial hearing abilities has not been investigated before. Our spatial SRT benefit results showed significant correlations, ranging from 0.24 to 0.54, for both noise-left and noise-right measurements, for question #6 (p ⬍ 0.05), and more strongly for question #7 and the GP localization category (p ⬍ 0.01, see Table 1). It is interesting to see that the highest correlations (0.48 and 0.54 for left and right, both p ⬍ 0.01) were found between the spatial SRT benefit test and question #7, which asked about head turning in the direction of a speaker. Compared to correlations between GP and PTA however, correlations with spatial SRT benefit were of the same order of magnitude and significance, or lower. This could indicate that at least a substantial part of these correlations was caused by underlying correlations with PTA. Nonetheless, the stepwise linear regression analysis showed that adding spatial SRT benefit to the regression model (after inclusion of PTA) improved the prediction of question #7 results by a 6% increment of adjusted R2 (p ⬍ 0.05). So, the spatial SRT benefit test has some added value in predicting self-reported binaural functioning (as measured by the GP), especially for listening to speech in spatial situations. To our best knowledge, the relevance of this test for self-reported spatial hearing performance has not been evaluated before. Considering the binaural SRT benefit test, the correlations between GP and the noise-left measurement (i.e. the measurement with noise from the left side) were between 0.29 and 0.39 (p ⬍ 0.05). For the noise-right measurement on the other hand the correlations were
188
T.E.M. Van Esch et al.
very low and even negative (ranging from ⫺ 0.08 to 0.09)4. This is confirmed by the scatter plots in Supplementary Figure 4 (Appendix A available online at http.//informahealthcare.com/doi/ abs/10.3109/14992027.2014.953216). Linear regression analyses nevertheless showed that results from the binaural SRT benefit test did not contribute significantly to the prediction of GP data. As a result, the binaural SRT benefit test does not appear to be an indicative measure of spatial hearing, when related to self-report on the GP questionnaire. Another finding from the linear regression analyses is that age proved to be a significant predictor for the GP localization category, accounting for 5% of explained variance (adjusted R2, p ⬍ 0.05). We found that older listeners reported (on average, after having accounted for their hearing loss) less problems. A hypothetical reason for this result is that it is more common among older listeners to have a hearing loss and that older listeners therefore report less problems due to a better acceptance of their disabilities. Lutman (1991) and Gatehouse (1991) described similar effects of the expectations of older listeners on their self-reported disability. Regarding the pure-tone audiogram, correlation and regression analyses showed that in the present data set only the pure-tone average was predictive of self-reported localization performance, accounting for 34%, 19%, and 34% of explained variance for GPloc, Q6, and Q7 respectively (adjusted R2 values). We repeated the regression analyses with a binaural weighted average of the audiogram (4:1 weighting of better:worse, see e.g. Lutman et al, 1987) and the results were almost identical. Slope of the audiogram, and asymmetry of hearing loss did not show any significant correlations with, or predictive power for, self-reported localization performance. However, the present study was not designed to examine the influence of asymmetry on spatial hearing, and due to the small number of listeners with asymmetric hearing loss (13 listeners), a significant correlation between asymmetry and self-reported localization in the full HI group was not expected in the present data. In general, the percentages of variance in GP results that could be explained by the psychophysical measures including the pure-tone audiogram were below 50% (see Table 2). These numbers have to be interpreted relative to the test-retest reliability of the GP. As the correlation between test and retest results of the localization category was 0.92, the percentage of explained variance from the regression models could not meaningfully exceed 84% (0.922) if the predicting variables were exactly accurate. Due to uncertainty in the predicting variables, the percentage of explained variance will be even lower. Nevertheless, the values we found may still seem low. There are important differences between virtual laboratory free-field measurements of spatial hearing compared with real life experiences that probably have played a role. Clearly, real-life localization involves much more than the detection of sounds (as measured with the pure-tone audiogram) and pure binaural hearing (as measured with the psychophysical tests). In the outside world, spatial hearing is influenced by level differences, reflections, background noise, and dynamic situations. The importance of the dynamic aspects of real life listening situations was argued to be important by Gatehouse & Noble (2004) and Noble & Gatehouse (2004). Gatehouse & Akeroyd (2006) suggested that a more dynamic localization test would have shown stronger correspondence to self-reported functioning. None of the above-mentioned aspects of real-life localization are captured by the auditory profile spatial hearing tests. Moreover, in the auditory profile spatial hearing tests averaged HRTFs were used, which do not exactly match the real world situation for individual listeners. Therefore, and also because of the measurement inaccuracy of each
measure, a considerable proportion of unexplained variance in the GP data was to be expected.
