Millions of adults are at risk of hearing loss resulting from exposure to occupational and recreational noises. Data from the combined National Health and Nutrition Examination Survey (NHANES) 2011–2012 and 2015–2016 datasets were used to establish the prevalence of occupational and recreational noise exposures through self-report questions. For recreational noise exposures, NHANES asked about the use of firearms, including the use of hearing protection devices (HPDs) while shooting, and off-work exposures to very loud noise. For work exposures, NHANES asked about exposures to loud and very loud noise. For four of these five questions, graded responses on a 5- or 7-point scale were available. Receiver-operating-characteristic analyses were used to optimize the criterion response for identification of hearing loss for each question with graded responses using the unweighted data. Correlations among the graded responses supported reduction to two measures: (1) rounds fired combined with use of HPDs while shooting and (2) work exposure to loud and very loud noise combined. Logistic-regression analyses of various measures of pure-tone hearing loss were performed to examine the effects of recreational and occupational noise exposures on hearing loss. The odds of hearing loss were significantly greater for those who reported recreational and combined noise exposures.

According to the U.S. Centers for Disease Control (CDC), about 17% of adults aged 20–69 years have noise-induced hearing loss (NIHL) (CDC, 2022a). Exposure to excessive noise is not solely an American phenomenon. The World Health Organization (WHO), for example, noted that “approximately 16% (7%–21% across different regions) of hearing loss in adults [worldwide] results from exposure to excessive noise in the workplace…[and] of persons 12–35 years, 50% are at risk for hearing loss due to exposure to unsafe levels of sounds in recreational settings.” [WHO (2021), p. 20]. Firearm shooting represents one of the most common hazardous recreational exposures in the U.S. [e.g., Meinke (2017) and Sonstrom Malowski (2022)]. Although there are many population estimates of the prevalence of occupational exposures and some, but many fewer, population estimates for recreational or off-work noise exposures, to our knowledge there are no such estimates for combined or all-cause noise exposures in the U.S. The purpose of the analyses presented here was to provide population estimates of such all-cause noise exposures for the U.S. and to examine their effects on the odds of having audiometrically defined hearing loss.

There is evidence from U.S. population studies that the odds of having hearing loss increase when both occupational and recreational noise exposures are considered. For example, Hoffman (2017) analyzed data from 3831 adults, aged 20–69 years, included in the National Health and Nutrition Examination Survey (NHANES) 2011–2012 dataset, a nationally representative cross-sectional sample of the U.S. population. In their study, “speech-frequency” hearing loss was deemed to be present when the pure-tone average (PTA) hearing threshold level (HTL) across 500, 1000, 2000, and 4000 Hz (denoted PTA4) exceeded 25 dB HL. In addition, “high-frequency” hearing loss was deemed to be present when the PTA across 3000, 4000, and 6000 Hz, denoted PTA346k, exceeded 25 dB HL. As is typical of most such studies, Hoffman (2017) found age and sex to be the most significant contributors to increased odds for hearing impairment. Combined exposure to high noise levels at work and outside of work also significantly increased the odds of hearing loss in adults, even after adjustment of the odds ratios (ORs) for several other demographic variables. The fully adjusted ORs for combined noise exposure reported by Hoffman (2017) were significant (p < 0.05) and ranged from 2.1 to 2.4, whereas those for work noise alone were smaller and either non-significant (OR = 1.4, PTA4) or barely significant (OR = 1.5, PTA346k). Importantly, responses to the NHANES questions about the use of firearms and the use of hearing protection were not considered in the combined noise exposures evaluated by Hoffman (2017). Both factors are likely to affect the overall exposure and the consequent risk of hearing loss from that exposure.

The NHANES datasets represent a large and rich set of data on noise exposure and hearing loss. The NHANES surveys include several questions about types of noise exposure, including firearms, occupational exposures, and off-work exposures, often including graded responses that provide an estimate of the severity of those exposures. For example, those who indicated that they had used firearms in their lifetime were asked about the number of rounds fired and how frequently they used hearing protection while shooting. Similarly, those who indicated that they had worked in loud or very loud noise were then asked about the number of years worked in such noise. The NHANES surveys also appropriately operationally defined terms such as “loud” and “very loud” noise exposures, which can then be used to approximate the noise levels involved. Finally, in addition to these questions about noise exposure, about 82% had complete audiograms.

The two most recently available NHANES datasets are the 2011–2012 and 2015–2016 datasets. These datasets were used in this paper to evaluate all-cause self-reported noise exposure for U.S. adults aged 20–69 years. Further details about these datasets and analyses follow.

The NHANES 2011–2012 and 2015–2016 surveys were conducted by the National Center for Health Statistics, Centers for Disease Control and Prevention to monitor the health and nutritional status of the civilian, non-institutionalized U.S. population. Sex and race/ethnicity were self-reported using federal guidelines in existence at that time. Because the surveys enquired about the respondent's sex, either male or female, rather than gender, we refer to that variable as sex. The surveys used a complex, multistage, stratified, cluster design with oversampling of targeted subgroups to produce nationally representative estimates. For the 2011–2012 and 2015–2016 cycles, NHANES oversampled Hispanic, non-Hispanic Asian, and non-Hispanic black individuals as well as all individuals (regardless of race or ethnicity) who were at or below either 130% of the poverty index (2011–2012) or 185% of the poverty index (2015–2016).

The NHANES 2011–2012 and 2015–2016 data were collected on full samples of 19 727 individuals, 11 279 of whom were adults ranging in age from 20 to 69 years. In both surveys, more respondents gave complete responses to the interview questions about noise exposure than had complete audiograms. To maximize use of available data, the data for noise exposure were examined first and separately from the audiometric data. The questions regarding noise exposure spanned survey item numbers AUQ300 to AUQ380/381. Initially, 4675 and 4767 adults had recorded responses to the first noise-exposure question for the NHANES 2011–2012 and 2015–2016 datasets, respectively, representing 9442 adults when combined. Responses of “do not know” or “refused” were treated as missing data for the noise-exposure questions. Two cases had missing data for all the noise-exposure questions and were removed from the analyses, resulting in a final total of N = 9440. The sample sizes tabulated by sex, age group, and race/ethnicity are provided in Table I.

TABLE I.

Sex, age, and race/ethnicity information for the 9440 adults from NHANES 2011–2012 and 2015–2016 included in the analyses of the noise-exposure information.

Age decade (yr) Mexican American Other Hispanic Non-Hispanic white Non-Hispanic black Non-Hispanic asian Other Total
Males                 
  20–29  142  97  294  228  168  41  970 
  30–39  159  103  335  187  133  40  957 
  40–49  101  91  283  196  141  29  841 
  50–59  117  93  302  219  123  33  887 
  60–69  140  144  246  276  97  21  924 
  Total  659  528  1460  1106  662  164  4579 
Females                 
  20–29  143  116  296  250  149  44  998 
  30–39  156  107  328  196  158  39  984 
  40–49  151  111  288  268  153  27  998 
  50–59  123  141  289  258  123  16  950 
  60–69  140  175  255  241  98  22  931 
  Total  713  650  1456  1213  681  148  4861 
Age decade (yr) Mexican American Other Hispanic Non-Hispanic white Non-Hispanic black Non-Hispanic asian Other Total
Males                 
  20–29  142  97  294  228  168  41  970 
  30–39  159  103  335  187  133  40  957 
  40–49  101  91  283  196  141  29  841 
  50–59  117  93  302  219  123  33  887 
  60–69  140  144  246  276  97  21  924 
  Total  659  528  1460  1106  662  164  4579 
Females                 
  20–29  143  116  296  250  149  44  998 
  30–39  156  107  328  196  158  39  984 
  40–49  151  111  288  268  153  27  998 
  50–59  123  141  289  258  123  16  950 
  60–69  140  175  255  241  98  22  931 
  Total  713  650  1456  1213  681  148  4861 

Audiometric testing was conducted in sound booths (model Delta 142; Acoustic Systems, Austin, TX) in mobile examination centers, which were transported to each survey location. Ambient noise met the standards for maximum permissible ambient noise levels for ears-covered testing (ANSI, 1999). During testing, background noise was monitored continuously. Thresholds were obtained using Interacoustics (Middlefart, Denmark) AD226 audiometers calibrated to ANSI S3.6 (ANSI, 1996) specifications.

