Risk of noise-induced hearing loss due to recreational sound: Review and recommendations

Noise exposure is one of the most frequent hazards present in the workplace, and several studies have estimated that approximately 22 × 10⁶ workers are exposed to potentially hazardous levels of noise in the United States (U.S.) alone (Kerns et al., 2018; Tak et al., 2009). Although regulations (OSHA, 1983) and recommended standards (ACGIH, 2006; NIOSH, 1998) have been in place for decades, the prevalence of noise overexposures and NIHL remains high in the U.S. and globally (Daniell et al., 2006; Nelson et al., 2005). NIHL is one of the most common occupational illnesses in the United States and Europe, with workers in manufacturing, construction, and the military at especially high risk for hearing loss (Alamgir et al., 2016; Masterson et al., 2014; Masterson et al., 2016; NIOSH, 2010; Schneider, 2005). NIHL has a profound impact on affected individuals, substantially reducing quality of life and impairing social and occupational abilities and relationships (Passchier-Vermeer and Passchier, 2000; Sataloff and Sataloff, 1996). Unlike other environmental problems, which have been reduced over time through regulation, noise exposures appear to be increasing over time (WHO, 1999). Additionally, noise exposure has historically been treated differently than other forms of environmental pollution (Hammer et al., 2014), and the approaches taken to addressing noise exposure issues in the U.S. have been less effective than those employed elsewhere (Hammer et al., 2017). While occupational noise exposure has long been recognized as a major risk factor for hearing loss, more recently researchers have suggested that recreational sound exposures (through, for example, activities such as attending concerts, listening to music, participation in noisy hobbies and recreational events, etc.) have the potential to increase the risk of hearing loss in individuals outside the workplace (McAlexander et al., 2015; Neitzel et al., 2012).

A variety of metrics are available with which to assess and quantify noise exposure. Measurements made with different metrics can result in divergent conclusions regarding NIHL risk, and the greater the variability, intermittency, or impulsiveness of a noise exposure profile, the greater the divergence between the metrics (Neitzel et al., 1999; Petrick et al., 1996). There is broad consensus among the scientific community that the best measure of the risk of NIHL from continuous exposure to noise is decibels (dB) of sound pressure level measured using the A-weighting network (e.g., dBA). For intense noise exposures (i.e., continuous exposures to noise greater than ∼110 dB, or to brief sounds of roughly 120 dB or greater), the C-weighting network (i.e., dBC) may be more appropriate (ACGIH, 2018b). However, the vast majority of human epidemiological data collected with regards to NIHL have been measured using dBA, and this remains the metric of choice for virtually all assessments of occupational noise.

There is general scientific agreement that the best metric for measuring average levels of noise—likely the best predictor of NIHL risk from chronic exposure to noise—is the equivalent continuous average sound pressure level, or L_EQ. The L_EQ can be computed over any period of time desired; for occupational noise exposure, the period chosen is nearly always 8 h, while for environmental noise exposure, the period chosen is often 24 h. When the exposure period is normalized to exactly 8 h, the L_EQ is referred to as L_EX. To account for noise levels that vary over time, the L_EQ uses a 3 dB time-intensity exchange rate (ER). That is, for every 3 dB change in the average exposure level, the allowable exposure duration is halved or doubled. So, for example, using a 3 dB ER, a noise exposure of 85 dBA on average for 8 h is considered to have identical risk of NIHL over time as an exposure of 88 dBA for 4 h, 91 dBA for 2 h, 94 dBA for 1 h, and so on (NIOSH, 1998).

Given the gaps in the literature regarding exposure limits for recreational sound, we conducted this review with the goal of determining an appropriate exposure limit to limit the risk of hearing loss risk due to recreational sound. To achieve this goal, we explored three specific questions. The first question related to what criteria and mechanisms are used to determine occupational and environmental noise exposure limits. Question two focused on whether there are differences in the nature and risk of hearing loss from occupational noise versus recreational sound. The third question was what is the most appropriate exposure limit for recreational sound given the available evidence from occupational and recreational exposures. Each of these questions is explored in a separate section below.

A variety of occupational and environmental noise exposure limits are in use around the world (Table I). The American Conference of Governmental Industrial Hygienists (ACGIH) has set its Threshold Limit Value (TLV) for noise (called “Audible Sound” as of 2018) as an 85 A-weighted decibel (dBA) 8 h-time-weighted average using a 3 dB time-intensity exchange rate (e.g., an L_EX) (ACGIH, 2018b). This TLV was established to “…protect the median of the population against a NIHL exceeding 2 dB after 40 years of occupational exposure for the average of 0.5, 1, 2, and 3 kHz” and “will not protect all workers from the adverse effects of noise exposure.” The TLV also sets a peak exposure limit of 140 dBC. The ACGIH TLV is essentially identical to the current Recommended Exposure Limit (REL) established by the National Institute for Occupational Safety and Health (NIOSH) in 1998 (NIOSH, 1998). The only substantive differences are in the exposure ceilings/peak limits (140 dBC for ACGIH, 140 dBA for NIOSH). The NIOSH REL is intended to reduce the excess risk of a material impairment (i.e., a Speech Intelligibility Index-weighted average hearing loss of 25 dB or greater across the audiometric frequencies of 1, 2, 3, and 4 kHz) to 8% of the exposed population over a 40-year working lifetime. NIOSH estimates a risk of material hearing impairment after a 40-year working lifetime of 1% at 80 dBA L_EX, 8% at 85 dBA L_EX, and 25% at 90 dBA L_EX (NIOSH, 1998).

