Shooting history and presence of high-frequency hearing impairment in swedish hunters: A cross-sectional internet-based observational study.

The aim of this cross-sectional study among Swedish hunters was to examine the association between shooting history and presence of high-frequency hearing impairment (HFHI). All hunters registered with an e-mail address in the membership roster of the Swedish Hunters' Association were invited via e-mail to a secure website with a questionnaire and an Internet-based audiometry test. Associations, expressed as prevalence ratio (PR), were multivariately modelled using Poisson regression. The questionnaire was answered by 1771 hunters (age 11-91 years), and 202 of them also completed the audiometry test. Subjective severe hearing loss was reported by 195/1771 (11%), while 23/202 (11%) exhibited HFHI upon testing with Internet-based audiometry. As many as 328/1771 (19%) had never used hearing protection during hunting. In the preceding 5 years, 785/1771 (45%), had fired >6 unprotected gunshots with hunting rifle calibers. The adjusted PR of HFHI when reporting 1-6 such shots, relative to 0, was 1.5 [95% confidence interval (CI) 1.1-2.1; P = 0.02]. We could not verify any excessive HFHI prevalence among 89 hunters reporting unprotected exposure to such gunshot noise >6 times. Nor did the total number of reported rifle shots seem to matter. These findings support the notion of a wide variation in individual susceptibility to impulse noise; that significant sound energy, corresponding to unprotected noise from hunting rifle calibers, seems to be required; that susceptible individuals may sustain irreversible damage to the inner ear from just one or a few shots; and that use of hearing protection should be encouraged from the first shot with such weapons.

Introduction

Recreational firearms can produce peak noise levels within the range of 156-164 decibel (dB) sound pressure level (SPL)[12] Human challenge experiments in the past have demonstrated that a single gunshot can cause a severe temporary threshold shift (TTS),[34] but the susceptibility seemed highly variable.[345] The recovery period after such acoustic trauma is unpredictable and a permanent threshold shift (PTS) cannot be excluded. Today, such challenge experiments in humans would be considered unethical. Although several studies have measured noise levels from different types of weapons,[126] there are only sparse observational data on dose-risk relationships and effect-modifying factors in humans engaged in real-life hunting.[78910]

Methods

This cross-sectional observational study among Swedish hunters aimed to examine shooting history (rifle calibers, hearing protection, and frequency of shots in total and shots without protection) in relation to the probability of having high-frequency hearing impairment (HFHI).

Study group

In total, 27063 hunters were contacted via e-mail between Feb 2012 and Aug 2013. The hunters’ e-mail addresses were provided by the Swedish Hunters’ Association, the largest association for hunters in Sweden. Nonresponders received two e-mail reminders.

Measurements

An Internet site was set up for the purpose of the study. The site presented a questionnaire and a hearing test to the participants. The participants entered the study by creating a personal log-in. To access the hearing test, the questionnaire had to be first completed.

Questionnaire

The questionnaire consisted of 143 questions and subqueries, of which a part was analyzed in this study. The questions concerned sociodemographic characteristics, hunting habits, caliber use, use and type of hearing protection, number of shots with and without hearing protection, and self-estimated hearing. The questionnaire also covered a wide array of possible confounding factors, e.g., exposure to noise at work, during military service, or in connection with leisure-time activities; ear diseases; and presence or previous occurrence of other diseases possibly affecting hearing ability. We also asked about regular use of anticoagulants and analgesics. Moreover, we probed into the use of cigarettes and smokeless tobacco (snuff). We also asked about family histories of hearing impairment, including use of hearing aids. To be able to proceed, all items were required to be answered. The questionnaire was evaluated in a pilot study, where the questionnaire and Internet-based hearing test combination was tried out on 560 Swedish hunters.[11] The response rate was only 29%, but among those who logged in to the study website, 92% completed the questionnaire and 50% carried out the InternetAudio test.[11]

Internet-based audiometry test (InternetAudio test)