Thresholds were obtained using a pulsed-tone stimulus and a standard clinical modified method of limits procedure for each ear at 500, 1000, 2000, 3000, 4000, 6000, and 8000 Hz. The first test ear was alternated across successive participants. HTLs were usually obtained using supra-aural TDH-49P headphones (Telephonics, Farmingdale, NY) but insert earphones (EARtone 3 A; Etymotic Research, Elk Grove Village, IL) were used when participants had collapsing ear canals. Noise was not used to mask the non-test ear. However, when marked interaural asymmetry was found, the poorer ear was re-tested with insert earphones.

All those with complete audiograms were included in the analyses of the audiometric data. In some cases, especially for the highest frequencies (4000–8000 Hz), “666” entries had been recorded to designate those for whom HTLs exceeded the limits of the audiometer. As in prior analyses of NHANES data [e.g., Agrawal (2008), (2009)], these entries were recoded as missing data. The audiometric data were analyzed for 3706 and 4019 adults in the NHANES 2011–2012 and 2015–2016 datasets, respectively, for a total of 7725 adults with complete audiograms. Although insert earphones were used in a few cases, as described above, all analyses of audiometric data presented here were based on the HTLs obtained using the TDH-49P headphones.

The surveys were approved by the National Center for Health Statistics Institutional Review Board, and all participants provided written consent. The audiometry examination was conducted by trained NHANES health technicians and included otoscopy, tympanometry, and air-conduction pure-tone audiometry.

For both surveys, the NHANES household questionnaire was administered by trained interviewers in the participant's home via computer-assisted personal interview. The entire protocol, including the survey instruments and data-coding procedures, and the complete dataset are publicly available online (CDC, 2022b,c).

All prevalence estimates and statistical analyses presented here were performed in accordance with the National Health and Nutrition Examination Survey: Analytic Guidelines, 2011–2014 and 2015–2016 published online on December 14, 2018, by the National Center for Health Statistics (NCHS). Statistical analyses were performed using sas version 9.4 (sas Institute, Inc., Cary, NC). This included the use of masked strata and cluster variables for the design and the use of the adjusted sample weights provided for each participant by the NCHS. Sample weights were adjusted as recommended by dividing each dataset's 2-year weights by 2. In addition, as recommended by the NCHS, the lowest common denominator guided the choice of which set of population weights to use. In this case, because most of the data of interest were obtained from those who were examined at the mobile examination center (MEC), the MEC weights were used. The full set of 19 727 respondents was included in the population-weighted analyses with the exposure group (N = 9440) and audiogram group (N = 7725) selected via domain specification in the sas analyses.

Variance estimation was accomplished using the recommended Taylor-series linearization method. The 95% confidence intervals (CIs) were those generated by sas 9.4 based on standard errors estimated by linear interpolation of a normal distribution [i.e., standard error = [(p*q)/N]1/2, where p is the proportion of interest and q = 1 − p, and N is the number of cases]. Hoffman (2010) noted that the standard error and resulting CIs so derived were valid estimates. All prevalence estimates presented in this report met the NCHS standard of reliability (relative standard error not exceeding 30%).

Two pure-tone averages served as the dependent variables in binary logistic-regression analyses. These pure-tone averages are the same used previously by Hoffman (2010) and Hoffman (2017) in analyses of the data from earlier NHANES cycles, namely, PTA4 and PTA346k. Both PTAs were defined for the better ear and the worse ear. The better-ear PTA was established here by computing the PTA for each ear and then selecting the minimum of the two values, whereas the worse-ear PTA was the maximum PTA of the two values. For PTA4, the fence between normal and impaired hearing was 20 dB HL, consistent with the most recent WHO guidelines (Stevens , 2013; WHO, 2021), whereas the fence for PTA346k was 25 dB HL, as in Hoffman (2010) and Hoffman (2017).

In addition to including sex, age, and race/ethnicity in the fully adjusted logistic-regression solution, measures of education level, diabetes, hypertension, and smoking were included, with each of these additional hearing-loss risk factors defined as in Hoffman (2017) for the analyses of the NHANES 2011–2012 data. Specifically, as detailed by Hoffman (2017), participants were classified positive for diabetes and assigned a value of 1 (+) (0 if negative, −) if they answered “yes” to “ever been told by a doctor or other health professional that you have diabetes” or “now taking diabetic pills to lower blood sugar” or had a 2-h fasting glucose readings ≥ 200 mg/dl. Participants were classified positive for hypertension and assigned a value of 1 (+) (0 or − if negative) if they answered “yes” to “Have you ever been told you have high blood pressure?” or “Are you taking a prescription for hypertension?” or, if during the MEC exam, the average of four blood pressure measurements was > 140 mm Hg (systolic) or > 90 mmHg (diastolic). Smoking history was defined as “nonsmoker” (assigned 0) if the respondent answered “no” to “Have you smoked at least 100 cigarettes in your life?” Current and former smokers were divided into two groups, <20 pack-years (assigned 1, +) and ≥20 pack-years (assigned 2, ++). In addition to the measures defined by Hoffman (2017), the response to a yes/no question about prior military service was included.

A minimum events-per-variable ratio of 10 is frequently recommended for logistic-regression analyses to ensure stability of the solution (Harrell , 1984; Harrell , 1996; Peduzzi , 1996). With either 10 or 12 independent variables used in these analyses, the events-per-variable ratio was never below 70.

There are a total of nine questions pertaining to noise exposure in the NHANES 2011–2012 and 2015–2016 interviews. Question AUQ380/381 asked about the frequency of HPD use in the past 12 months when exposed to very loud sounds or noise, with AUQ380 specifically asking about both work and off-work exposures, whereas AUQ381 was confined to off-work exposures only, asking the respondent to exclude use of firearms from outside-work exposures. Given the differences in these two questions and that both ask only about “the past 12 months” rather than lifetime exposures as for all other noise-exposure questions, responses to AUQ380/381 were not considered further in these analyses.

There were slight changes in wording for two of the remaining questions from 2011–2012 to 2015–2016. AUQ330 in 2011–2012 and AUQ331 in 2015–2016 both enquire about exposure to loud noise at work. In both cases, the definition of “loud noise” is tied to the need to speak “in a raised voice to be heard” but only AUQ331 appropriately qualified this enquiry by adding “by someone three feet away when not using hearing protection.” Question AUQ330 did not specify the distance between the speaker and the listener, nor did it note that this question applied to situations in which hearing protectors were not worn. Despite these differences in wording, the response distributions were nearly identical (65.4% “yes” for AUQ330 and 67.1% “yes” for AUQ331) and did not differ statistically [Rao-Scott chi-square (1) = 0.50, p = 0.48]. Moreover, for those answering affirmatively, the follow-up question about the duration of such exposure (AUQ340) showed no differences in response distributions between the two datasets [Rao-Scott chi-square (6) = 1.67, p = 0.95]. Those who responded affirmatively about exposure to “loud” noise were then asked about exposure to “very loud” noise (AUQ350 and AUQ360/361). In both surveys, “very loud” noise was defined as “so loud you have to shout in order to be understood by someone standing 3 feet away from you.” The more recent version, AUQ361, was also tied to the reference condition without the use of HPDs. Despite these differences in questions, the response distributions for AUQ350 did not differ significantly across the two NHANES datasets with 35.9% (2011–2012) and 32.5% (2015–2016) indicating that they had been exposed to very loud noise at work [Rao-Scott chi-square (1) = 1.21, p = 0.27]. In addition, there were slight but non-significant differences in the distributions of the responses to the follow-up question about the duration of such exposures to very loud noise at work, AUQ360/361 [Rao-Scott chi-square (6) = 11.9, p > 0.06]. Based on these analyses, the pooling of responses for these noise-exposure questions across the two NHANES surveys was considered appropriate.

