Download Application of the online hearing screening test Earcheck: Online

 Application of the online hearing screening test Earcheck: Online speech‐in‐noise testing and SRT performance among teenage students. April 2016 Commissioned by: Date Version Status Authors April 2016 1.0 Final version Marya Sheikh Rashid, MSc, AMC Dr. Irene Jacobi, AMC Dr.ir. Jan A.P.M. de Laat, LUMC Prof.dr.ir. Wouter A. Dreschler, AMC Acknowledgements De Nationale Hoorstichting Zicht Online Key words Earcheck, teenagers, speech‐in‐noise, screening test Remarks An evaluation of Earcheck in teenage students. 1 TABLE OF CONTENTS
1. Introduction ........................................................................................................................................ 3 2. Study objectives ................................................................................................................................. 4 3. Methodology ...................................................................................................................................... 5 4. 3.1. Description of the test ................................................................................................................ 5 3.2. Study design ............................................................................................................................... 5 3.3. Study population ........................................................................................................................ 6 3.4. Sample size and recruitment ...................................................................................................... 6 3.5. Study procedure ......................................................................................................................... 7 3.6. Ambient noise level measurements at the test rooms .............................................................. 7 3.7. Equipment .................................................................................................................................. 7 3.8. Data analysis ............................................................................................................................... 8 Results ................................................................................................................................................ 9 4.1. Subjects ...................................................................................................................................... 9 4.2. Categorized test and retest results ............................................................................................ 9 4.3. Test‐ and retest SRT and overall learning effect ...................................................................... 10 4.4 Mean SRT (test and retest) and effect of age (gender and education) .................................... 11 4.6 New categorization of Earcheck results ................................................................................... 13 4.7 Age‐corrected outcome and conditional retest ....................................................................... 14 5. Discussion ......................................................................................................................................... 15 6. Main findings of this study ............................................................................................................... 15 7. Implications of an age‐dependent cut‐off values ............................................................................ 16 8. Conclusion ........................................................................................................................................ 18 References ................................................................................................................................................ 19 2 1. Introduction Teenagers and young adults are frequently exposed to noise in their daily life, during leisure time (the use of mp3‐players, smartphones and tablets, visits to concerts, discos, cafes, cinemas), and to environmental noise, gaming, sports, etc. [Torre 2008, Zhao 2010, Portnuff 2011, Serra 2005]. It comes as no surprise if teenagers and young adults have an increased risk for noise‐induced hearing loss (NIHL). However, according to the literature, there is no consensus concerning the effect of leisure noise on hearing loss [Carter, 2010]. Some studies show an increase in NIHL due to the exposure of leisure noise [Meyer‐Bisch 1996, Niskar 2001, Shargorodsky 2010]. Other studies, however, did not find evidence for this relationship [Henderson 2011, Mostafapour 1998, Zhao 2010, Jin 2013]. NIHL usually affects intelligibility for speech in noise. Ideally, a speech‐in‐noise test, instead of pure‐
tone audiometry, can be used to measure this specific hearing ability. A speech‐in‐noise test is a valuable test, which can be used easily for online hearing screening, reaching the general public in a simple and effective way [Smits 2005, Smits 2013, Jansen 2013, Leensen 2013]. Earcheck (Dutch: Oorcheck), is a Dutch online speech‐in‐noise test, specifically designed and tested for the detection of high‐frequency hearing losses, including NIHL. The target group of Earcheck are teenagers and young adults from the age of 12 to 24 years. The test reaches large numbers (about 20,000 participants a year), and gives an insight in SRT performance. In addition, the test may create more awareness concerning exposure to leisure noise and hearing loss in this specific risk group. In 2015, the performance of teenagers and young adults on the Earcheck online speech‐in‐noise test was analyzed, comprising data of the preceding five years [Sheikh Rashid, Report Online Hearing Tests 2010‐2014, Nationale Hoorstichting, 2015]. The main objective was to get insight into their speech discrimination abilities based on the internet survey data. An interesting finding was that SRT scores tend to decrease (improve) with age for adolescents until around the age of 20 years. This finding may be explained by developmental changes in adolescence (maturation of speech understanding in noise, see also Elliott, 1979). However to confirm this explanation, additional research is required. 3 2. Study objectives The aim of this study is to investigate speech discrimination abilities of teenagers, by the use of the online speech‐in‐noise test, Earcheck. The main objective is to explore the relation of SRT scores with age, gender, and level of education. Earcheck will be evaluated among normal‐hearing secondary school‐aged teenagers, with the participants of 18 years and older as a reference group. The following research questions will be answered: 1.