We developed the InternetAudio test, with headphones connected to a regular home computer, and validated the test against clinical audiometry.[12] Biological calibration against a normal hearing reference person was required before the actual hearing test could be executed, to compensate for the differences in computer equipment and in the headphones’ frequency range variation, as well as the ambient background noise levels. The calibration was an analogue to the clinical zero dB reference value.[13] The participants were instructed to each ask a normal hearing person to be a reference person. The reference person had be a person who estimated that his/her hearing was normal, had not been exposed to excessive noise, and had not suffered from any ear disease. The age of the reference person was not to exceed 35 years. For the calibration to be accepted, five separate criteria, described in an earlier report,[12] had to be met. The reference person's hearing level (RP-HL) was considered to be the 0 level of dB for this computer, headphones, and environment. Hence, the hearing test measured the differences in frequency-specific hearing levels (HLs) between the tested participant and the reference person.

Six frequencies, 0.5 kilohertz (kHz), 1 kHz, 2 kHz, 4 kHz, 6 kHz, and 8 kHz were tested in the actual hearing test. Tones were presented to each ear separately in random order and at random intervals varying between 4 s and 6 s. The sound levels ranged 0-60 dB RP-HL. The test person determined his/her HL by pressing the keyboard space bar when a tone was detected. If the tone was not heard, it was presented at a level 10 dB higher. If the tone was heard, it was presented at a level 5 dB lower. To avoid ambiguity, the space bar had to be pressed within 1 s after presentation of the tone. If more than five responses were recorded outside these time limits, the test was automatically classified as being invalid and the test person was asked to do the test again. The final and approved hearing result was saved on the Internet site server and presented to the participant as an audiogram on the computer screen.

When participants had done more than one valid InternetAudio test (n = 102), we chose the last one, on the assumption that the final test was the test with which the participant was most satisfied.

The InternetAudio test has earlier been formally validated, with a clinical pure-tone air-conducted audiometry as the gold standard, in 72 volunteers with a broad range of hearing ability.[12] The Pearson's correlation coefficients were 0.94 and 0.93 for the right and the left ear, respectively, while the sensitivity and specificity for any hearing loss, as defined by Heijbel-Lidén[14] were 75% and 96%, respectively. The Pearson reliability coefficient was 0.99.[12]

The InternetAudio test was further validated while used under authentic conditions in the present investigation. We contacted all participants who had completed the test at home and had given an affirmative answer to a question in the questionnaire about their willingness to undergo a pure-tone clinical audiometry at the department of audiology at the Karolinska University Hospital, in Stockholm. Thirteen (13) participants completed the latter examination. In the comparison between the InternetAudio test and reference audiometry, adjustments were done as indicated by our first validation study.[12] This meant that if in the InternetAudio test there was no measured response at a given frequency, an arbitrary level of 65 dB (RP-HL) was given, as this test does not examine higher dB values. This adjustment was also made in the clinical audiogram when the decibel hearing level (dB HL) value surpassed 60.

After the exclusion of one participant with an invalid InternetAudio test, the Pearson correlation coefficients between results (average HL across all six frequencies) from the two tests were 0.95, 95% confidence interval (CI) (0.83, 0.99) for the left ear and 0.76, 95% CI (0.33, 0.93) for the right ear, which corresponds with the CIs of the formal validation test.[12] In Figure 1, box-and-whisker diagrams depict, for each tested frequency, the distribution of intraindividual differences. A negative value indicates that the clinical pure-tone audiogram dB HL value was higher than that obtained with the InternetAudio test. For all frequencies, there was a general tendency toward higher dB values in the clinical audiogram. Although the most extreme values (marked with a + sign) show vast deviations for single individuals, the mean differences were small.