It is generally the case that, for a speaker-listener distance of 3 feet, a noise level of 80–85 dBA requires a raised voice to communicate whereas the need to shout typically does not arise until noise levels approximate 90 dBA (Miller, 1974; Robinson and Casali, 2000; Ahmed , 2004; Neitzel , 2009). Self-report ratings of noise levels using raised or shouted speech compare favorably to those measured acoustically (Neitzel , 2009; Schlaefer , 2009).

In addition to enquiries about work-related noise exposures, both NHANES surveys asked whether respondents ever used firearms for target shooting, hunting, work, or military service (AUQ300 and AUQ310), and these questions were identical across surveys. Both surveys also asked whether the respondent ever had off-work exposure to very loud noise (AUQ370), again using the need to shout to be understood by someone three feet away as the operational definition of “very loud” noise. There was no reference to use of hearing protection while trying to communicate in such off-work noise in either survey. Examples given for the off-work exposures included “power tools, lawn mowers, farm machinery, cars, trucks, motorcycles, motorboats, or loud music.”

Finally, there were two questions in each survey regarding the use of HPDs. One question, AUQ320, immediately followed the question about firearm use and asked how often HPDs (ear plugs or earmuffs) were used when shooting firearms. The other question about HPD use in the past 12 months, AUQ380/381, was dropped from the analyses for the reasons noted above.

Despite slight differences in the wording of some survey items, the responses across the two NHANES datasets were remarkably similar, with population-weighted response distributions typically differing by less than a few percentage points between surveys. As a result, all analyses reported here are based on the combined NHANES 2011–2012 and 2015–2016 datasets. The distribution of sex, age, and race/ethnicity of those in the combined dataset is shown in Table I.

Figure 1 shows the percentage of adults who reported various noise exposures, separately for males (blue bars) and females (pink bars). The prevalence of HPD use while shooting is shown in the top-right panel. Across all types of exposures, fewer females than males reported hazardous noise exposures. Females also reported more frequently using HPDs all the time (always) while shooting.

FIG. 1.

(Color online) The percentage of adults in the combined dataset (N = 9440) who reported various noise exposures and HPD use, shown separately for males (blue bars) and females (pink bars). All percentages are population estimates. Here, and in what follows, error bars show 95% CIs.

FIG. 1.

(Color online) The percentage of adults in the combined dataset (N = 9440) who reported various noise exposures and HPD use, shown separately for males (blue bars) and females (pink bars). All percentages are population estimates. Here, and in what follows, error bars show 95% CIs.

Close modal

The prevalence estimates provided in Fig. 1 should not be interpreted as estimates of those experiencing solely the exposure type noted. Many of those who reported firearm use, for example, experienced other types of noise exposure, such as work noise. To illustrate the prevalence of exposures of more than one type, the percentage of males and females experiencing either firearm or loud work-noise exposure, both exposures, or neither exposure, is shown in Fig. 2. About 8% of females and 36% of males experienced exposures both to firearms and loud work noise. Indeed, more males experienced both types of noise exposure than experienced either alone.

FIG. 2.

(Color online) The population-weighted percentages of males (blue) and females (pink) experiencing exposure to firearms only, loud work noise only, both, or neither.

FIG. 2.

(Color online) The population-weighted percentages of males (blue) and females (pink) experiencing exposure to firearms only, loud work noise only, both, or neither.

Close modal

In creating a measure of all-cause noise exposure, the goal was to capture as much of the information about noise exposure as possible, including information on the use of hearing-protection devices, while also eliminating redundant exposure information. Recall that the two questions pertaining to use of firearms, number of rounds fired and frequency of use of HPDs while shooting, made use of a five-point response scale. The population-weighted correlation between number of rounds fired and use of HPDs while shooting was r = 0.73. For the question about HPD use while shooting, responses with higher values reflect less frequent use (i.e., 1 = always and 5 = never). The positive correlation indicates that those who fired the most rounds also tended to use HPDs less frequently. Given the correlation between these two measures, a single combined measure was desired and is described below. A combined measure was also desirable in that the use of HPDs while shooting may be as critical to the severity of the exposure as the number of rounds fired.

For work noise, the questions about the length of exposure to loud and very loud noise made use of identical seven-point response scales listing exposure durations from less than 3 months (1) to 15 or more years (7). The population-weighted correlation between the length of time exposed to loud and very loud work noise was r = 0.81; those who reported longer durations of lifetime exposure to loud work noise tended also to report longer durations of lifetime exposure to very loud work noise. From review of the 7 × 7 response matrix for both questions, it was clear that for many respondents, the exposures to loud and very loud work noise were two separate exposures having different durations. On the other hand, 48.4% of those with both loud and very loud noise exposures at work reported the same duration for the two. In these cases, given the sequential nature of these questions in NHANES, it is impossible to determine whether this reflects exposure to very loud noise for a given duration or for loud noise and very loud noise each for that duration. Given these issues and the correlation between the responses for those who experienced both loud and very loud work noise exposures, a single work-noise exposure was desired that would reflect the overall work-noise exposure.

Our approach to combining the responses to questions about the number of rounds fired and use of HPDs while shooting, as well as the responses to work exposures to loud and very loud noise, was as follows. As a first step, each graded response for the noise-exposure questions was converted to a binary response by determining the cut-off response that was best for the identification of hearing loss. From the response alternatives for each question, it is not obvious where to establish the cut-off along each ordinal scale to dichotomize the responses. Here, an empirical approach was used, evaluating the responses along each scale using receiver-operating-characteristic (ROC) analyses. ROC analyses were performed on the unweighted data for each of the four dependent variables of interest here: (1) better-ear PTA4 > 20 dB HL; (2) worse-ear PTA4 > 20 dB HL; (3) better-ear PTA346k > 25 dB HL; and (4) worse-ear PTA346k > 25 dB HL. The results of those ROC analyses are shown in Table II. From those analyses, the optimal cut-off responses for the four questions were: (1) 100+ rounds fired; (2) HPD use while shooting of half the time or less; (3) exposure to loud noise at work for 5 or more years; and (4) exposure to very loud noise at work for 3 or more years. These responses were set to 1 for each question with all other responses set to 0 (including those who never had such exposures).

TABLE II.

Results of ROC analyses to determine the optimal response criterion for each graded noise-exposure variable for the identification of various PTA criteria for hearing loss using the unweighted data. The z values are for one-sided tests of area under curve (AUC) > 0.5 and all were significant (p < 0.001). The optimal response criterion was determined based on the maximum Youden Index for each ROC. The percent-correct accuracy of the identification of hearing loss at that criterion is also provided.

