Audiometric notch and extended high-frequency hearing threshold shift in relation to total leisure noise exposure: An exploratory analysis.

BACKGROUND: Studies investigating leisure noise effect on extended high frequency hearing are insufficient and they have inconsistent results. The aim of this study was to investigate if extended high-frequency hearing threshold shift is related to audiometric notch, and if total leisure noise exposure is associated with extended high-frequency hearing threshold shift.
MATERIALS AND METHODS: A questionnaire of the Ohrkan cohort study was used to collect information on demographics and leisure time activities. Conventional and extended high-frequency audiometry was performed. We did logistic regression between extended high-frequency hearing threshold shift and audiometric notch as well as between total leisure noise exposure and extended high-frequency hearing threshold shift. Potential confounders (sex, school type, and firecrackers) were included.
RESULTS: Data from 278 participants (aged 18-23 years, 53.2% female) were analyzed. Associations between hearing threshold shift at 10, 11.2, 12.5, and 14 kHz with audiometric notch were observed with a higher prevalence of threshold shift at the four frequencies, compared to the notch. However, we found no associations between total leisure noise exposure and hearing threshold shift at any extended high frequency.
CONCLUSION: This exploratory analysis suggests that while extended high-frequency hearing threshold shifts are not related to total leisure noise exposure, they are strongly associated with audiometric notch. This leads us to further explore the hypothesis that extended high-frequency threshold shift might be indicative of the appearance of audiometric notch at a later time point, which can be investigated in the future follow-ups of the Ohrkan cohort.

Introduction

Hearing impairment due to exposure to loud noise is increasingly common nowadays. The typical early sign of hearing impairment due to noise exposure is an audiometric notch at the frequencies of 3, 4, and 6 kHz with recovery at 8 kHz,[1] which is identified by the Niskar criteria.[2] In the past decades, many studies have investigated occupational noise exposure and hearing, and its effect on hearing impairment is well known. However, along with the emerging technologies, the concerns regarding hearing impairment, especially in young people, caused by leisure noise exposure such as by personal listening devices (PLDs), discotheque visiting, music festival attending, and so on have increasingly arisen.

Audiometry is performed routinely at conventional frequencies (between 0.25 and 8 kHz) for the measurement of hearing threshold shift. However, studies investigating the effect of leisure noise exposure on conventional frequency hearing threshold shift reported inconsistent results.[3456789101112131415161718] Frequencies below 8 kHz are known to be most important for speech intelligibility; however, extended high frequencies (>8 kHz) also play an important role.[19] Therefore, the current hearing assessment with only conventional frequency audiometry would be insufficient. Moreover, researches have indicated that hearing thresholds in extended high frequency (8–20 kHz) might be affected by noise earlier,[2021] which means extended high-frequency audiometry may identify individuals with beginning hearing loss not yet visible in conventional audiometry. Several studies have explored leisure noise exposure and extended high-frequency hearing threshold shift with contradictory findings. Sulaiman et al.[222324] reported significant differences of hearing thresholds between people exposed to PLD or not at most high frequencies, whereas Silvestre et al.[25] reported no associations between exposure to personal stereo devices and hearing threshold shift at all high frequencies. One explanation for the inconsistent results may be that these studies only considered one leisure time activity without accounting for many other activities. To our best knowledge, there are only two studies[2627] so far investigating total leisure noise and extended high-frequency hearing threshold shift, which took several leisure time activities into consideration. Both studies showed no relationship between total leisure noise exposure and extended high-frequency hearing threshold shift. However, the leisure time activities considered were limited and may not accurately represent the true exposure.

Therefore, there were two main objectives of our study: first, to investigate if hearing threshold shift at high frequencies is associated with audiometric notch identified by Niskar criteria as the indicator for hearing impairment; second, to explore the association between total leisure noise exposure and hearing threshold shift at each extended high-frequency taking as many leisure time activities as possible into account.