Figure 1

Comparison between frequency-specific HLs obtained with the InternetAudio test, self-administered at home, and those yielded on professionally conducted clinical pure-tone audiometry in 12 volunteering participants. Box plot of the distributions of intraindividual differences in hearing levels (dB — Y axis), frequency by frequency (kHz — X axis). ◊ = Mean, bold line = Median, bottom of box = First quartile, top of box = Third quartile, whiskers = Range + = Outliers

Evaluation of impact of nonparticipation

To examine possible selection bias due to nonparticipation, we sent an e-mail to all 13884 invitees who had not responded within 18 months after the initial invitation. The e-mail contained three simple questions about exposure and outcome:

In total, how many shots, on average, do you fire per year (with an open-ended response)?

How many of these shots are fired without any hearing protection (in percent)?

How would you rate your hearing ability (good/slightly reduced/severely reduced)?

The study was approved by the Stockholm Regional Ethics Vetting Board.

Analysis

Measure of exposure

Participants self-estimated the annual number of fired training shots and hunting shots separately, categorized as shown in Table 1. The following classification of weapon types was used:

Table 1

Hunting-related characteristics of study participants who answered the questionnaire (n = 1771), and the subset who completed the InternetAudio test (n = 202)

Variables	Questionnaire n/%	InternetAudio test n/%
Overall	1771/100	202/100
Number of shots without protection in past 5 years, per caliber
Hunting rifle calibers
0	733/41	96/48
1-6	253/14	22/11
>6	785/45	84/42
Lighter hunting weapons
0	655/37	69/34
1-6	361/20	44/22
>6	755/43	89/44
Shotguns
0	1023/58	117/58
1-6	116/7	11/5
>6	632/36	74/37
Number of shots in total on average per year, per caliber
Hunting rifle calibers
0-50	50/3	4/2
51-100	445/25	47/23
101-200	422/24	48/24
>200	854/48	103/51
Lighter rifle calibers
0	543/31	60/30
1-90	495/28	49/24
91-400	467/27	55/27
>400	266/15	38/19
Shotguns
0-100	1600/90	185/92
101-300	118/7	12/6
301-500	24/1	2/1
>500	29/2	3/1
Number of hunting days per year
0	18/1	2/1
1-5	94/5	10/5
6-10	273/15	37/18
11-20	457/26	45/22
>20	929/53	108/53
Ear protection use during hunting
Always	686/39	70/35
Often	372/21	39/19
Sometimes	227/13	26/13
Rarely	158/9	25/12
Never	328/19	42/21
Type of ear protection used during hunting
Earplugs	87/n. a	12/n. a
Earmuffs	78/n. a	6/n. a
Electronic earmuffs	1171/n. a	134/n. a
Electronic earplugs	107/n. a	8/n. a

Hunting rifle calibers, including Magnum calibers;

Lighter rifle calibers; and

Shotgun calibers [Table 2].

Table 2

Examples of rifles in the three categories: hunting rifle calibers, lighter rifle calibers, and shotguns

Hunting rifle calibers	Lighter rifle calibers	Shotguns
Most common calibers
6.5×55 mm	.22 long rifle (LR)	.20
.308 Winchester
7×57 mm
30-06 Springfield		.16
7 mm Remington Magnum
.300 Winchester Magnum
.338 Winchester Magnum		.12
375 Holland and Holland
Uses
Larger game	Small game, training	Game: bird, rabbit, hare, badger, fox; and training
Unweighted peak SPL (dB), microphone placed 3 cm from the shooter’s left ear*
163 (30-06)	141 (.22cal rifle)	161 (.12 gauge)

*According to reference 2

Of particular interest was the number of shots fired without hearing protection in the preceding 5 years, and the participants self-reported these numbers separately for each weapon type. Shots fired with hunting rifle calibers generate the highest dB SPL peak levels. The number of unprotected shots (categorized into 0, 1-6, and >6) with these weapons was the exposure of main interest in this study. This shot categorization was chosen in order to have reasonable, equally sized groups for the statistical analysis.