PTA event Noise exposure variable AUC (95% CI) z Optimal response % accuracy
Better-ear PTA4 > 20 dB HL  Number of rounds fired  0.55  4.95  100+ rounds  74.3 
(0.53–0.56) 
  Use of HPD when shooting  0.55  5.77  HPD used half time or less  75.6 
(0.53–0.57) 
  Duration of exposure to loud noise at work  0.58  8.54  5+ years of exposure  77.9 
(0.56–0.60) 
  Duration of exposure to very loud noise at work  0.57  8.70  3+ years of exposure  80.2 
(0.56–0.59) 
Worse-ear PTA4 > 20 dB HL  Number of rounds fired  0.54  5.68  100+ rounds  69.3 
(0.53–0.55) 
  Use of HPD when shooting  0.54  6.08  HPD used half time or less  70.4 
(0.53–0.56) 
  Duration of exposure to loud noise at work  0.57  10.35  5+ years of exposure  72.9 
(0.56–0.59) 
  Duration of exposure to very loud noise at work  0.56  10.18  3+ years of exposure  74.4 
(0.55–0.58) 
Better-ear PTA346k > 25 dB HL  Number of rounds fired  0.57  9.06  100+ rounds  70.6 
(0.55–0.58) 
  Use of HPD when shooting  0.57  10.13  HPD used half time or less  72.3 
(0.56–0.59) 
  Duration of exposure to loud noise at work  0.59  12.3  5+ years of exposure  74.4 
(0.58–0.60) 
  Duration of exposure to very loud noise at work  0.57  11.37  3+ years of exposure  75.4 
(0.56–0.59) 
Better-ear PTA346k > 25 dB HL  Number of rounds fired  0.55  8.84  100+ rounds  63.7 
(0.54–0.57) 
  Use of HPD when shooting  0.56  9.9  HPD used half time or less  65.5 
(0.55–0.57) 
  Duration of exposure to loud noise at work  0.59  14.28  5+ years of exposure  67.6 
(0.57–0.60) 
  Duration of exposure to very loud noise at work  0.57  12.34  3+ years of exposure  67.0 
(0.55–0.58) 
PTA event Noise exposure variable AUC (95% CI) z Optimal response % accuracy
Better-ear PTA4 > 20 dB HL  Number of rounds fired  0.55  4.95  100+ rounds  74.3 
(0.53–0.56) 
  Use of HPD when shooting  0.55  5.77  HPD used half time or less  75.6 
(0.53–0.57) 
  Duration of exposure to loud noise at work  0.58  8.54  5+ years of exposure  77.9 
(0.56–0.60) 
  Duration of exposure to very loud noise at work  0.57  8.70  3+ years of exposure  80.2 
(0.56–0.59) 
Worse-ear PTA4 > 20 dB HL  Number of rounds fired  0.54  5.68  100+ rounds  69.3 
(0.53–0.55) 
  Use of HPD when shooting  0.54  6.08  HPD used half time or less  70.4 
(0.53–0.56) 
  Duration of exposure to loud noise at work  0.57  10.35  5+ years of exposure  72.9 
(0.56–0.59) 
  Duration of exposure to very loud noise at work  0.56  10.18  3+ years of exposure  74.4 
(0.55–0.58) 
Better-ear PTA346k > 25 dB HL  Number of rounds fired  0.57  9.06  100+ rounds  70.6 
(0.55–0.58) 
  Use of HPD when shooting  0.57  10.13  HPD used half time or less  72.3 
(0.56–0.59) 
  Duration of exposure to loud noise at work  0.59  12.3  5+ years of exposure  74.4 
(0.58–0.60) 
  Duration of exposure to very loud noise at work  0.57  11.37  3+ years of exposure  75.4 
(0.56–0.59) 
Better-ear PTA346k > 25 dB HL  Number of rounds fired  0.55  8.84  100+ rounds  63.7 
(0.54–0.57) 
  Use of HPD when shooting  0.56  9.9  HPD used half time or less  65.5 
(0.55–0.57) 
  Duration of exposure to loud noise at work  0.59  14.28  5+ years of exposure  67.6 
(0.57–0.60) 
  Duration of exposure to very loud noise at work  0.57  12.34  3+ years of exposure  67.0 
(0.55–0.58) 

Next, given the correlations noted above for the responses to each pair of questions, the pair of questions for firearm exposure (number of rounds and frequency of HPD use) and the pair for work noise (years exposed to loud noise and to very loud noise) were combined. Specifically, the binary values for each correlated pair were combined to form a single measure for each type of noise exposure, firearm use and overall work-noise exposures. Figure 3 shows the percentage of males (blue bars) and females (pink bars) with exposures categorized based on the two binary measures of firearm exposure (top) and on the two binary measures of work-noise exposure (bottom). Not surprisingly, based on the response distributions presented in Fig. 1, males reported more combined exposures than females. About 24% of males but only 3% of females both fired 100+ rounds and used HPDs less than half of the time. These individuals were considered at greatest risk for hearing loss from the use of firearms and were assigned a value of 1 for the Firearm/HPD noise-exposure type. All others were assigned a value of 0, including those who never used firearms. Similarly, about 18% of males and 4% of females reported experiencing both loud noise for 5+ years and very loud noise for 3+ years at work. These individuals were considered at greatest risk for hearing loss from work-noise exposures and were assigned a value of 1 for the Work Noise noise-exposure type. All others, were assigned a value of 0, including those who reported never experiencing loud or very loud work noise or reported never having worked.

FIG. 3.

(Color online) Population-weighted prevalence of various noise exposure types and combinations for males (blue bars) and females (pink bars). Values of “1” and “0” along the x axis signify the presence or absence, respectively, of each noise-exposure condition as listed at the right of the x axis.

FIG. 3.

(Color online) Population-weighted prevalence of various noise exposure types and combinations for males (blue bars) and females (pink bars). Values of “1” and “0” along the x axis signify the presence or absence, respectively, of each noise-exposure condition as listed at the right of the x axis.

Close modal

With one exception, preliminary population-weighted logistic-regression analyses with better-ear and worse-ear PTA values as dependent variables indicated that it was only the combined conditions (“1 1” at far right) in the top and bottom panels that significantly increased the odds for hearing loss. The lone exception occurred for PTA346k > 25 dB HL in the worse ear for work-noise exposure to loud noise for 5+ years. This exposure significantly increased the odds for hearing loss by itself. Nonetheless, only those Firearm/HPD and Work Noise exposures that were positive for both conditions (“1 1”) were considered here to be positive noise exposures of each type.

In the end, three variables were used to capture self-reported all-cause noise exposure. In addition to the Firearm/HPD and Work Noise noise-exposure types described above, responses to the question about off-work exposure to very loud noise were included. This noise-exposure type is referred to here as Off-Work Noise. All three noise-exposure types were binary variables with values of 1 indicating that the respondent had experienced that type of exposure, as defined above, in his or her lifetime.

Given the three binary noise-exposure types, there are eight possible exposure scenarios: no exposure to any of the three exposure types, three involving one type, three involving various pairs of types, and one involving all three types. The prevalence of each of these noise-exposure scenarios is shown in Fig. 4. About 55% of males and 85% of females did not experience any of the exposure types. Noise exposure for each type alone and for each combination of types was greater for males than for females. About 15% of males and 1% of females reported experiencing various combined exposures to noise.

FIG. 4.

(Color online) Population-weighted prevalence of various noise exposure types and combinations for males (blue bars) and females (pink bars). Values of “1” and “0” along the x axis signify the presence or absence, respectively, of each of the three noise-exposure types listed at the right of the x axis.

FIG. 4.

(Color online) Population-weighted prevalence of various noise exposure types and combinations for males (blue bars) and females (pink bars). Values of “1” and “0” along the x axis signify the presence or absence, respectively, of each of the three noise-exposure types listed at the right of the x axis.

Close modal

Importantly, the noise-exposure types shown in Fig. 4 are mutually exclusive. For example, Off-Work Noise exposure (“0 0 1”) means that this was the only exposure experienced. If one sums the prevalence estimates for all four noise-exposure types in Fig. 4 that include Off-Work Noise, the total is 21.0% for males and 7.9% for females, the same values as shown in the bottom panel of Fig. 1.