Do we find an age‐dependent effect of Earcheck results in normal‐hearing teenage students aged 12‐18 years? 2.
Is there an effect of repeated testing? 3.
Does the level of education of the 12‐18 year olds affect the outcome? 4.
Are the effects of age, level of education, and learning large enough to introduce correction factors on the original cut‐off value for fail or pass, and if so, how can this be done? 4 3. Methodology 3.1.
Description of the test Earcheck (link: www.oorcheck.nl) is specifically designed for adolescents aged 12 to 24 years, a population frequently exposed to recreational noise. Earcheck is an objective functional online speech‐
in‐noise test, giving insight in the ability to recognize speech in noise, a hearing ability of great importance in everyday communication. The speech‐in‐noise test serves as an appropriate screening test, as it has shown to be sensitive for high‐frequency hearing loss. Thus the test is especially applicable for high‐frequency hearing loss, e.g. noise‐induced hearing loss. Earcheck uses nine different monosyllabic Dutch words, randomly presented in a low‐pass filtered interfering noise. On screen, nine response buttons are shown. A tenth button, saying ‘not recognized’ is given to reduce guessing of the respondents. To allow users to become acquainted with the words, a preliminary sequence of the words is presented preceding testing. After the presentation of a word, the user’s task is to identify the word that was presented by clicking on the corresponding button on the screen. The level of the noise is fixed and the level of the presented words varies according to an up‐down procedure, with a step size of 2 dB. This is based on the testing procedure according to Plomp & Mimpen (Plomp & Mimpen, 1979), except for the first stimulus of Earcheck that is presented only once at a fixed signal‐to‐noise ratio level of ‐10 dB. A list of 27 presented words is used to estimate the signal‐to‐noise ratio (SNR) at which 50% of the word material is correctly reproduced. This is defined as the speech reception threshold (SRT), and is calculated by taking the arithmetic average of the SNRs of the last 20 presentation levels. Although test results can be qualified in three categories: good (green), insufficient (orange), and poor (red), we decided to present the test results to the user as either good (green) or bad (orange and red). The cut‐off value for a good Earcheck score in a well‐controlled setting was ‐18.4 dB SRT (Leensen et al.,2011b). The categories insufficient and poor (second cut‐off value of ‐12.7 dB SRT), are distinguished by a different shade of red in the data register. This information is not provided to the user, but is used for the eventual recommendation given to the user. For research purposes, the exact SRT values are available. At the start of a test session, a word without noise is presented repeatedly, and users can use their PC volume control or a slider on screen to adjust the volume to a comfortable level . The presentation level selected is used for the presentation of all consecutive test words. The test signal is presented to both ears (binaural presentation in free‐filed or diotic presentation with headphones). The use of headphones is recommended in order to obtain a more reliable test result. Before the test starts, some information about the user is collected (for statistical reasons), including age, gender, and subjective rating of one’s hearing in general (a categorical scale is used with the following options: good, less good, or bad). The test can be performed in less than five minutes, including introduction and instructions, test result, and recommendations. 3.2.
Study design The study protocol was approved by the medical ethics committee of the University of Amsterdam (identification code 2015_297). To fulfil the purpose of this study, we intended to objectify the hearing ability of teenagers by hearing tests, following a cross‐sectional study design. Data were collected in February 2016. Quantitative data regarding speech reception in noise were obtained by 5 the participants’ responses to the online hearing test (Earcheck), performed in a quiet room at school. The Earcheck was performed twice (test and retest). In addition, pure‐tone audiometry tests were also performed at the schools as a reference standard. Information on age, gender, and education level was collected. The aim was to include normal‐hearing young adults as a control group. The adults underwent the same tests. 3.3.
Study population Subjects involved in this study were normal‐hearing (NH) teenage students, recruited from a lower and a higher secondary school: Zandvliet College at Den Haag, and Haarlem College at Haarlem. A control group of normal‐hearing young adults were included as well, recruited from Avans Hogeschool at Breda. In order to be eligible to participate in this study, a subject had to meet all of the following criteria: -
Age 12 to 18 years Normal hearing (Hearing thresholds of 25 dB or better at 250 and 500 Hz, and hearing thresholds of 20 dB or better at 1, 2, 3, 4, and 6 kHz) Secondary school student Native speaker of the Dutch language In order to be eligible to participate in this study as a control, the same inclusion criteria were used, except that the age range was between 18 and 25 years and the level of education was college or university student. Subjects with language or hearing problems were excluded. Furthermore, subjects with missing or unreliable Earcheck measurements (intra‐test standard deviations greater than 3 dB) were excluded as well. 3.4.