Covariates

Graded variables

As age showed an approximately linear relationship with the outcome, age was treated as a discrete numerical variable in the analyses. However, in order to convey a better picture of the age distribution, it is categorized in Table 3. This Table also displays the categorizations of place of residence and education level. Self-assessed hearing status was categorized according to the response categories in the questionnaire (“good,” “slightly reduced,” “severely reduced,” “don’t know”), along with a special category for missing answers.

Table 3

Characteristics of study participants who answered the questionnaire (n = 1771) and the subset who completed the InternetAudio test (n = 202)

Variables	Questionnaire n/%	InternetAudio test n/%
Overall	1771/100	202/100
Sex
Men	1610/91	177/88
Women	161/9	25/12
Age (years)
11-35	245/14	22/11
36-47	366/21	51/25
48-56	366/21	44/22
57-65	385/22	46/23
66-91	409/23	39/19
Place of residence
City	203/11	16/8
Suburb	189/11	20/10
Medium-sized town	268/15	32/16
Municipality	383/22	35/17
Rural area	728/41	99/49
Highest education level
Elementary school	225/13	35/17
High school	648/37	70/35
University	898/51	97/48
Self-assessed auditory status
Good	628/36	58/29
Slightly reduced	940/53	121/60
Severely reduced	195/11	23/11
Don’t know	2/1	0
Missing	6	0
Non-hunting-related noise*
Yes	1345/76	162/80
No	426/24	40/20
A family history of impaired hearing
Yes	477/27	46/23
No	1134/64	136/67
Don’t know	149/8	20/10
Missing	11	0
Ear disease**
Yes	328/19	29/14
No	1443/81	173/86
History of CVD***
Yes	500/28	52/26
No	1271/72	150/74
Tobacco use
Yes	418/24	47/23
No	1353/76	155/77
Other disease****
Yes	669/38	71/35
No	1102/62	131/65
Anti-inflammatory/anticoagulant drug use
Yes	446/25	47/23
No	1325/75	155/77

*History of noise at work past or present, noise during army service, tinnitus after army service, or tinnitus after leisure activities, **History of Ménière’s, sudden deafness, vestibular schwannoma, ear surgery, chronic otitis media, or ear disease as a child, ***History of myocardial infarction and/or hypertension and/or use of anticoagulants, ****History of meningitis, migraine, head trauma, epilepsy, joint disease, or cancer

Dichotomous variables

Classified as exposed to “non-hunting-related noise” were participants who answered affirmatively to at least one of the following questions: If present or past work was believed to have damaged the hearing ability by noise exposure, or if hearing deteriorated during military service, or if military service or leisure activities had caused tinnitus. Only those participants who indicated present use of cigarettes and/or smokeless tobacco were classified as “tobacco users.” Those who self-reported hypertension, previous myocardial infarction, or use of anticoagulants were categorized as positive for cardiovascular disease (CVD+). If there was a history of Ménière's disease, sudden deafness, vestibular schwannoma, ear surgery, chronic otitis media, or ear disease as a child, the participant was classified as having “ear disease.” Similarly, the classification for “other disease” was positive if any affirmative answer was given with regard to a history of meningitis, migraine, head trauma, epilepsy, diabetes, joint disease, or cancer. If an affirmative answer was given with regard to the use of anticoagulants, painkillers, or anti-inflammatory medication at least twice a week in the last month, the participant was classified as an “anticoagulant/anti-inflammatory drug user.” When an affirmative answer was given to a question about any family members with a hearing impairment in need of hearing aid before 65 years of age, the participant was classified as having “a family history of hearing impairment.”

Measures of outcome

Our primary goal was to capture HFHI, and for this reason we only considered hearing levels for the frequencies 4 kHz and 6 kHz, typically affected by severe acoustic trauma. RP-HL values above 20 dB for either of these two frequencies on either ear were categorized as HFHI.