Logistic-regression analyses were performed to examine the effect of each noise-exposure variable on the odds of having hearing loss. These analyses were performed separately for each of the four definitions of hearing loss: better-ear and worse-ear PTA4 > 20 dB HL and better-ear and worse-ear PTA346k > 25 dB HL. All four analyses were repeated several times to examine the effects of Firearms/HPD, Work Noise, and Off-Work Noise, as in previous analyses of NHANES data [e.g., Hoffman (2010) and Hoffman (2017)] and for all seven noise-exposure scenarios shown in Fig. 4, which include separate exposure types, paired exposure types, and all three types combined. Finally, in some cases both fully adjusted and unadjusted logistic-regression analyses were performed. The fully adjusted model (adjusted ORs) included covariates of sex, race/ethnicity, education level, diabetes, hypertension, and smoking, as in Hoffman (2017). These covariates were selected by Hoffman (2017) based on research suggesting that these variables were related to hearing loss. Prior military service and annual income were also included as covariates here. The ORs and 95% CIs for all the variables in the fully adjusted model are provided in the supplemental material.1

Figure 5 shows the ORs and 95% CIs for the fully adjusted logistic-regression model with better-ear (black circles) and worse-ear (gray circles) PTA4 > 20 dB HL as the event predicted. The type-3 analyses of main effects for each variable, as well as the ORs and 95% CIs in tabular form, are provided in the supplemental material1 (Tables S1 and S2). Although there were small differences between the solutions for the better-ear and worse-ear PTA4, the overall pattern of ORs across all variables was quite similar. The effects observed are consistent with the analyses of the NHANES 2011–2012 dataset by Hoffman (2017) using PTA4 > 25 dB HL as the indicator of the presence of hearing loss. Hearing loss was most strongly associated with age and sex, older adults and males having higher odds of having hearing loss than younger adults and females. For race/ethnicity, the well-established difference between Non-Hispanic Blacks (NHB) and Non-Hispanic Whites (NHW) was also apparent, as were the smaller but significant effects of level of education and diabetes. Finally, turning to the noise-exposure measures shown at the bottom of Fig. 5, significant effects were found for the better-ear and worse-ear PTA4 for Work Noise and Off-Work Noise, either of these exposures increasing the odds of hearing loss by about 50%–60%. Firearms/HPD did not significantly increase the odds for hearing loss, perhaps because the effects of this occur mainly at high frequencies (Humes , 2006; Moore, 2020).

FIG. 5.

ORs and 95% CIs for the better-ear (black circles) and worse-ear (gray circles) for the PTA4 > 20 dB HL criterion for hearing loss.

FIG. 5.

ORs and 95% CIs for the better-ear (black circles) and worse-ear (gray circles) for the PTA4 > 20 dB HL criterion for hearing loss.

Close modal

Similar analyses were performed for PTA346k exceeding 25 dB HL for the better and worse ears. Figure 6 presents the results of the fully adjusted logistic-regression models for this case. The type-3 analyses of main effects for each variable, as well as the ORs and 95% CIs in tabular form are provided in the supplemental material1 (Tables S1 and S2). The significant effects of sex, age, race/ethnicity, level of education, and diabetes on the odds of having hearing loss in either ear were nearly identical to those for PTA4 (Fig. 5). For PTA346k, the effect of income was also significant, those making $75 000/year or more having lower odds of having hearing loss than those making less than $20 000/year. Significant effects of noise exposure were confined to the worse ear and were found for Firearms/HPD and Work Noise. Each increased the odds of having hearing loss by about about 40%–50%. Off-Work Noise did not significantly affect the odds of having hearing loss for either ear.

FIG. 6.

ORs and 95% CIs for the better-ear (black circles) and worse-ear (gray circles) for the PTA346k > 25 dB HL criterion for hearing loss.

FIG. 6.

ORs and 95% CIs for the better-ear (black circles) and worse-ear (gray circles) for the PTA346k > 25 dB HL criterion for hearing loss.

Close modal

The logistic-regression analyses presented in Figs. 5 and 6, for the most part follow the prevailing approach to analyzing the effects of noise exposure on hearing loss and do not indicate the separate effects of each exposure type. For example, those with Off-Work Noise exposure, may also have experienced Work Noise, Firearm/HPD, or all three types of exposure. As a result, it would be erroneous to conclude from Fig. 5 that Firearms/HPD or Work Noise per se increased the odds of PTA346k in the worse ear exceeding 25 dB HL. Rather, one could only conclude that Firearms/HPD or Work Noise exposures alone or in combination with other noise exposures increased the odds of hearing loss.

To better examine the role of each noise-exposure type alone, as well as various combinations of noise-exposure types, the logistic-regression analyses were repeated for the seven positive noise-exposure types shown in Fig. 4 with no-noise-exposure (“0 0 0” in Fig. 4) serving as the reference group. Figure 7 presents the unadjusted ORs for the better (black circles) and worse (gray circles) ears for PTA4 > 20 dB HL (top) and PTA346k > 25 dB HL (bottom). Tables of the type-3 main effects and the ORs are provided in supplemental material1 (Tables S3 and S4). For both PTA criteria, the unadjusted ORs in Fig. 7 indicate that Off-Work Noise only was the sole noise-exposure type that failed to have a significant effect for at least one of the four definitions of hearing loss.

FIG. 7.

Unadjusted ORs and 95% CIs for each of the seven noise-exposure types (Fig. 4) with the no-noise-exposure group as the reference. The top half shows results for PTA4 > 20 dB HL and bottom half for PTA346k > 25 dB HL.

FIG. 7.

Unadjusted ORs and 95% CIs for each of the seven noise-exposure types (Fig. 4) with the no-noise-exposure group as the reference. The top half shows results for PTA4 > 20 dB HL and bottom half for PTA346k > 25 dB HL.

Close modal
TABLE III.

Population-estimated prevalence of various noise-exposure types, individually and in combination, for 9440 adults (4579 males and 4861 females). The prevalence estimates were applied to the population-weighted totals for males (110.1 × 106) and females (119.2 × 106) aged 20–69 years to generate the number of U.S. adults experiencing various noise-exposure types.

Noise-exposure type Male % ± 95% CI Male (millions) ± 95% CI Female % ± 95% CI Female (millions) ± 95% CI
No Noise  54.1  1.5  59.5  1.7  85.8  1.0  102.2  1.2 
Firearm/HPD only  13.5  1.0  14.9  1.1  2.8  0.5  3.3  0.6 
Work Noise only  7.6  0.8  8.4  0.9  3.4  0.5  4.1  0.6 
Off-Work Noise only  10  0.9  11.0  1.0  6.7  0.7  8.0  0.9 
Work Noise and Firearm/HPD  3.7  0.6  4.1  0.6  —  —  —  — 
Work Noise and Off-Work Noise  3.6  0.6  4.0  0.6  0.8  0.3  1.0  0.3 
Firearms/HPD and Off-Work Noise  4.3  0.6  4.7  0.7  0.4  0.2  0.5  0.2 
All 3 noise-exposure types  3.1  0.5  3.4  0.6  —  —  —  — 
Noise-exposure type Male % ± 95% CI Male (millions) ± 95% CI Female % ± 95% CI Female (millions) ± 95% CI
No Noise  54.1  1.5  59.5  1.7  85.8  1.0  102.2  1.2 
Firearm/HPD only  13.5  1.0  14.9  1.1  2.8  0.5  3.3  0.6 
Work Noise only  7.6  0.8  8.4  0.9  3.4  0.5  4.1  0.6 
Off-Work Noise only  10  0.9  11.0  1.0  6.7  0.7  8.0  0.9 
Work Noise and Firearm/HPD  3.7  0.6  4.1  0.6  —  —  —  — 
Work Noise and Off-Work Noise  3.6  0.6  4.0  0.6  0.8  0.3  1.0  0.3 
Firearms/HPD and Off-Work Noise  4.3  0.6  4.7  0.7  0.4  0.2  0.5  0.2 
All 3 noise-exposure types  3.1  0.5  3.4  0.6  —  —  —  — 

The results of the fully adjusted logistic-regression analyses for PTA4 > 20 dB HL are shown in Fig. 8. Tables of type-3 analyses of main effects and for each fully adjusted model are presented in the supplemental material1 (Tables S5 and S6). As in Fig. 5, large effects of age are apparent with smaller but significant effects of sex, race/ethnicity, level of education, and diabetes. Of greatest interest here, however, are the effects of noise-exposure type shown at the bottom of Fig. 8. None of the three noise-exposure types representing the effects of Firearms/HPD, Work Noise, or Off-Work Noise alone were significant. Only Work Noise combined with Off-Work Noise or all three noise-exposure types combined had significant effects on hearing loss, with these effects confined primarily to the worse ear. The ORs for the worse-ear PTA4 indicate that those who experienced combined Work Noise and Off-Work Noise exposure types had odds for hearing loss that were three times greater than those with no noise exposure. For the worse ear and all three noise-exposure types combined, the odds for hearing loss based on PTA4 > 20 dB HL were about eight times greater than for those with no noise exposure.