Materials and Methods

Study design and participants

Data of our study were collected in the framework of the Ohrkan cohort study (O I: Ohrkan baseline study, O II: the 1st follow-up of Ohrkan study, O III: the 2nd follow-up of Ohrkan study), which aims to monitor noise exposure and hearing threshold shift among adolescents and young adults and is conducted in the city of Regensburg (Bavaria, Germany) for 10 years from 2009.[17] A subpopulation of participants from O III study was recruited by convenience sampling (from January 2015 to July 2016) for this study.

Informed consent of all participants was obtained for study participation, and the study was approved by the Ethics Committee of the University of Regensburg.

Data collection and audiometry

All participants were surveyed with a standardized questionnaire asking information about their sociodemographics (e.g., age, sex, school type, and so on) as well as the time they spent on different leisure time activities (e.g., listening to portable music players, playing video games, watching movies, playing an instrument, visiting discotheques, and so on) per day, per week, or per month during the last 12 months.[17] Four new leisure time activities (visiting music festivals, attending private parties with loud music, fitness courses with music, and shooting) were updated in the follow-up questionnaire compared to the baseline survey due to their relevance for young adults.

The audiometric measurements were conducted in two sound-treated test rooms in the Department of Otorhinolaryngology at the University Hospital Regensburg. The audiometry was performed either with a Madsen/GN Resound Audiometer Aurical and Holmco PD95 headphones or with a Madsen Itera II Diagnostic — Audiometer and Sennheiser HDA200 headphones. Both devices comply with IEC 60645-1: 2017 RLV.[28] Hearing threshold levels for each ear were determined for both conventional frequencies from 0.125 to 8 kHz (0.125, 0.250, 0.5, 1, 2, 3, 4, 6, and 8 kHz) as well as extended high frequencies from 9 to 16 kHz (9, 10, 11.2, 12.5, 14, and 16 kHz) with 5 dB increment. Thresholds at 18 and 20 kHz were also measured but not used in analysis because the audiometry at very high frequencies was not very accurate due to the limitations of measurement devices. Additionally, there were too many missing values at 18 and 20 kHz.

In addition to the pure tone audiometry, tympanometry using MADSEN Zodiac 901 Tympanometer (Otometrics A/S, DK-2630 Taastrup, Denmark) was also performed to exclude participants with middle ear problems. A failure of tympanometry was defined as (a) a peak of the curve at pressure below −100 daPa (decapascal) (low pressure) or (b) a peak of the curve at pressure above +100 daPa (excess pressure), or (c) no peak observable. People wearing hearing aids were also excluded.

Hearing threshold shift and audiometric notch

Hearing threshold for each extended high frequency tested was divided into two categories: participants with hearing threshold above 15 dB in at least one ear were considered as having hearing threshold shift at this frequency. Otherwise (≤15 dB) they were considered as having a normal hearing threshold.[29]

Audiometric notch at 3, 4, and 6 kHz[3031] was identified using the Niskar criteria.[2] Namely, all the following three criteria had to be met at least in one ear:

Threshold values at 0.5 and 1 kHz were ≤15 dB hearing level (better);

The maximum (poorer) threshold value at 3, 4, or 6 kHz was at least 15 dB higher (poorer) than the highest (poorest) threshold value for 0.5 and 1 kHz; and

The threshold value at 8 kHz was at least 10 dB lower (better) than the maximum (poorest) threshold value for 3, 4, or 6 kHz.

Total leisure noise exposure

Total leisure noise exposure was estimated with the same algorithm developed in O I study.[6] Here, sound pressure levels (SPLs) for a variety of leisure time activities were obtained from literature research and experts enquiry. Compared to the baseline study with only 18 leisure time activities, three of the four new activities — “visiting music festivals (SPL: 95 dB(A)),” “attending private parties (SPL: 100 dB(A))” and “attending courses with loud music such as fitness and dance class (SPL: 89.2 dB(A))” were included when calculating total leisure noise exposure in our study and corresponding SPLs of them were acquired through literature research[3233] and expert enquiring as well. Activities associated with SPL lower than 80 dB(A) were not taken into account for statistical analysis (e.g., shooting, one of the four new activities mentioned above).