Statistical methods

We modeled the probability of having HFHI as function of the exposure and the covariates described above. We used this model to estimate prevalence ratios (PRs) of HFHI associated with different exposure and covariate levels, as a consistent and conservative alternative to prevalence odds ratios.[15] The actual model fit was performed via an approximate Poisson model with robust standard errors,[16] which provides less biased estimates than logistic regression.[17] The PRs reported below were obtained as the exponentiated regression coefficients of the approximating Poisson model. To account for the binary outcome we used the robust sandwich estimator for modeling of the variance. Because of the dominant effect of age on hearing impairment, we performed an extensive goodness-of-fit analysis where we compared the model with a linear term for age to more complex models involving quadratic term, spline function, or categorization of age. In the regression analysis we used a model with a linear term for age. The statistical analyses were conducted using SAS software, version 9.4, of the SAS System for Windows (SAS Institute Inc., 100 SAS Campus Drive Cary, NC, USA)

Results

Responders

Figure 2 shows the passage of invitees through the study. Of the 27063 e-mail addresses received from the Swedish Hunters’ Association membership roster, 9123 did not lead to functioning e-mail accounts and bounced back, while 2119 reached individuals who actively declined participation. In all, 1937 persons started the survey, 1771 of them completed the questionnaire, and 202 also completed the InternetAudio test. Of the 13 who completed an additional clinical audiometry for validation purposes, one participant was excluded due to technical evidence suggesting invalid performance on the InternetAudio test.

Figure 2

Flow diagram

Men constituted the overwhelming majority among participants [Table 3]. The youngest participant was 11 years of age and the oldest 91 years of age. Mean age was 52.9 (SD 14.4) years. The overall education level (51% with university education) was higher than in the Swedish population as a whole (25% in 2012).[18] Hearing impairments, both the participants’ own and among relatives, were fairly common: Only about one-third self-reported “good hearing” (even less among those who completed the hearing test), and 11% reported severely reduced hearing.

Hunting-related exposures among the 1771 participants who answered the questionnaire and the subset of 202 participants who completed the hearing test are shown in Table 1. The proportions of hunters who had never fired any unprotected shots were around 40% for both hunting rifle calibers and lighter hunting weapons. For shotguns, 58% in both groups had never fired any such shots. Between 36% and 45% in both groups had fired more than 6 unprotected shots with all types of weapons in the preceding 5 years. About one-fifth of the participants never used hearing protection [Table 1], but among the 1443 who did, electronic, level-dependent earmuffs (electronic earmuffs) were the most popular (81%).

Nonresponders

Of the 3746 hunters who answered the three questions posed to nonresponders, 42% indicated slightly reduced hearing and 7% severely reduced hearing. Forty-two percent (42%) had never fired any unprotected shots in the preceding year. Fifty-six percent of (56%) the nonresponders had fired more than 6 unprotected shots last five year. This figure is consistent with the proportions among both the 1771 responders to the questionnaire and the 202 who carried out the InternetAudio test (around 65% in the preceding 5 years). This suggests that the study participants, on average, perceived their hearing ability as being slightly worse than the invitees’ in general, but also that their shooting pattern was not dramatically different.

Association between exposure to unprotected noise from hunting rifle caliber weapons and HFHI

Of the 202 participants who completed the InternetAudio test, 97 (48%) were found to have HFHI [Table 4]. HFHI was significantly linked to exposure to unprotected shooting noise from hunting rifle caliber weapons in the preceding 5 years. The HFHI prevalence among participants with 1-6 such reported shots was 50% higher than among those with none. However, among 89 hunters who reported more than 6 such shots, the prevalence of HFHI was essentially the same as in the reference category with no unprotected shots. Thus, there was no consistent dose-response relationship between the accumulated audio trauma over time and HFHI prevalence. While age was strongly associated with HFHI prevalence (50% increase in HFHI prevalence with each 10-year age increment), age was not linked to the exposure to unprotected noise from hunting rifle caliber weapons (data not shown). Hence, age did not confound the crude association between unprotected hunting rifle caliber noise exposure and HFHI prevalence. Moreover, further multivariable adjustments, as described in a footnote of Table 4, did not materially change the observed crude prevalence ratios, despite suggested crude associations between HFHI and several of the covariates (a reported previous history of ear disease, CVD, non-hunting-related noise exposure, highest attained education, and tobacco use - data not shown). Of note, for all investigated calibers, the reported average total yearly number of shots (of which the overwhelming majority were fired with ear protection in use) was unrelated to the HFHI prevalence (data not shown), and this aspect of the shooting habits did not confound the observed association between exposure to unprotected shooting noise from hunting rifle calibers and the prevalence of HFHI.