Sample size and recruitment For this research we wanted to explore the relationship of SRT with age, gender, and level of education, by means of a multiple regression analysis. As we had three predictors, we needed at least 30 to 45 students (according to the rule‐of‐thumb to include 10 to 15 participants per predictor). Because we’re interested in small effects, we included at least 70 normal‐hearing teenage students aged 12‐18 years (about 10 students per age‐category), with different education levels. In order to detect a relevant difference of 3 dB in mean SRT scores between teenagers and young adults, at least 9 participants are needed per category (Power analysis using Enquiry Advisor: SD=2 dB, alpha: 0.05, power:80%, 2‐sided t‐test). Therefore 10 young adults had to be included in the control group. Participants were recruited by contacting the schools, and receiving consent from the school management. A letter was sent, containing information regarding the purpose and procedure of the study, the online hearing test, and the audiogram. During the study, invited subjects were always allowed to contact the researcher by phone or by e‐mail. 6 3.5.
Study procedure Tests were performed at the schools, where two quiet rooms were organized. Before testing, the sound levels of the rooms were inspected. Audiometry was performed by two trained test operators. Earcheck was performed individually, with minimal instructions of the researchers. An audiogram (air‐conduction, at the frequencies 250, 500, 1000, 2000, 3000, 4000, 6000, and 8000 Hz) was taken to confirm whether the teenagers are normal‐hearing. The audiogram was performed in a quiet room at the school and took approximately 10 minutes. In case of anomalies the participant received individual advice by a hearing professional. The Earcheck tests (test and retest) were performed in the same, quiet room. Results of the Earcheck (either pass or fail) were directly shown to the participants. Participants were made clear that the results of the test were indicative only, and that the audiogram would give more insight in their hearing status. 3.6.
Ambient noise level measurements at the test rooms Ambient noise level measurements were performed at the test rooms, with the use of a DVM805 digital sound level Meter (Applicable standard: IEC651 type 2). The sound level at the room at Avans Hogeschool, where the control subjects were measured, was 37 dBA. The sound levels at both rooms at Zandvliet college were 35 dBA. The sound level measurements of the rooms at Haarlem College were measured with the use of a sound level meter (B&K 2260) (table 1). The audiometric test conditions of all test locations met the requirements of international standards for hearing screening (i.g. unmasked air conduction starting at 500 Hz; ISO 8253, part I), if sound attenuating caps were used in combination with headphones. Table 1. Ambient sound level measurements at Haarlem College Mid‐frequency tertsband (Hz) Maximal ambient noise level according to ISO 8253 (part I) Lmax (dB SPL) Mean attenuation TDH39 (Amplivox cups) 125 51 9 Maximal
permissible level using audiocups 60 Room 1 Room 2 LZeq LZF max LZeq LZF max 30,9 42,8 27,0 42,4 250 37 13 50 29,8 38,7 25,7 36,4 500 18 24 42 26,1 36,9 19,3 37,4 1000 23 30 53 21,9 29,9 18,4 32,1 2000 30 39 69 16,2 27,3 15,2 30,7 4000 36 44 80 15,6 26,9 14,9 25,1 8000 33 35 68 17,0 23,3 16,8 18,6 3.7.
Equipment Audiometry was performed by two trained test operators, using calibrated clinical audiometers (AC40 and Decos audioNigma), in combination with TDH 39 headphones with sound attenuating caps (Amplivox audiocups). For the Earcheck measurements, a research laptop (HP) and a tablet (Surface) were used, with Sennheiser HDA 200 and Sennheiser HD330 headphones. The control group of students performed Earcheck on their own mobile phone, in combination with Sennheiser HD330 headphones. 7 3.8.