FIG. 8.

Fully adjusted ORs and 95% CIs showing effects of each noise type (bottom seven variables) on odds for PTA4 > 20 dB HL.

FIG. 8.

Fully adjusted ORs and 95% CIs showing effects of each noise type (bottom seven variables) on odds for PTA4 > 20 dB HL.

Close modal

The results of the fully adjusted logistic-regression analyses for PTA346k > 25 dB HL are shown in Fig. 9. Tables of type-3 analyses of main effects and tabular presentation of ORs for each fully adjusted model are presented in the supplemental material1 (Tables S5 and S6). As in Fig. 6, large effects of age were apparent with smaller significant effects of sex, race/ethnicity, level of education, and diabetes. The effects of noise-exposure type on the odds of hearing loss shown at the bottom of Fig. 9 indicate that it is primarily Firearms/HPD noise exposure that significantly increased the odds of hearing loss for PTA346k > 25 dB HL and usually just in the worse ear. For the worse-ear PTA346k, those whose noise exposure was confined to only Firearms/HPD had odds for hearing loss that were 1.44 times greater than for those with no noise exposure. The odds for hearing loss in the worse ear for PTA346k increased to 1.66 relative to no noise exposure when Firearms/HPD and Off-Work Noise were combined. Finally, when all three noise-exposure types (Firearms/HPD, Work Noise, Off-Work Noise) were combined, the odds for hearing loss increased to 5.19 and 4.54 times that of the no-noise group for the worse ear and better ears, respectively.

FIG. 9.

Fully adjusted ORs and 95% CIs showing effects of each noise type (bottom seven variables) on odds for PTA346k > 25 dB HL.

FIG. 9.

Fully adjusted ORs and 95% CIs showing effects of each noise type (bottom seven variables) on odds for PTA346k > 25 dB HL.

Close modal

The foregoing analyses provide the most recent estimates of occupational and recreational noise exposures from self-report surveys for U.S. adults. Evaluation of self-reported exposures to “recreational” noise was confined to those using firearms and those being exposed to noise outside of work. The question about exposure to firearm noise, however, included “for your job or in military service” along with “for target shooting” and “hunting.” Only the latter two would be considered “recreational” exposures. In contrast, the question about off-work noise exposures specifically enquires about exposures to very loud noise “outside of a job” and lists recreational noise sources, specifically “power tools, lawn mowers, farm machinery, cars, trucks, motorcycles, motor boats, or loud music.”

Among the various types of noise exposure, use of firearms had the highest prevalence, 63.7% (95% CI: 59.4.0%, 68.1%) of males and 30.9% (27.0%, 34.7%) of females having used firearms during their lifetime (Fig. 1). About 1/3 of those using firearms never used HPDs (Fig. 1). For males, the prevalence of never using HPDs among those who had used firearms was 33.8% (95% CI: 30.4%, 37.1%) while for females it was 36.9% (95% CI: 32.8%, 41.0%).

Exposure to noise at work was the next most prevalent exposure type among both males and females (Fig. 1), 47.6% (95% CI: 44.1%, 51.1%) and 20.4% (95% CI: 18.2%, 22.6%) of males and females, respectively, experiencing exposure to loud work noise. Off-work noise was the least prevalent exposure type. Even here, 21.0% (95% CI: 18.8%, 23.2%) of males and 7.9% (95% CI: 6.8%, 9.1%) of females experienced such exposures (Fig. 1).

There were strong and significant correlations between the number of rounds fired and the use of HPDs while shooting; those who reported shooting the most rounds were more likely to report seldom or never using HPDs while shooting. Likewise, the duration of work exposure to loud noise was correlated with the duration of work exposure to very loud noise among those who experienced both exposure levels. In addition to being potentially redundant, the collinearity among these two pairs of variables is undesirable statistically in the logistic-regression analyses. In such cases, it is common to combine the correlated variables in some fashion. This was done here, after first optimizing the cut-off response for the detection of hearing loss for each noise-exposure type, by considering only those with positive noise exposures for both variables to be positive for a particular noise-exposure type. Thus, only those with both 100+ rounds fired in a lifetime and use of HPDs while shooting less than or equal to half the time were considered positive for the Firearms/HPD noise-exposure type. Likewise, only those who reported both loud work noise for 5+ years and very loud work noise for 3+ years were considered positive for the Work Noise noise-exposure type. The results of the fully adjusted logistic-regression models for PTA4 and PTA346k, better-ear and worse-ear, shown in Figs. 5 and 6, demonstrated that experiencing the noise-exposure types defined here significantly increased the odds of having hearing loss. Whereas Work Noise and Off-Work Noise had the greatest influence for PTA4 (Fig. 5), Work Noise and Firearm/HPD exposure significantly increased the odds of hearing loss at higher frequencies (PTA346k), especially for the worse ear (Fig. 6). The latter observation on the greater sensitivity of PTA346k than PTA4 to the effects of firearm noise is consistent with the tendency of firearm exposure to affect hearing at and above 3000 Hz [e.g., Humes (2006), Meinke (2017), and Moore (2020)].

The logistic-regression analyses whose results are presented in Figs. 5 and 6 followed the conventional approach to examining the effects of noise exposure on the odds of having hearing loss [e.g., Hoffman (2010) and Hoffman (2017)]. For example, all those who answered affirmatively to the question about exposure to very loud off-work noise were considered as positive for Off-Work Noise. However, many who answered affirmatively to this question also had exposures to Work Noise, Firearm/HPD, or both (Fig. 4). The same applies to categorization of the Work Noise and Firearm/HPD noise-exposure types and the interpretation of the results in Figs. 5 and 6, as these exposures also often reflected combined exposures (Fig. 4). A substantial percentage of males and females experienced one or more of the noise-exposure types (Fig. 4). Further, nearly 15% of males experienced combinations of two or three noise-exposure types.

Importantly, as shown in Figs. 8 and 9, combined noise-exposure types significantly increased the odds of hearing loss, regardless of hearing-loss definition, and this was especially true for the worse-ear PTAs. Those who were positive for the Firearms/HPD, Work Noise, and Off-Work Noise noise-exposure types combined had odds for hearing loss that were 4–6 times higher than for those with no noise exposure. In only one case, PTA346k for the worse ear and Firearms/HPD exposure type (Fig. 9), did a given noise-exposure type alone increase the odds of having hearing loss.