Table 4

Associations, expressed as PRs, between exposures of main interest (unprotected shots with, respectively, hunting rifle calibers, lighter rifle calibers, and shotguns) and the probability of HFHI (a hearing impairment >20 dB on 4 kHz and/or 6 kHz). The analysis is restricted to participants who completed the InternetAudio test (n = 202)

Variables	Prevalence of HFHI in category n/N (%)	PR with 95% CI (within parentheses)
		Crude	P value	Age and sex adjusted	P value	Fully adjusted*	P value
Overall	97/202 (48)
Number of unprotected shots in the preceding 5 years per caliber
hunting rifle calibers
0	30/69 (43)	1 (reference)		1 (reference)		1 (reference)
1-6	28/44 (64)	1.5 (1.0-2.1)	0.03	1.5 (1.1-2.1)	0.02	1.5 (1.0-2.2)	0.05
>6	39/89 (44)	1.0 (0.7-1.4)	0.97	1.1 (0.8-1.4)	0.73	0.9 (0,6-1,3)	0.67
Lighter rifle calibers
0	59/117 (50)	1 (reference)		1 (reference)		1 (reference)
1-6	7/11 (64)	1.3 (0.8-2.0)	0.34	1.5 (1.0-2.3)	0.07	1.2 (0.7-2.1)	0.43
>6	31/74 (42)	0.8 (0.6-1.1)	0.26	0.9 (0.7-1.2)	0.47	0.9 (0.7-1.2)	0.51
Shotguns
0	41/96 (43)	1 (reference)		1 (reference)		1 (reference)
1-6	10/22 (45)	1.1 (0.6-1.8)	0.81	1.3 (0.8-2.0)	0.28	1.1 ((0.7-1.9)	0.65
>6	46/84 (55)	1.3 (0.9-1.7)	0.11	1.1 (0.9-1.5)	0.37	1.3 (0.9-1.9)	0.11
Age (continuous)
Per 10 years		1.5 (1.4-1.7)	<0.01	1.5 (1.4-1.7)**	<0.01	1.6 (1.4-1.8)**	<0.01

*Poisson model included age; sex; number of shots with no ear protection with, respectively, hunting rifle calibers, lighter hunting calibers, and shotguns; non-hunting-related noise exposure; education level; tobacco use, CVD (history of myocardial infarction and/or hypertension and/or use of anticoagulants); ear diseasel; and family history of hearing impairment, **Not age-adjusted

Associations between exposure to unprotected noise from rifles with other calibers and HFHI

Crude associations between HFHI prevalence and unprotected shooting noise from rifles with other calibers were, at most, only suggested [Table 4]. Adjustments for age resulted in moderate changes of the HFHI prevalence ratio estimates, but in the multivariately adjusted full model, which included adjustment for exposure to hunting rifle caliber noise, the associations were statistically nonsignificant.