Conclusions We conclude that the MAA test and the spatial SRT benefit test are indicative measures for self-reported binaural functioning. The MAA test best represents self-reported localization, while the SRT benefit test was found to be the most important for self-reported listening to speech in spatial situations. The binaural SRT benefit test was not found to be relevant for self-reported spatial hearing performance.
Acknowledgements Data from the present study have been published before by Van Esch et al (2013). Materials of several of the tests from the preliminary auditory profile are available from the HearCom website (www. hearcom.eu): Matrix sentence frameworks in the four languages, and the Gothenburg profile questionnaire in the four languages. We thank Daniel Berg for technical support and implementation of the tests on the platform (OMA), and Kirsten Wagener for her work on consolidation of the speech tests. Finally we thank the test subjects for their participation. The study was supported by grants from the European Union FP6, Project 0004171 HEARCOM. The information in this document is provided as is, and no guarantee or warranty is given that the information is fit for any particular purpose. The user thereof uses the information at his/her sole risk and liability. The authors alone are responsible for the content and writing of the paper. Parts of this work have been presented at the International Hearing Aid Research Conference, August 16, 2008, Lake Tahoe, California: Evaluation of the ‘Auditory Profile’ test battery in an international multi-centre study.
Notes 1. NL-AMC: no. 05/127 # 05.17.0934, dated August 3, 2005; HZO-DE: “Klinische Tests zur Bestimmung individueller Hördefizite und Kommuniationsfähigkeiten”, dated November 15, 2006; ISVR-UK: 791, dated February 13, 2007; SE-LINK: M83-06; VUMC-NL: MEC05/12 - 2006/171, dated November 2, 2006. 2. Spearman’s rank correlations (which are less sensitive to outliers) were calculated to investigate whether the significant correlation for noise-left measurements were influenced by outliers. We found that Spearman’s correlations between GP and BILD were very similar to, and of the same significance as, Pearson’s rank correlations: 0,314 (p ⬍ 0.01), 0,284 (p ⬍ 0.05) and 0,419 (p ⬍ 0.01) for the localization subscale and questions #6 and #7 respectively. This indicates that the influence of outliers on the correlation values was negligible. Also for the other variables (PTA, MAA, and spatial SRT benefit) Pearson’s correlations and Spearman’s rank correlations were nearly identical (data not shown). 3. Van Esch et al (2013) found that the low-pass noise condition of the MAA test was too hard for a considerable number of listeners. Additionally, they reported a significant learning effect for the MAA test. 4. Van Esch et al (2013) found in the same data set that listeners perceived significantly more binaural benefit with noise from the left side (after correction for their ear-specific hearing thresholds),
Psychophysical and self-reported measures of spatial hearing which agrees with the generally known right-ear advantage for speech perception (see Tervaniemi & Hugdahl, 2003, for a review). The present analyses suggest that the binaural benefit with noise from the left side was also more closely related to subjective binaural functioning. If both findings can be replicated in other studies, it implies that the right-ear advantage is related to everyday binaural functioning.
Declaration of interest: The authors report no conflicts of interest.
References Arlinger S., Billermark E., Oberg M., Lunner T. & Hellgren J. 1998. Clinical trial of a digital hearing aid. Scand Audiol, 27, 51–61. Athalye S.P. 2010. Factors affecting speech recognition in noise and hearing loss in adults with a wide variety of auditory capabilities. PhD thesis, Institute of Sound and Vibration Research, Southampton, UK. Brand T. & Hohmann H. 2002. An adaptive procedure for categorical loudness scaling. J Acoust Soc Am, 112, 1597–1604. Brand T. & Kollmeier B. 2002. Efficient adaptive procedures for threshold and concurrent slope estimates for psychophysics and speech intelligibility tests. J Acoust Soc Am, 111, 2801–2810. Durbin J. & Watson G.S. 1971. Testing for serial correlation in least squares regression. III. Biometrika, 58, 1–19. Flamme G.A. 2001. Examination of the validity of auditory traits and tests. Trends Amplif, 5, 111–138. Gatehouse S. 1991. The role of non-auditory factors in measured and selfreported disability. Acta Oto-Laryngol, 249–256. Gatehouse S. & Akeroyd M. 2006. Two-eared listening in dynamic situations. Int J Audiol, 45, S120–S124. Gatehouse S. & Noble W. 2004. The speech, spatial and qualities of hearing scale (SSQ). Int J Audiol, 43, 85–99. Goverts S.T. & Houtgast T. 2010. The binaural intelligibility level difference in hearing-impaired listeners: The role of supra-threshold deficits. J Acoust Soc Am, 127, 3073–3084. Grantham D.W., Hornsby B.W. & Erpenbeck E.A. 2003. Auditory spatial resolution in horizontal, vertical, and diagonal planes. J Acoust Soc Am, 114, 1009–1022. Hafter E.R., Saberi K., Jensen E.R. & Briolle F. 1992. Localization in an echoic environment. Adv Biosci, 83, 555–561. Hagerman B. 1982. Sentences for testing speech intelligibility in noise. Scand Audiol, 11, 79–87. Hausler R., Colburn S. & Marr E. 1983. Sound localization in subjects with impaired hearing. Spatial-discrimination and interaural-discrimination tests. Acta Otolaryngol, Suppl. 400, 1–62. Johansson M.S. & Arlinger S.D. 2002. Binaural masking level difference for speech signals in noise. Int J Audiol, 41, 279–284. Koopman J., Houben R., Dreschler W. & Verschuure J. 2006. Development of a speech in noise test (Matrix). 8th EFAS Congress; 10th DGA Congress; 6–9 June 2006, Heidelberg, Germany.