To quantify total leisure noise exposure per person, we used the information from the questionnaire used in the 2nd follow-up on time spent on defined leisure time activities. With the time and SPL information, the mean equivalent A-weighted continuous SPL of all activities (L Aeq,tot) was calculated. Then, L Aeq,tot was extrapolated to a 40-h working week (L Aeq,40h) to make it comparable with the occupational noise exposure limits. Because there are no health-based limits available for exposure to leisure noise, the regulatory limit for occupational noise exposure of 85 dB(A), which is widely applied in many countries referred here.[34] However, due to the potential overestimation of leisure noise risk,[3536] we assumed 90 dB(A) might be a more appropriate limit. Therefore, L Aeq,40h was finally classified into three categories: (a) <85 dB(A), (b) 85–90 dB(A), and (c) ≥90 dB(A). In addition, the two categories of L Aeq,40h (<90 dB(A) or ≥90 dB(A)) were also used in regression analysis as a comparison.

Covariates

Apart from total leisure noise, several factors that may potentially influence hearing ability were considered. According to the results of O I study,[6] sex, school type, and firecracker exposure were included as covariates.

Statistical analysis

Descriptive analysis of all participants’ characteristics including sociodemographics, firecracker exposure, as well as the total leisure noise was conducted.

Then, we did a descriptive analysis (mean, standard deviation, min, and max) of hearing thresholds for both ears and both genders at extended high frequencies. Non-parametric test was used to detect if there is a significant difference of hearing thresholds between left and right ears or between male and female.

To study if hearing threshold shift at high frequencies is associated with audiometric notch, bivariate logistic regression model was used for each extended high frequency. The independent variable was hearing threshold shift (hearing threshold above 15 dB in at least one ear) at a certain extended high frequency, and the dependent variable was an audiometric notch identified by Niskar criteria.

As a comparison to the previous O I study,[6] we did logistic regression between total leisure noise exposure and audiometric notch. Then, to explore the association between total leisure noise and hearing threshold shift at extended high frequencies, logistic regression analysis was performed using three models for each frequency: bivariate model without covariates, multivariate model 1 adjusted for sex and school type, and multivariate model 2 adjusted for sex, school type, and firecracker exposure. The effect modification of sex and total leisure noise exposure was not included in the models due to the insignificant results of the Wald test.

The significant level of all tests was P < 0.05 and all analyses were performed with the statistical software package SAS version 9.4 (SAS Institute Inc., Cary, NC 27513-2414, USA).

Results

In total, there were 281 participants in the study who had both data on extended high frequency thresholds and questionnaire information. Among them, three were excluded due to middle ear problems or wearing hearing aids. For the final 278 study participants (53.2% female), most were 20 (41.4%) or 21 (43.9%) years old (ranging from 18 to 23 years). Table 1 shows the distribution of sociodemographics and noise exposure (total leisure noise and firecrackers) of all participants. 73.4% participants were categorized to having total leisure noise exposure for a 40-h working week exceeding 85 dB(A) and 34.2% exceeding 90 dB(A).

Table 1

Distribution of sociodemographic variables and noise exposure in 278 participants

Variables	Categories	n (%)
Age (years) in O III	18 or 19	3 (1.1)
	20	115 (41.4)
	21	122 (43.9)
	22 or 23	38 (13.7)
Sex	Male	130 (46.8)
	Female	148 (53.2)
School years at baseline	12	184 (66.2)
	10–11	52 (18.7)
	9	42 (15.1)
Firecracker exposure in O III	Yes	174 (62.6)
	No	104 (37.4)
Total leisure noise exposure in O IIIa (dB(A))	<80	25 (9.0)
	≥80 to <85	49 (17.6)
	≥85 to <90	109 (39.2)
	≥90	95 (34.2)

O III: the 2nd follow-up of Ohrkan study. aThe mean equivalent A-weighted continuous sound pressure level is extrapolated to a 40-h working week (L Aeq,40h).