Discussion

The most salient finding of this cross-sectional epidemiological study is the significant, 50% higher prevalence of HFHI among hunters who reported having fired 1-6 shots from hunting rifle caliber weapons without any hearing protection — doubtlessly exposed to the highest sound energy levels occurring in sports hunting — compared to those who reported not having fired any such shots without protection in the past 5 years. The observed result persisted after multivariable control for a large number of conceivable confounding factors, including age. In the fairly large group of hunters who reported more than 6 shots (twice as many hunters as those who reported 1-6 shots) the HFHI prevalence was, however, close to that observed among those reporting no shots. Accordingly, there was no linear relationship between the accumulated number of such ototrauma iterations and the outcome of the InternetAudio test. Further, exposure to unprotected shooting noise from weapons that generate less sound energy was unrelated to HFHI prevalence. Moreover, the reported total annual number of shots, the vast majority of them fired with hearing protection in place, did not seem to be related to HFHI, regardless of caliber.

The observation that one or a few unprotected high-energy impulse sound blasts from hunting rifle caliber weapons seemed to have an effect on the audiogram, in contrast to a history of no unprotected shots, is expected and is supported in the previous literature.[4578910192021] Taylor and Williams[7] were the first to compare audiograms in sports hunters and nonshooting controls, and found significant differences in all frequencies above 3 kHz. Nondahl et al. reported 57-100% increased odds of a marked high-frequency (4 kHz, 6 kHz, and 8 kHz) hearing loss among hunters and an increasing prevalence with increasing number of years of shooting.[8] In an American study among workers exposed to occupational noise, recreational shooting was associated with an average of approximately 5-10 dB higher dB HL in the 3-6 kHz frequencies, though with no statistically significant relationship to the number of unprotected shots fired per year.[9] Similarly, a Norwegian survey found the same frequencies to be most affected by impulse noise (mostly shooting).[10]

It is, however, difficult to explain why the prevalence of HFHI was lower among hunters who reported >6 unprotected shots than among those with less exposure. We believe that it would be a mistake to dismiss the observed result as a chance finding. The most plausible explanation is individual differences in the susceptibility to ototraumatic events such as intense exposure to impulse noise. Those hunters who have vulnerable inner ears might stop shooting without protection already after a few shots, when the adverse effects of impulse noise, such as temporary hearing loss and tinnitus, become apparent. Hunters with resistant inner ears might tend to ignore the risk of unprotected shots as they do not have the same unpleasant experiences. If they at all get any TTS, they may recover more quickly than do susceptible individuals. In historical provocation experiments,[24] some volunteers did not show any TTS when exposed, whereas others exhibited very marked and pronounced TTS. The latter individuals had detectable hearing loss for a long time (>2 months) after exposure, even though the depth of the hearing loss diminished over time.[3] This between-individual variation in recovery to impulse noise was also seen by Luz and Hodge[22] and by Dancer et al.[4]

In general, the prevalence of HFHI was rather high also in the reference categories, the members of which were not exposed to unprotected shots. This could, in part, be explained by the age structure among participating hunters. The association between age and HFHI prevalence was clearly evident in our data, and the effect of age has been shown to easily override the effect of long-time noise exposure with dB levels of 85-90.[23] With an essentially linear relationship between age and HFHI prevalence, the removal of confounding ought to have been rather efficient in our analyses with age as a continuous variable. This notwithstanding, the crude prevalence ratios did not change materially after adjustment for age because age was basically unrelated to the shooting exposures. The generally high HFHI prevalence could also be partly explained by our study being restricted to hunters. All participants had a positive shooting history, and those serving as the reference categories, who were not exposed to unprotected shots in analyses of one type of weapon, were invariably exposed to other types of shooting noise. This lack of contrast could have impeded our ability to verify small effects of shooting noise.

Therefore, the absence of a relationship between HFHI prevalence and the use of other types of firearms in this study should not be taken as evidence of their audiological safety. In addition, the InternetAudio test, and possibly even the pure-tone audiogram, may be too blunt an instrument to expose the full inner ear damage caused by the noise from these types of weapons. Despite normal audiometry, Pawlaczyk-Luszczynska et al.[21] were able to demonstrate effects on the cochlea caused by unprotected shooting noise from small-caliber firearms with transient-evoked otoacoustic emissions.