Supplementary material available online Supplementary Appendix Figures 1–4.
189
Kramer S.E., Kapteyn T.S., Festen J.M. & Tobi H. 1996. The relationships between self-reported hearing disability and measures of auditory disability. Audiology 35, 277–287. Kramer S.E., Kapteyn T.S. & Festen J.M. 1998. The self-reported handicapping effect of hearing disabilities. Audiology, 37, 302–312. Lutman M.E. 1991. Hearing disability in the elderly. Acta Oto-Laryngol, 239–248. Lutman M.E., Brown E.J. & Coles R.R.A. 1987. Self-reported disability and handicap in the population in relation to pure-tone threshold, age, sex, and type of hearing loss. Brit J Audiol, 21, 45–58. Middlebrooks J.C. 1999a. Individual differences in external-ear transfer functions reduced by scaling in frequency. J Acoust Soc Am, 106, 1480–1492. Middlebrooks J.C. 1999b. Virtual localization improved by scaling nonindividualized external-ear transfer functions in frequency. J Acoust Soc Am, 106, 1493–1510. Mills A.W. 1958. On the minimum audible angle. J Acoust Soc Am, 30, 237–246. Noble W., Ter-Horst K. & Byrne D. 1995. Disabilities and handicaps associated with impaired auditory localization. J Am Acad Audiol, 6, 129–140. Noble W. & Gatehouse S. 2004. Interaural asymmetry of hearing loss, Speech, Spatial and Qualities of Hearing Scale (SSQ) disabilities, and handicap. Int J Audiol, 43, 100–114. Ringdahl A., Eriksson-Mangold M. & Andersson G. 1998. Psychometric evaluation of the Gothenburg Profile for measurement of experienced hearing disability and handicap: applications with new hearing-aid candidates and experienced hearing-aid users. Br J Audiol, 32, 375–385. Silzle A. 2007. Generation of quality taxonomies for auditory virtual environments by means of systematic expert survey. PhD thesis, Ruhr-Universität Bochum, Germany. Tervaniemi M. & Hugdahl K. 2003. Lateralization of auditory-cortex functions. Brain Res Rev, 43, 231–246. Tyler R.S., Perreau A.E. & Ji H. 2009. Validation of the spatial hearing questionnaire. Ear Hear, 30, 466–474. Van Esch T.E., Kollmeier B., Vormann M., Lyzenga J., Houtgast T. et al. 2013. Evaluation of the preliminary auditory profile test battery in an international multi-centre study. Int J Audiol, 52, 305–321. Villchur E. 1974. Simulation of the effect of recruitment on loudness relationships in speech. J Acoust Soc Am, 56, 1601–1611. Wagener K., Brand T. & Kollmeier B. 1999a. Development and evaluation of a German sentence test part I: Design of the Oldenburg sentence test. Z Audiol, 38, 4–15. Wagener K., Brand T. & Kollmeier B. 1999b. Development and evaluation of a German sentence test part II: Optimization of the Oldenburg sentence test. Z Audiol, 38, 44–56. Wagener K., Brand T. & Kollmeier B. 1999c. Development and evaluation of a German sentence test part III: Evaluation of the Oldenburg sentence test. Z Audiol, 38, 86–95. Wagener K., Brand T. & Kollmeier B. 2006. International cross-validation of sentence intelligibility tests. 8th EFAS Congress; 10th DGA Congress, 6–9 June 2006, Heidelberg, Germany. Wilson R.H., Hopkins J.L., Mance C.M. & Novak R.E. 1982. Detection and recognition masking-level differences for the individual CID W-1 spondaic words. J Speech Hear Res, 25, 235–242.
Copyright of International Journal of Audiology is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