Table 2 shows the mean, standard deviation, minimum, and maximum of hearing thresholds at each extended high frequency for left and right ear, respectively as well as the hearing threshold shift at each frequency. The highest (worst) mean hearing threshold shift as well as hearing threshold shift percentage was found at 16 kHz for both ears. At most frequencies, there were no significant differences between left and right ear hearing thresholds. Only at 10 and 11.2 kHz, left ear hearing thresholds were significantly worse than right ear thresholds. The mean hearing thresholds of male and female at each extended high frequency are shown separately for left and right ear in Figures 1 and 2. The differences between male and female were not significant at most frequencies except for that at 10 kHz in the left ear and at 10 and 14 kHz in the right ear.

Table 2

Description of hearing threshold level (dB) at each extended high frequency and bivariate logistic regression for the association of hearing threshold shift and audiometric notcha

Frequency (kHz)	Right ear			Left ear			Hearing threshold shiftb N (%)	Odds ratio (95% CI)
	Mean ± SD	Min	Max	Mean ± SD	Min	Max
9	7.03 ± 6.80	0	40	7.82 ± 8.15	0	45	40 (14.39)	2.33 (0.59–9.19)
10c	5.31 ± 7.21	−5	55	6.44 ± 8.12	−5	45	31 (11.15)	5.08 (1.40–18.48)
11.2c	5.41 ± 6.79	0	45	6.85 ± 8.18	−5	55	31 (11.15)	7.72 (2.2–27.07)
12.5	4.64 ± 7.02	0	50	5.43 ± 8.22	0	55	30 (10.79)	8.07 (2.30–28.33)
14	6.58 ± 9.04	0	40	6.04 ± 9.55	0	55	50 (17.99)	4.11 (1.20–14.06)
16	7.52 ± 10.58	0	50	8.35 ± 11.14	0	50	72 (25.90)	1.67 (0.48–5.89)

Identified by Niskar criteria: threshold values at 0.5 and 1 kHz were <15 dB hearing level (better), and the maximum (poorer) threshold value at 3, 4, or 6 kHz was at least 15 dB higher (poorer) than the highest (poorest) threshold value for 0.5 and 1 kHz, and the threshold value at 8 kHz was at least 10 dB lower (better) than the maximum (poorest) threshold value for 3, 4, or 6 kHz. bParticipants with hearing threshold level >15 dB in at least one ear in each frequency. cThe difference of hearing threshold level between right and left ear is significant.

Figure 1

Mean hearing threshold level (dB) at extended high frequencies in left ear by sex

Figure 2

Mean hearing threshold level (dB) at extended high frequencies in right ear by sex

Among 278 participants, 11 (4.0%) were found with audiometric notch identified by Niskar criteria. Table 2 presents the odds ratios [95% confidence interval (CI)] for association between hearing threshold shift at each extended high frequency and audiometric notch. Associations were found at 10, 11.2, 12.5, and 14 kHz, with the odds ratios (95% CI) of 5.08 (1.40–18.48), 7.72 (2.21–27.07), 8.07 (2.30–28.33), and 4.11 (1.20–14.06), respectively. The proportion of participants with hearing threshold shift at every extended high frequency was much higher than that of participants with an audiometric notch.

No association between total leisure noise exposure (both cut-off values) and audiometric notch was observed (not shown). There were no associations between total leisure noise exposure and hearing threshold shift at any extended high frequency either in bivariate or multivariate models. Table 3 shows the odds ratios (95% CI) of associations between extended high-frequency hearing threshold shifts with total leisure noise exposure categorized into three exposure groups (<85 dB(A), between 85 and 90 dB(A), ≥90 dB(A)). The odds ratios of having hearing threshold shift for people with exposure to leisure noise ≥90 dB(A) were above unity at most frequencies, although not significant, while the odds ratios were below unity in people with exposure between 85 and 90 dB(A) compared to people with exposure <85 dB(A). When using 90 dB(A) as cut-off value to divide total leisure noise exposure into two groups, similarly, the odds ratios (95% CI) of having hearing threshold shift for people with exposure to leisure noise ≥90 dB(A) were above unity at most frequencies compared to people with <90 dB(A) noise exposure (not shown).