Data analysis Data were analyzed using IBM SPSS Statistics 22. Descriptive statistics were applied on hearing thresholds, age, gender, level of education, and Earcheck test results, in categories and SRT scores. SRT results in dB were measured on a continuous scale, and are described as means and corresponding standard deviations. The relation between mean SRT scores for the first test and the retest were assessed by means of a Pearson’s r correlation. A paired t‐test and repeated measures analysis were conducted on the test‐retest SRT’s. To explore SRT score as a function of age (in years), gender (male/female), and level of education (low/high), multiple regression analyses were performed. To compare SRT scores of teenagers and young adults, Mann‐Whitney‐U‐tests were performed. 8 4. Results 4.1.
Subjects A total of 104 subjects participated. Twenty‐three (22%) were excluded from analyses based on their audiogram (hearing thresholds greater than 25 dB at 250 or 500 Hz, or hearing thresholds greater than 20 dB at 1, 2, 3, 4, or 6 kHz): Seven subjects (7%) had a hearing loss at both ears, seven had a hearing loss at the right ear, and nine subjects (9%) had a loss at the left ear. One of the remaining 81 (78%) normal‐hearing subjects did not perform a retest of the Earcheck and was therefore excluded from analyses. Of the 80 NH subjects with two Earcheck test runs, eight (8%) had an intra‐test standard deviation for the Earcheck of greater than 3 dB. They were excluded as well, leaving 72 subjects (69%) for data analyses. Table 2 shows the mean hearing thresholds PTA 0.5‐1‐2 kHz and PTA 1‐2‐4 kHz of the 72 subjects for each age group and by ear. The group of adults ranged from 18 to 20 years of age (mean 19 years, SD 0.94). In the following, the score of the best ear will be used for all analyses. The means (SD) of PTA 0.5‐1‐2 kHz and PTA 1‐2‐4 kHz of the best ear are also given in table 2 by age group. Table 2. Age group, number of participants, and mean (SD) of PTA0.5/1/2 and PTA1/2/4 in dB HL for right ear, left ear, and better ear. Age group 12 13 14 15 16 17 adults N 10 8 11 17 8 8 10 PTA right ear PTA0.5/1/2 PTA1/2/4 4.50 (4.38) 4.33 (4.81)
7.50 (5.84) 6.35 (4.45)
3.93 (4.43) 1.67 (5.69)
3.23 (4.77) 1.47 (4.81)
4.79 (3.14) 2.70 (3.75)
5.83 (3.14) 1.46 (2.12)
5.83 (5.62) 4.66 (4.11)
PTA left ear PTA0.5/1/2 PTA1/2/4 3.50 (4.26) 3.25 (3.86)
5.83 (4.63) 5.00 (3.65)
3.33 (6.67) 1.74 (6.41)
4.12 (4.53) 2.89 (4.53)
3.33 (5.56) 2.92 (3.42)
6.04 (4.79) 2.71 (2.98)
3.67 (5.82) 2.83 (4.81)
PTA better ear PTA0.5/1/2 PTA1/2/4 2.33 (4.32) 2.25 (4.23)
5.21 (4.67) 4.17 (4.15)
2.12 (5.63) 0.53 (5.76)
2.16 (3.67) 0.64 (4.64)
2.08 (4.78) 1.15 (2.67)
4.38 (3.56) 1.25 (2.36)
3.50 (5.90) 1.63 (4.33)
4.2.
Categorized test and retest results For the 72 subjects who were included for further analyses, table 3 shows the number (%) of subjects who scored “good” versus “insufficient or poor” in their first and second Earcheck respectively. Based on the cut‐off score of ‐18.4 dB, five of the NH subjects had a bad result for both tests. Table 3 . Number (%) of subjects who scored good (bad) on their first (second) Earcheck, N=72 Retest good N (%) Retest bad N (%) Total N (%) First test good N (%)
45 (62%) 7 (10%)
52 (72%)
First test bad N (%)
15 (21%)
5 (7%)
20 (28%)
Total N (%) 60 (83%)
12 (17%)
72 (100%)
9 4.3.