Ideally, the dataset used to formulate a particular metric of interest, such as the noise exposure types described here, should be different from the dataset used to evaluate that metric. Not doing this can bias the evaluation in favor of the adequacy of the metric. To evaluate whether this was the case here, the two independent NHANES datasets, 2011–2012 and 2015–2016, were evaluated separately to see if similar noise-exposure types would be derived. As noted in the results, there were no significant differences in the response distributions across surveys for the various firearm-related and work-noise-related questions, despite slight differences in the wording of the noise-exposure questions across surveys. When correlations were examined among the four noise-exposure questions with ordinal responses, the correlations for the two firearm-related questions (AUQ310 and AUQ320) were identical (r = 0.73) for the two datasets. Further, the correlations for the two work-noise-related questions with ordinal responses (AUQ340 and AUQ360/361) were nearly identical with r = 0.79 for the 2011–2012 dataset and r = 0.82 for the 2015–2016 dataset. Finally, when ROC analyses were performed separately for each NHANES dataset and for each of the four noise-exposure variables having ordinal responses, a total of eight ROC analyses for each noise-exposure variable (4 PTAs as dependent variables × 2 datasets), the cut-off values were typically the same as those shown in Table II for the combined NHANES dataset. When the ROC analyses did not yield the same optimal cut-off criterion for a given noise exposure and PTA, the cut-offs for the separate analyses differed by no more than one response from those shown in Table II for the pooled data. The resulting differences in accuracy between the cut-off criteria were minimal (≤3% points). Overall, it seems unlikely that the results reported here for the combined dataset were biased by deriving and evaluating the noise-exposure types from the same dataset. Nonetheless, such bias remains possible and should be considered a possible limitation of the present findings.

Population-weighted prevalence estimates for each of the eight noise-exposure types considered here, seven positive types and one negative (no noise exposure), are presented in Table III for the 9440 adults with self-reported noise-exposure information available. The eight noise-exposure types are mutually exclusive, summing to 100% (99.9% here due to rounding). Table III also shows the population estimates in millions based on 110.0 × 106 males and 119.2 × 106 females ranging in age from 20 to 69 years for the combined NHANES dataset. As shown in Figs. 8 and 9, it was mainly the worse-ear PTAs that showed significant effects of noise-exposure type on the odds of having hearing loss. For PTA346k > 25 dB HL, Firearm/HPD significantly increased the odds of hearing loss in the worse ear when acting alone, in combination with Off-Work Noise, and in combination with Off-Work and Work Noise. Based on the prevalence estimates and counts in Table III, 23.0 × 106 males and 4.3 × 106 females experienced these noise-exposure types and had greater odds of PTA346k exceeding 25 dB HL in the worse ear than those without self-reported noise exposure. Similarly, for PTA4 > 20 dB HL, Work Noise significantly increased the odds of hearing loss in the worse ear when combined with Off-Work Noise or Off-Work Noise and Firearms/HPD exposures. Based on the prevalence estimates and counts in Table III, 7.4 × 106 males and 1.0 × 106 females experienced these noise-exposure types and had greater odds of PTA4 exceeding 20 dB HL in the worse ear than those without self-reported noise exposure. Combined, this represents a total of four out of seven noise-exposure types that significantly increased the odds of hearing loss for either PTA for the worse ear for 27 × 106 males and 2.8 × 106 females.

Generally, hearing loss defined based on the worse-ear PTA was more sensitive to the effects of combined noise-exposure types than hearing loss based on the better-ear PTA. Further, for PTA346k > 25 dB HL, three of the four noise-exposure types that included the Firearms/HPD noise-exposure type significantly increased the odds of hearing loss. These findings are as expected if it was the case that the worse ear was usually the ear in closer proximity to the end of the barrel, increasing the risk for that ear [e.g., Keim (1969) and Meinke (2017)]. Population estimates from the combined NHANES dataset indicated that for 62.7% (95% CI: 61.1%, 64.3%) of males and 56.1% (95% CI: 54.5%, 57.7%) of females the worse ear was the left ear, which is usually the most exposed ear when a rifle is fired from the right shoulder (Keim, 1969).

Although the focus here has been on the impact of noise exposure on hearing loss, many other variables increased the odds of hearing loss, as shown in Figs. 5, 6, 8, and 9. Regardless of which PTA, which ear, and which categorization of noise exposures was used, age and sex had the strongest effects on the odds of having hearing loss, as has been observed in analyses of NHANES data many times (Agrawal , 2008, 2009; Hoffman , 2010, Hoffman , 2017). Likewise, the effects of race/ethnicity, level of education, and diabetes were consistent with prior analyses of NHANES data. Generally, the effects of noise exposure illustrated in Figs. 5 and 6 were somewhat stronger than reported previously for each noise type, likely due to the way in which the noise-exposure types were defined here. In addition, the results of the logistic-regression analyses shown in Figs. 8 and 9 indicated that only the Firearms/HPD exposure type increased the odds of hearing loss when acting alone. All other significant effects, usually for the worse ear only, were observed for three of the four combined-exposure scenarios among the seven noise-exposure scenarios. Occupational noise, defined here as Work Noise, only significantly increased the odds of hearing loss for the worse ear and only when combined with recreational noise exposures (Off-Work or Off-Work and Firearms/HPD).

The impact of noise-exposure type, as defined here, was apparent for all four definitions of hearing loss: better-ear and worse-ear PTA4 > 20 dB HL and PTA346k > 25 dB HL. Changing the definition of hearing loss is likely to affect the results, as shown here and many times previously. There has been renewed interest in recent years in “sub-clinical” hearing loss or “hidden hearing loss” [see review by Kohrman (2020)]. One way this has been explored is through even more stringent audiometric definitions of hearing loss, such as defining normal hearing as all thresholds from 500 to 8000 Hz being less than 15, 20, or 25 dB HL (e.g., Ridley 2018; Spankovich , 2018; Cruickshanks , 2020; Humes, 2021). Of course, the effects of noise-exposure types identified here may not apply to other definitions of normal and impaired hearing, including those targeting “sub-clinical” hearing loss.

The overall prevalence of hearing loss varies markedly with the definition of hearing loss. For better-ear and worse-ear PTA4, for example, the observed prevalence values for the combined NHANES 2011‐2012 and 2015–2016 datasets were 11.9% (95% CI:10.5–13.3%) and 21.6% (95% CI: 19.6–23.6%), respectively. For better-ear and worse-ear PTA346k, on the other hand, the observed prevalence values were 21.3% (95% CI: 19.5–23.1%) and 34.1% (95% CI: 31.5–36.6%), respectively. Clearly, as has been noted many times previously, the prevalence of “hearing loss” depends critically on the audiometric definition of that condition.

The association between the presence of hearing loss, defined as described above, and noise exposure relied on logistic-regression analyses and the use of ORs, as is the case for many previous studies of this type. Such analyses, however, are not without limitations [e.g., Norton and Dowd (2018)]. The ORs that emerge from such analyses diverge increasingly from an interpretation in terms of probabilities or relative risk as the prevalence of the dependent variable exceeds 10% [e.g., McNutt (2003)] with some suggesting that ORs should not be used at all when the prevalence is greater than 20%–30% (Altman , 1998). The prevalence of hearing loss noted above ranged from about 12% to 34%, depending on the definition, and the use of relative risk ratios or average marginal effects may be more appropriate in future analyses of noise exposure and hearing loss [e.g., Norton and Dowd (2018)].

This work was supported, in part, by the Medical Research Council (UK, Grant No. G0701870) and the High-Performance Computing facility at Indiana University. We thank the reviewers for helpful comments on earlier versions of this paper.

1

See supplementary material at https://www.scitation.org/doi/suppl/10.1121/10.0016552 for tables of type-3 analyses and ORs for each fully adjusted logistic-regression model.