Table 3

Odds ratios (95% CI) in logistic regression for the association of total leisure noise exposure and hearing threshold shift at extended high frequencies

Frequency (kHz)	Total leisure noise (dB(A))	N (%)	Bivariate modela	Multivariate model 1b	Multivariate model 2c
9	<85	74 (26.6)	1.00	1.00	1.00
	85 to <90	109 (39.2)	0.71 (0.30–1.70)	0.68 (0.28–1.65)	0.71 (0.29–1.73)
	≥90	95 (34.2)	1.25 (0.55–2.86)	1.20 (0.52–2.77)	1.26 (0.54–2.93)
10	<85	74 (26.6)	1.00	1.00	1.00
	85 to <90	109 (39.2)	0.81 (0.32–2.07)	0.82 (0.32–2.13)	0.90 (0.34–2.36)
	≥90	95 (34.2)	0.95 (0.37–2.42)	0.93 (0.36–2.40)	1.02 (0.39–2.67)
11.2	<85	74 (26.6)	1.00	1.00	1.00
	85 to <90	109 (39.2)	0.73 (0.28–1.89)	0.78 (0.30–2.07)	0.82 (0.31–2.19)
	≥90	95 (34.2)	1.04 (0.42–2.63)	1.07 (0.42–2.72)	1.12 (0.44–2.89)
12.5	<85	74 (26.6)	1.00	1.00	1.00
	85 to <90	109 (39.2)	0.93 (0.35–2.43)	0.95 (0.36–2.53)	0.98 (0.37–2.64)
	≥90	95 (34.2)	1.08 (0.41–2.84)	1.09 (0.41–2.90)	1.13 (0.42–3.03)
14	<85	74 (26.6)	1.00	1.00	1.00
	85 to <90	109 (39.2)	0.72 (0.34–1.52)	0.70 (0.33–1.50)	0.74 (0.34–1.59)
	≥90	95 (34.2)	0.73 (0.34–1.59)	0.71 (0.33–1.55)	0.75 (0.34–1.65)
16	<85	74 (26.6)	1.00	1.00	1.00
	85 to <90	109 (39.2)	0.93 (0.48–1.82)	0.95 (0.48–1.87)	1.00 (0.50–1.98)
	≥90	95 (34.2)	0.91 (0.46–1.82)	0.91 (0.45–1.82)	0.96 (0.48–1.95)

No covariate in bivariate model, only total leisure noise exposure. bAdjusted for sex and school type. cAdjusted for sex, school type and firecracker exposure.

Discussion and Conclusion

Our study investigated the association between extended high frequency hearing threshold shift and audiometric notch as well as the effect of total leisure noise exposure on hearing threshold shift at high frequencies. Our results show that hearing threshold shift at 10, 11.2, 12.5, or 14 kHz has a strong and stable association with audiometric notch, with extended high-frequency hearing threshold shift being much more frequent than audiometric notch. We did not find any association between total leisure noise exposure and hearing threshold shift at any extended high frequency.

To the best of our knowledge, this study is the first one to explore the relationship between extended high-frequency hearing threshold shift and audiometric notch. The associations between hearing threshold shift at 10, 11.2, 12.5, and 14 kHz with audiometric notch were clear. At the same time, threshold shift at these high frequencies was more frequent than the notch. On the basis of these observations, the hypothesis that extended high-frequency threshold shift might be indicative of the appearance of audiometric notch at a later time point can be studied in the future follow-ups of the Ohrkan cohort. If this hypothesis is confirmed, it would help to detect noise-induced hearing impairment earlier and would allow more efficient prevention measures.