Test‐ and retest SRT and overall learning effect The SRT scores for the first test and the retest are presented in table 4 and figures 1 and 2. Test and retest SRT’s correlated positively (r=0.309, p=0.008). A paired t‐test and repeated measures analysis were conducted on the test‐retest SRT’s. The SRT scores of the first test (mean=‐19.22 , SD=1.38) and the retest (mean=‐19.70, SD=1.48) differed significantly (t=2.421 p=0.018 and F(1,71)=5.863, p=0.018), indicating a systematic improvement and thus an overall learning effect. Repeated measures analysis on the test‐ and retest SRT with age group as a covariate showed a significant main effect of age group (F(1,71)=8.508 p=0.005). Here, the effect of test sequence (test versus retest) did not reach significance (F(1,70) =3.331, p=0.072), and there was no significant interaction effect of age group with the test‐retest SRTs (F(1,70)=2.334 p=0.131). Table 4. Mean SRT (SD) of test and retest and the average of test and retest in dB SNR by age group. Mean SRT (SD) in dB SNR Age group First test Retest Mean of test‐retest 12 ‐17.92 (1.53) ‐19.61 (1.11)
‐18.76 (0.94) 13 ‐19.10 (0.89) ‐19.58 (0.95)
‐19.34 (0.75) ‐19.52 (0.90) 14 ‐19.46 (1.09) ‐19.59 (1.13)
‐19.37 (1.42) 15 ‐19.21 (1.69) ‐19.55 (1.80)
‐19.31 (1.36) 16 ‐19.17 (0.94) ‐19.44 (2.00)
‐19.66 (1.19) 17 ‐19.69 (0.73) ‐19.63 (2.01)
‐20.31 (0.90) adults ‐20.05 (1.34) ‐20.57 (0.96)
‐19.46 (1.16) Total ‐19.22 (1.38) ‐19.70 (1.48)
Figure 1. Mean SRTs and standard deviations of test (black line) and retest (gray line) as a function of age. 10 P75
-16
P25
P50
-17
-18
-19
-20
-21
-22
12
13
14
15
16
17
adults
years
Figure 2. SRT distribution in percentiles (25th, 50th, 75th) of first test (left) and retest (right) by age group.
4.4 Mean SRT (test and retest) and effect of age (gender and education) Since the test sequence did not reach significance, a regression analysis was conducted on the SRT mean of test and retest. The mean SRT improved significantly by 0.218 dB per year of age (95% CI: ‐
0.364; ‐0.072, p=0.004, table 5), while effects of gender or education were insignificant. Mann‐Whitney tests on the age groups (with a Bonferroni correction of α < 0.008) showed that the 12‐
year olds differed significantly from the adults (U=11.5, p=0.002, r=0.65). Table 5. Regression analysis, with SRT (mean of both tests in dB SNR) as dependent variable, and age, gender, and level of education as predictive factors. β p 95% CI ‐18.794
‐14.456
Constant* ‐16.625 .000 ‐.364
‐.072
Age (in years) ‐.218 .004 Gender .355 .467 .366 .479 Level of education ‐.613
1.323
‐.660
1.391
2 The explained variance is R =0.123. 11 4.5 SRT with reference to gender and education In the following, effects of gender and education are described in more detail. Table 6 presents the number of students categorized according to gender and secondary school level (excluding the group of adults). There were 41 female and 21 male students. Thirty‐two of them were students at a higher level secondary school, while 30 attended a secondary school of lower level. In table 7, the mean SRTs (SD) are given by age group according to level of education and gender respectively. Table 6. Gender and level of education by numbers. Gender Female Male Total School level Higher 18 14 32 Lower 23 7 30 Total 41 21 62 Table 7. Age groups, numbers, and mean SRT (SD) by gender and level of education. Age group 12 13 14 15 16 17 N f/m 7/3 3/5 8/3 13/4 4/4 6/2 gender female male ‐18.51 (0.72) ‐19.36 (1.30)
‐19.38 (0.78) ‐19.32 (0.82)
‐19.27 (0.92) ‐19.24 (0.98)
‐19.27 (1.52) ‐19.74 (1.16)
‐19.30 (1.05) ‐19.32 (1.79)
‐19.66 (1.18) ‐19.66 (1.70)
N l/h
4/6
3/5
8/3
12/5
2/6
1/7
level of education lower higher ‐19,01 (1.38) ‐18,60 (0.61)
‐19.20 (0.47) ‐19.42 (0.92)
‐19.74 (0.79) ‐18.95 (1.09)
‐19.50 (1.38) ‐19.09 (1.65)
‐19.16 (1.64) ‐19.35 (1.43)
‐20.09 ‐19.60 (1.27)
In figure 3, regression lines are given for SRT as a function of age: • Overall for all subjects (n=62) • Separate for male (n=21) and female (n=41) participants (left panel) • Separate for students at higher (n=32) and lower (n=30) level secondary schools (right panel) For the calculation of these regression lines the ages (in months) have been used rather than the age categories. 12 Figure 3. Regression lines for mean SRTs as a function of age: All pupils (black line, N=62), females (left panel: dark grey dotted line, N=41), males (left panel: light grey dotted line, N=21), higher (right panel: dark grey dashed line, N=32), and lower (right panel: light grey dashed line, N=30) level secondary schools.