Even if it is carried out meticulously, the InternetAudio test cannot compete in accuracy with a clinical pure-tone audiogram. However, a previous validation study under ideal conditions[12] and a small validation substudy within the present study both suggest a good correlation with the clinical pure-tone audiogram results serving as the gold standard. Unfortunately, due to limited willingness among participating hunters and logistic constraints (only hunters living in Stockholm County could be invited), the sample size of the present validation substudy is too small to allow precise statistical estimations of the agreement. It is also not possible to rule out instances of total failure to follow the instructions: For instance, leaving the InternetAudio test running unattended, thus generating a false positive result indicating severe hearing loss for all frequencies; or by serving as both reference person and testee, resulting in an InternetAudio test with perfect hearing. In addition, the InternetAudio test has a tendency, in both validity evaluations, to underestimate the hearing thresholds in all frequencies, and by doing so potentially missing out on hearing loss.

The cross-sectional study design is associated with certain limitations. First and foremost, the timing of exposure and outcome is unobservable, and the direction of possible causal effects is far from obvious. The observed association between 1-6 unprotected shots with a hunting rifle and HFHI prevalence in the present study is interpreted as a possible direct causal effect, while the lack of a linear positive dose-effect trend is interpreted as reverse causation, i.e., the outcome (HFHI) has affected the exposure (number of unprotected shots). Second, since participants were asked to recall shooting habits in the past, recall bias is a viable possibility. Vulnerable hunters who develop hearing loss might remember unprotected shots better than resistant hunters whose hearing has remained unaffected. This could potentially lead to spurious associations. Third, the observational study is open to confounding, i.e., incorrect attribution of causal effects. The rich information on covariates, however, has allowed us to control for most conceivable confounding factors, but residual confounding due to insufficiently fine categorization, or confounding from unmeasured causal factors (e.g., genetic factors) cannot be totally excluded. The low participation rate and partial nonresponse pose a threat to the external validity of our results, and if the participation or response inclination was linked to both exposure and outcome, internal validity is also at risk. Participants indicated slightly more problems with their hearing ability but only a marginally higher incidence of unprotected shooting compared to the nonparticipants who answered our simple e-mail questionnaire of three questions. However, the proportion of all nonparticipants who provided this information was too small to allow any trustworthy generalization to the entire Swedish population of sports hunters. In any case, the possibility of selection bias is an important caveat. In a pilot study,[11] lack of time was the major reason given for nonparticipation, and full participation in this study admittedly demanded both motivation and time. This was clearly reflected by the successive attrition between entry into the study (n = 1937) and the successful completion of the hearing test (n = 202). As expected, our participants had a higher average education level than the Swedish population on the whole.

Keeping all caveats in mind, we still dare to state that our results are consistent with the following tentative conclusions:

There is a wide variation in individual susceptibility to high-energy impulse noise;

Significant sound energy, corresponding to unprotected noise from hunting rifle calibers, seems to be required for the development of HFHI;

Susceptible individuals may sustain long-lasting or possibly irreversible damage to the inner ear from just 1 or a few shots (and then stop using these weapons without protection); and

The mandatory use of adequate hearing protection should be strongly encouraged already from the first shot with such weapons.

The mechanisms behind the variation in susceptibility warrant further studies.

List of abbreviations

Temporary threshold shift (TTS), permanent threshold shift (PTS), high-frequency hearing impairment (HFHI), hearing impairment (HI), speech reception thresholds (SRT), decibel (dB), decibel hearing level (dB HL), kilohertz (kHz), reference persons’ hearing level (RP-HL), cardiovascular disease (CVD), Internet-based internet audiometry (InternetAudio test), prevalence ratio (PR), pure-tone average (PTA), confidence interval (CI).

Financial support and sponsorship

Grants from the Swedish Research Council, the Acta Oto-Laryngologica Foundation, and the Tysta Skolan Foundation.

Conflicts of interest

There are no conflicts of interest.