We found no association between total leisure noise exposure and audiometric notch identified by the Niskar criteria, which is consistent with the previous O I study.[2] In the past decades, many previous studies have investigated the leisure noise effect on hearing threshold shift at conventional frequencies but the results were inconsistent. Several studies revealed associations between leisure noise exposure and conventional frequency hearing threshold shift,[458111215] whereas many others did not.[3679101314161718] Additionally, though audiometric notch identified by Niskar criteria is known to be the typical early indicator of hearing impairment induced by noise and excessive noise exposure is the most common cause of it,[1] some other factors such as genetic factors may also lead to a similar audiometric notch.[2] Another study by Schlauch and Carney even showed high false-positive rates when using audiometric notch as sign of hearing impairment due to noise exposure.[37]

Our study did not reveal any association between total leisure noise exposure and extended high-frequency hearing threshold shift. There are fewer studies about leisure noise exposure and extended high-frequency hearing threshold shift. Most reported significant results,[1522232427383940] whereas only a few studies showed no relationship between leisure noise and extended high-frequency hearing threshold shift.[252641] Compared to the previous studies, our study had obvious strength. First, most previous studies only focused on one or a few leisure time activities such as PLDs, nightclub or discotheque visiting, instrument playing, and so on, whereas our survey considered comprehensively a wide range of leisure time activities to estimate the total leisure noise exposure as realistic as possible. Furthermore, we investigated not only the time spent on each leisure time activity, but also the SPLs corresponding to them to try to calculate the more precise mean equivalent A-weighted continuous SPL of each participant, whereas most previous studies only considered the time of leisure noise exposure.

There are several possible reasons that might explain why we found no association between total leisure noise exposure and hearing ability in our study. First is the limitation of cross-sectional design. We used leisure noise exposure data from O III study instead of exposure data at baseline for the reason that the new questionnaire covers more leisure activities and better represents the true exposure compared to the previously used version. However, as it takes a while for the noise exposure to harm the hearing, the harming effect of the noise exposure measured at the 2nd follow-up may not appear yet. In addition, the retrospective self-reported questionnaire survey of leisure time activities in hours per day, week, or month during the last 12 months may lead to recall bias from participants when estimating their average exposure time of leisure time activities. In addition, as the interindividual variation in hearing thresholds might be too large to disclose the slight effect from different exposure categories of leisure noise, one of the solutions is to increase the study sample size as well as to do a long-term evaluation of young people’s hearing and total leisure noise exposure to further confirm the results. Furthermore, the estimations of SPL for leisure time activities were obtained from literature research and experts enquiry, which may be rather uncertain and may have influenced the results. A standardized and recognized SPL assessment system needs to be established for further studies.

Finally, another possible explanation for the insufficient evidence of hearing threshold shift caused by leisure noise exposure is that our assessment of leisure noise’s harm to hearing is based on an occupational noise limit. It results in the potential overestimation of the true risk of leisure noise such as loud music exposure, as the differences of spectrum as well as temporal and dynamic variation between leisure noise and occupational noise are distinct.[3536] When using 85 and 90 dB(A) as the cut-off values of leisure noise exposure in our study, compared to the reference group (people with lowest noise exposure level), the odds ratios of having hearing threshold shift for people with exposure between 85 and 90 dB(A) were below unity, although not significant, while the odds ratios were above unity in people with exposure above 90 dB(A) at most extended high frequencies. This was similar with the results of using 90 dB(A) as cut-off value. Our results indicated that 90 dB(A) might be a more appropriate limit for leisure noise exposure compared to the occupational noise limit of 85 dB(A), and it needs to be further confirmed in future studies.

In conclusion, our study found that hearing threshold shift at 10, 11.2, 12.5, and 14 kHz is associated with audiometric notch in cross-sectional data, leading us to further explore the hypothesis of an indicator function of the extended high-frequency audiometry. However, we did not observe the association between leisure noise exposure and extended high-frequency hearing threshold shift in our study. This may be due to the cross-sectional design, sample size, young age, and reporting error, but it might show up in the further follow-up of the Ohrkan study. Thus, we could not exclude the potential usefulness of extended high-frequency audiometry in the early detection of noise-induced hearing impairment. The cohort design of the Ohrkan study with a follow-up of all participants in the next few years will enable us to confirm our current findings at the four high frequencies mentioned above, and to further explore the potential association between leisure noise exposure and extended high-frequency hearing threshold shift.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.