4.6 New categorization of Earcheck results Initially, the cut‐off score for a good Earcheck SRT in a controlled laboratory setting was ‐18.4 dB (table 3, p.9). According to the results of the previous regression analysis on the present study group (table 5), the effect of age on the Earcheck outcome can be predicted by a factor of ‐0.2 dB per year of age. To control for the effect of age, a correction of 0.2 dB per year was applied to the original overall cut‐
off score of ‐18.4 dB, resulting in a cut‐off value of ‐17.2 dB SNR for the 12‐year olds, that decreases to a cut‐off value of ‐18.4 dB SNR for young adults of 18 years and older. In Table 8, the proposed cut‐off scores based on an SRT improvement of 0.2 dB per year of age are presented by age group, next to the SRT distributions in percentiles. As can be seen in table 8, the 75% or 90% percentiles of the subjects tested are at lower scores than the cut‐off values (except for the 15‐year olds), indicating that the vast majority of the SRTs are within the proposed cut‐off value in all age groups. Table 8. SRT distribution of mean of first and second Earcheck in percentiles by age group. To the right, the proposed Earcheck cut‐off SRTs are given based on a correction factor of 0.2 dB per year of age. N=72 Percentiles age‐corrected cut‐off score
Age group 10th 25th
50th
75th
90th
12 ‐20.59 ‐19.33 ‐18.68
‐18.03
‐17.51
‐17.2
13 ‐20.45 ‐20.14
‐19.07
‐18.88
‐17.4
‐20.63 ‐20.14
‐19.64
‐17.88
‐17.6
14 ‐19.05
‐21.41 ‐20.48
‐19.41 ‐16.75
‐17.8
15 ‐18.84
‐20.86 ‐20.23 ‐19.84 ‐18.0
16 ‐18.23 17 ‐20.86 ‐20.56
‐20.07 ‐18.72 ‐18.2
‐21.57 ‐21.02
‐20.32
‐18.98
‐18.4
adults ‐19.47
13 4.7 Age‐corrected outcome and conditional retest Table 11 shows the test and retest Earcheck results according to the new categorization with a correction factor of 0.2 dB per year of age. In figure 4, the SRT distribution of the first Earcheck is plotted after application of the age‐dependent cut‐off scores in percentiles. As can be seen in table 11, 90% scored a ‘good’ Earcheck in the first run after correcting for the effect of age. One of the 72 subjects failed to reach the age‐corrected Earcheck criterion in both test. (The latter concerns a 15‐
year old pupil with a first SRT of ‐14.91 dB SNR and a retest SRT of ‐17.63 dB SNR, where the proposed age‐corrected cut‐off value is ‐17.8 dB.) In a previous Earcheck internet test for the discrimination of speech in noise, the learning effect from test to retest was compensated for by a test procedure that automatically repeated the test if the result was poor (conditional retest). Since we also found a significant learning effect in this study, we recommend to implement the same procedure in future versions of Earcheck. In figure 5, the effects of both improvements have been calculated. The distribution of the deviations from the age‐corrected cut‐off scores are plotted in percentiles: For the 90% who scored a ‘good’ score in the first Earcheck, the first SRT result was taken, and for the 10 % who scored ‘poor or insufficient’ in the first run, the retest SRT was taken. As can be seen in figure 5 and table 11, with this procedure 99% of the subjects obtained a good score. Table 11. Number (%) of subjects who scored good (bad) on their first (second) Earcheck, according to age‐dependent criteria. N=72 First test good N (%) First test bad N (%)
Total N (%) Retest good N (%)
58 (81%) 6 (8%)
64 (89%)
Retest bad N (%)
7 (10%)
1 (1%)
8 (11%)
Total N (%)
65 (90%)
7 (10%)
72 (100%)
Figure 4. Distribution in percentiles (10th, 25th, 50th, 75th, 90th) of the age‐corrected SRT scores for the first Earcheck by age group. The reference line represents the cut‐off value of ‐18.4 dB. Figure 5. Distribution in percentiles (10th, 25th, 50th, 75th, 90th) of deviations from age‐corrected SRT cut‐off score by age group according to the 14 Earcheck procedure: In case of a ‘poor’ first SRT score, the retest SRT was taken. 