1.
Agrawal
,
Y.
,
Platz
,
E. A.
, and
Niparko
,
J. K.
(
2008
). “
Prevalence of hearing loss and differences by demographic characteristics among US adults: Data from the National Health and Nutrition Examination Survey, 1999–2004
,”
Arch. Intern. Med.
168
,
1522
1530
.
2.
Agrawal
,
Y.
,
Platz
,
E. A.
, and
Niparko
,
J. K.
(
2009
). “
Risk factors for hearing loss in US adults: Data from the National Health and Nutrition Examination Survey, 1999 to 2002
,”
Otol. Neurotol.
30
,
139
145
.
3.
Ahmed
,
H. O.
,
Dennis
,
J. H.
, and
Ballal
,
S. G.
(
2004
). “
The accuracy of self-reported high noise exposure level and hearing loss in a working population in Eastern Saudi Arabia
,”
Int. J. Hyg. Environ. Health
207
,
227
234
.
4.
Altman
,
D. G.
,
Deeks
,
J. J.
, and
Sackett
,
D. L.
(
1998
). “
Odds ratios should be avoided when events are common
,”
Brit. Med. J
317
,
1318
.
5.
ANSI
(
1996
). S3.6-1996,
Specifications for Audiometers
(
American National Standards Institute
,
New York
).
6.
ANSI
(
1999
). S3.1-1999,
Maximum Permissible Ambient Noise Levels for Audiometric Test Rooms
(
American National Standards Institute
,
New York
).
7.
CDC
(
2022a
). https://www.cdc.gov/ncbddd/hearingloss/noise.html (Last viewed December 8, 2022).
10.
Cruickshanks
,
K. J.
,
Nondahl
,
D. M.
,
Fischer
,
M. E.
,
Schubert
,
C. R.
, and
Tweed
,
T. S.
(
2020
). “
A novel method for classifying hearing impairment in epidemiological studies of aging: The Wisconsin Age-Related Hearing Impairment Classification Scale
,”
Am. J. Audiol.
29
,
59
67
.
11.
Harrell
,
F. E.
, Jr.
,
Lee
,
K. L.
,
Califf
,
R. M.
,
Pryor
,
D. B.
, and
Rosati
,
R. A.
(
1984
). “
Regression modelling strategies for improved prognostic prediction
,”
Stat. Med.
3
,
143
152
.
12.
Harrell
,
F. E.
, Jr.
,
Lee
,
K. L.
, and
Mark
,
D. B.
(
1996
). “
Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors
,”
Statist. Med.
15
,
361
387
.
13.
Hoffman
,
H. J.
,
Dobie
,
R. A.
,
Ko
,
C. W.
,
Themann
,
C. L.
, and
Murphy
,
W. L.
(
2010
). “
Americans hear as well or better today compared with 40 years ago: Hearing threshold levels in the unscreened adult population of the United States, 1959–1962 and 1999–2004
,”
Ear Hear.
31
,
725
734
.
14.
Hoffman
,
H. J.
,
Dobie
,
R. A.
,
Losonczy
,
K. G.
,
Themann
,
C. L.
, and
Flamme
,
G. A.
(
2017
). “
Declining prevalence of hearing loss in US adults aged 20 to 69 years
,”
JAMA Otolaryngol. Head Neck Surg.
143
,
274
285
.
15.
Humes
,
L. E.
(
2021
). “
Further evaluation and application of the Wisconsin age-related hearing impairment classification system
,”
Am. J. Audiol.
30
,
359
375
.
16.
Humes
,
L. E.
,
Joellenbeck
,
L. M.
, and
Durch
,
J. S.
(
2006
).
Noise and Military Service: Implications for Hearing Loss and Tinnitus
(
National Academies Press
,
New York
), p.
338
.
17.
Keim
,
R. J.
(
1969
). “
Sensorineural hearing loss associated with firearms
,”
Arch. Otolaryngol.
90
,
581
584
.
18.
Kohrman
,
D.
,
Wan
,
G.
,
Cassinotti
,
L.
, and
Corfas
,
G.
(
2020
). “
Hidden hearing loss: A disorder with multiple etiologies and mechanisms
,”
Cold Spring Harbor Perspect. Med.
10
,
a035493
.
19.
McNutt
,
L. A.
,
Wu
,
C.
,
Xue
,
X.
, and
Hafner
,
J. P.
(
2003
). “
Estimating the relative risk in cohort studies and clinical trials of common outcomes
,”
Am. J. Epidem.
157
,
940
943
.
20.
Meinke
,
D. K.
,
Finan
,
D. S.
,
Flamme
,
G. A.
,
Murphy
,
W. J.
,
Stewart
,
M.
,
Lankford
,
J. E.
, and
Tasko
,
S.
(
2017
). “
Prevention of noise-induced hearing loss from recreational firearms
,”
Sem. Hear.
38
,
267
281
.
21.
Miller
,
J. D.
(
1974
). “
Effects of noise on people
,”
J. Acoust. Soc. Am.
56
,
729
764
.
22.
Moore
,
B. C. J.
(
2020
). “
Diagnosis and quantification of military noise-induced hearing loss
,”
J. Acoust. Soc. Am.
148
,
884
894
.
23.
Neitzel
,
R. L.
,
Daniell
,
W.
,
Sheppard
,
L.
,
Davies
,
H.
, and
Seixas
,
N.
(
2009
). “
Comparison of perceived and quantitative measures of occupational noise exposure
,”
Ann. Occup. Hyg.
53
,
41
54
.
24.
Norton
,
E. C.
, and
Dowd
,
B. E.
(
2018
). “
Log odds and the interpretation of logit models
,”
Health Serv. Res.
53
,
859
878
.
25.
Peduzzi
,
P.
,
Concato
,
J.
,
Kemper
,
E.
,
Holford
,
T. R.
, and
Feinstein
,
A. R.
(
1996
). “
A simulation study of the number of events per variable in logistic regression analysis
,”
J. Clin. Epidemiol.
49
,
1371
1379
.
26.
Ridley
,
C. L.
,
Kopun
,
J. G.
,
Neely
,
S. T.
,
Gorga
,
M. P.
, and
Rasetshwane
,
D. M.
(
2018
). “
Using thresholds in noise to identify hidden hearing loss in humans
,”
Ear Hear.
39
,
829
844
.
27.
Robinson
,
G. S.
, and
Casali
,
J. G.
(
2000
). “
Speech communications and signal detection in noise
,” in
The Noise Manual
,
5th ed
., edited by
E.
Berger
,
L.
Royster
,
J.
Royster
,
D.
Driscoll
, and
M.
Layne
(
American Industrial Hygiene Association
,
Fairfax, VA
), pp.
567
600
.
28.
Schlaefer
,
K.
,
Schlehofer
,
B.
, and
Schuz
,
J.
(
2009
). “
Validity of self-reported occupational noise exposure
,”
Eur. J. Epidemiol.
24
,
469
475
.
29.
Sonstrom Malowski
,
K.
,
Gollihugh
,
L. H.
,
Malyuk
,
H.
, and
Le Prell
,
C. G.
(
2022
). “
Auditory changes following firearm noise exposure, a review
,”
J. Acoust. Soc. Am.
151
,
1769
1791
.
30.
Spankovich
,
C.
,
Gonzalez
,
V. B.
,
Su
,
D.
, and
Bishop
,
C. E.
(
2018
). “
Self-reported hearing difficulty, tinnitus, and normal audiometric thresholds, the National Health and Nutrition Examination Survey 1999–2002
,”
Hear. Res.
358
,
30
36
.
31.
Stevens
,
G.
,
Flaxman
,
S.
,
Brunskill
,
E.
,
Mascarenhas
,
M.
,
Mathers
,
C. D.
, and
Finucane
,
M.
(
2013
). “
Global and regional hearing impairment prevalence: An analysis of 42 studies in 29 countries
,”
Euro. J. Pub. Health
23
,
146
152
.
32.
World Health Organization
(
2021
).
World Report on Hearing
(
World Health Organization
,
Geneva
).

Supplementary Material