5. Discussion Discussion of results The present results of the SRTs of 12‐17 year‐old students and young adults are in line with early studies that showed developmental changes in the SRTs from the age of 3 years to adulthood [Goldman & Fristoe & Woodcock as cited in Elliot 1979, Report Online Hearing Tests 2010‐2014]. In our data, the effect of age on the SRT was independent of the level of education and gender, and a learning effect from test to retest was present in all age groups. It should be noted that 22% (23/104) of the subjects who participated in the present study had to be excluded according to their PTA’s. The evaluation of the present data set did not support the hypotheses that the maturation effect would be smaller for students with a higher education level. The levels of education did not differ in SRT, neither in the overall results nor age‐dependently. In view of the small numbers in the subgroups, however, the insignificance of gender and education has to be taken with reservation. While the effects of gender and the level of education could be disregarded, the data indicated age‐dependent cut‐off values. 6. Main findings of this study With respect to the four research questions introduced in chapter 2, the following results have been found: 1. We found an age‐dependent effect of Earcheck results that possibly can be attributed to maturation of the auditory system in normal‐hearing teenage students aged 12‐18 years. This effect was ‐0.2 dB SNR per year of age, and most prominent in the results of the first test, while insignificant in the second test. 2. Repeated testing revealed a small but consistent learning effect (0.5 dB). 3. The level of education does not show differences in SRT, neither in the overall results nor in age‐dependency. In our group we found no support for hypotheses that the maturation effect would be smaller for students with a higher education level. 4. The effects of gender and the level of education can be disregarded, but the age effect is too large to ignore. Age‐dependent cut‐off values should be introduced. Based on the present data and supported by previous studies, a correction factor of ‐0.2 dB per year of age is proposed for Earcheck SRT cut‐off scores. The learning effect can be compensated for by a test procedure that automatically repeats the test if the result is poor (conditional retest). Such a procedure has been introduced successfully in other internet tests for the discrimination of speech in noise. 15 7. Implications of an age‐dependent cut‐off values To control for the found effect of age, linear regression indicated a correction factor of 0.2 dB per year of age, which resulted in an age‐corrected SRT cut‐off score of ‐17.2 dB for the 12‐year olds, with a decrease to ‐18.4 dB for (young) adults of 18 years and older, in a controlled setting (for a home‐
based setting, an extra correction factor of 1.2 dB should be applied). The age correction of ‐0.2 dB per year of age that we retrieved from our study group is in line with the outcome of previous Earcheck studies [Leensen et al.,2011b; Report online hearing tests 2010‐2014: Five years of Oorcheck & Hoorscan], which revealed an SRT of ‐18.4 dB. With current knowledge this seems to be a valid criterion to control for the unwanted effect of maturation on the Earcheck outcome. However, larger samples of subjects with equally balanced subgroups should be evaluated for more exact compensation values. Retrospective analysis of Earcheck data (2010‐2014), applying the new age‐dependent cut‐off values The performance of teenagers and young adults on the online Earcheck speech‐in‐noise test in the last five years was studied [Sheikh Rashid, Report Online Hearing Tests 2010‐2014, Nationale Hoorstichting, 2015]. According to this study, the proportion of users (aged 12‐25 years) with a good result was 74.5%, while a substantial percentage of 25.5% attained an “insufficient or poor” result (18.5% attained an insufficient result, and 7% attained a poor result), implying a large number of users with a difficulty in understanding speech in noise. An explanation for these unexpected high percentages is the use of an unrepresentative, biased and anonymous convenience sample of teenagers, performing the test in an uncontrolled home setting. Another explanation, however, may be the use of the original cut‐off value for pass/fail categories. Therefore, the same data were reanalyzed retrospectively, applying the newly defined, age‐dependent cut‐off values on all test results, as presented in table 8. Table 12 shows the distributions, also given per year, and per age year (figures 6 and 7). As a result of the new age‐dependent cut‐off values, the percentage teenage users with insufficient or poor scores decreased with 5%, indicating more reliable estimates. Table 12. Distribution of test results, when applying the original, and the age‐dependent cut‐off values, for all years and all teenagers (N=96.803). Good Insufficient Poor Original cut‐off
74.5% 18.5% 7% Age‐dependent cut‐off
79.1%
14.5%
6.4%
16 Figure 6. Test results (per category) for the original and age‐dependent cut‐off values (per year), for all teenagers Figure 7. Test results (per category) for the original and age‐dependent cut‐off values (per year), per age year 17 8. Conclusion The results of this study indicate that an age‐dependent cut‐off value should be used and propose a relatively simple correction factor of 0.2 dB per year between the age of 12 and 18. If the proposed age‐dependent cut‐off values are applied in a large sample of teenagers and young adults, the percentage of subjects with insufficient hearing drops from 18,5% to 14,5%. The percentage of subjects with poor hearing remains about stable (small decrease from 7% to 6.4 %). The resulting data seems to be more realistic, although these data still suffer from inclusion bias and testing in poorly controlled conditions. 18 References Carter L, Williams W, Black D, Bundy A. The Leisure‐Noise Dilemma: Hearing Loss or Hearsay? What Does the Literature Tell Us? Ear & Hearing 2014;35;491–505. Elliot LL: Performance of children aged 9 to 17 years on a test of speech intelligibility in noise using sentence material with controlled word predictability. J Acoust Soc Am 1979; 66: 651–653. Elliot LL, Connors S, Kille E, Levin S, Ball K, Katz D: Children’s understanding of monosyllabic nouns in quiet and in noise. J Acoust Soc Am 1979; 66: 12–21. Henderson, E., Testa, M. A., Hartnick, C. (2011). Prevalence of noise‐induced hearing‐threshold shifts and hearing loss among US youths.Pediatrics, 127, e39–e46. Jansen, S. (2013). Efficient and sensitive hearing assessment based on speech perception. Jin, S. H., Nelson, P. B., Schlauch, R. S., et al. (2013). Hearing conservation program for marching band members: A risk for noise‐induced hearing loss? Am J Audiol, 22, 26–39. Leensen, M.C, Dreschler W.A. The applicability of a speech‐in‐noise screening test in occupational hearing conservation. Int J Audiol. 2013, 52 (7): 455‐65. Leensen, M.C., Dreschler W.A. Speech‐in‐noise screening tests by internet, part 3: test sensitivity for uncontrolled parameters in domestic usage. Int J Audiol. 2013 Jul 2. [Epub ahead of print]. Meyer‐Bisch C. 1996. Epidemiological evaluation of hearing damage related to strongly amplified music (personal cassette players, discotheques, rock concerts): High‐definition audiometric survey on 1364 subjects. Audiology, 35, 121 – 142. Mostafapour S.P., Lahargoue K. & Gates G.A. 1998. Noise‐induced hearing loss in young adults: The role of personal listening devices and other sources of leisure noise. Laryngoscope , 108, 1832 – 1839. Niskar A.S., Kieszak S.M., Holmes A.E., Esteban E., Rubin C. et al. 2001. Estimated prevalence of noise‐
induced hearing threshold shifts among children 6 to 19 years of age: The Third National Health and Nutrition Examination Survey, 1988 – 1994, United States. Pediatric , 108, 40 – 43. Portnuff, C.D.F., Fligor, B.J., Arehart, K.H.2011. Teenage Use of Portable Listening Devices: A Hazard to Hearing? J Am Acad Audiol, 22 (10), 663–677. Serra M, Biassoni E, Richter U, Minoldo G, Franco G, Abraham S, Carignani J, Joekes S, Yacci M. Recreational noise exposure and its effects on the hearing of adolescents. Part I: An interdisciplinary long‐term study. International Journal of Audiology 2005; 44:65‐/73. Shargorodsky J., Curhan S.G., Curhan G.C. & Eavey R. 2010. Change in prevalence of hearing loss in US adolescents. JAMA , 304, 772 – 778. 19 Sheikh Rashid M, Leensen M, de Laat J, Dreschler W. Online Hearing Tests 2010‐2014, AMC 2015. Smits C, Houtgast T. Results From the Dutch Speech‐in‐Noise Screening Test by Telephone. Ear & Hearing 2005;26;89–95. Smits, C., Goverts, T., Festen, J.M. 2013. The digits‐in‐noise test: Assessing auditory speech recognition abilities in noise. J. Acoust. Soc. Am, 133 (3), 1693‐1706. Torre III, P. 2008. Young adults’ use and output level settings of personal music systems. Ear Hear, 29, 791‐799. Zhao F., Manchaiah V.K., French D. & Price S.M. 2010. Music exposure and hearing disorders: An overview. Int J Audiol , 49, 54 – 64. 20