Germany, Japan (Shaikh, 1999), Australia (NOHSC, 2000), and many other countries (Suter, 2003) have adopted an 85 dBA exposure limit essentially identical to those of ACGIH and NIOSH. However, unlike the ACGIH and NIOSH standards, the standards from these other countries do not provide any easily accessible documentation of the excess risk of hearing impairment upon which the limits are based. New Zealand (Australia, 2014) mentions that “AS/NZS 1269.4 predicts a 10 dB hearing loss for 95% of a noise-exposed population if they are exposed for 40 years to the noise standard of 85 dB(A) for 8 h per day,” with no specification of the audiometric frequencies at which this loss is expected to occur in the code of practice.

Occupational exposures to noise in most European countries are set according to European Union Directive 2003/10/EC (European Parliament and Council, 2003). This directive set limits for daily (i.e., exposure level for 8 h daily, or L_EX) or weekly (i.e., the average L_EX across five working days in a single working week) exposures. Three limits are specified by the directive: a lower exposure action value of 80 dBA L_EX, an upper exposure action value of 85 dBA L_EX, and an exposure limit of 87 dBA L_EX. Like the ACGIH and NIOSH limits, Directive 2003/10/EC specifies peak exposure limits as well as 8-h limits; these peak limits are 135 dBC for the lower exposure action value, 137 dBC for the upper exposure action value, and 140 dBC for the exposure limit. Note that while most EU countries have adopted Directive 2003/10/EC, not all have; for example, Sweden has set the L_EX exposure limit at 85 dBA, rather than 87 dBA (Arbetsmiljöverket, 2005). Unlike the ACGIH and NIOSH standards, Directive 2003/10/EC does not specify the excess risk of hearing impairment upon which the limits are based.

The U.S. Occupational Safety and Health Administration (OSHA) has a less protective Permissible Exposure Limit (PEL) of 90 dBA TWA with a 5 dB exchange rate (OSHA, 1983). Exposure at the OSHA PEL is expected to result in an excess risk of material hearing impairment of ∼25% at the audiometric frequencies 1, 2, 3, and 4 kHz after a 40-year working lifetime (NIOSH, 1998).

A guidance document developed by the International Organization for Standardization [ISO 1999-2013 (ISO, 2013)] provides information and equations that enable the prediction of noise-induced permanent threshold shifts at various audiometric frequencies and for varying exposure durations. The Introductory statement of ISO 1999-2013 (ISO, 2013) explicitly notes that the standard can be used to estimate the hearing impact of daily exposures to occupational and/or nonoccupational noise. The standard further states that nonoccupational noise exposure must be considered when evaluating noise-induced permanent threshold shift, except in the case where nonoccupational exposure is negligible in relation to occupational exposure. The standard also notes:

“The selection of maximum tolerable or maximum permissible noise exposures and protection requirements, as well as the selection of specific formulae for impairment risk assessment or compensation purposes, require consideration of ethical, social, economic, and political factors not amenable to international standardization. Individual countries differ in their interpretation of these factors and these factors are therefore considered outside the scope of this International Standard” (ISO, 2013).

Exposures described in the ISO standard are quantified in terms of L_EX8h, for a given number of years of exposure, and the standard is specifically described as applying to “noise at frequencies less than approximately 10 kHz which is steady, intermittent, fluctuating, [or] irregular.” Annex D of the standard includes tables with examples of noise-induced permanent threshold shift predicted at the audiometric frequencies of 0.5, 1, 2, 3, 4, and 6 kHz for L_EX8h exposures of 85, 90, 95, and 100 dBA for exposure durations of 10 to 40 years. Using these tables, it is possible to compute the average estimated noise-induced permanent threshold shift for the median of a population. For example, a 10 year exposure to a daily average of 85 dBA L_EX8h is expected to result in an average noise-induced permanent threshold shift at 1, 2, 3, and 4 kHz of 2.75 dBHL for the median individual, compared to average shifts of 5.25, 10.75, and 17.75 dBHL at 90, 95, and 100 dBA L_ex8h, respectively. Estimated average noise induced permanent threshold shifts at 1, 2, 3, and 4 kHz, derived from Annex D (ISO, 2013) are shown for different durations in Fig. 1.

Note that expected noise-induced permanent threshold shift increases most sharply in the early years of exposure, and that the rate of loss slows after that exposure window. This suggests that the first 10–20 years of exposure are the most critical in terms of risk of hearing loss, though hearing loss does continue to accrue with additional exposure time. These values reflect the predicted average hearing loss for the median of a population. Care should be taken when interpreting these predictions and applying to any individual; barring prior knowledge of an individual's susceptibility to NIHL, healthcare providers are encouraged to consider each individual as carrying higher susceptibility to NIHL than the median.

There are two primary environmental noise standards in use in the world today that are focused specifically on the prevention of NIHL, and at least one other national standard (Table II). The first primary standard is a Recommended Exposure Limit established by the U.S. Environmental Protection Agency in 1974 (EPA, 1974). This EPA limit is a 24-h L_EQ, or L_EQ24h, of 70 dBA using a 3 dB exchange rate (rounded down from 71.4 dBA to provide an additional margin of safety). For comparison to occupational exposure limits, this L_EQ24 value of 70 dBA is energetically equivalent to an 8-h exposure (L_EX using the conventional occupational exposure nomenclature) of 75 dBA (rounded down from 76.4 dBA to provide an additional margin of safety), assuming that the other 16 h of exposure in a given 24-h period have average levels of <60 dBA (EPA, 1974). When the EPA recommended this exposure limit in 1974, the rationale provided was that exposures to an L_EQ24 of 71.4 dBA over any 24-h period would result in a 5 dBHL or less shift in 4 kHz hearing threshold levels in 96% of the exposed population after a 40-year exposure. The EPA then added an additional margin of safety to the limit by rounding down to a recommended L_EQ24h of 70 dBA. In other words, this limit is intended to be completely protective against any measurable hearing loss in virtually all of the population over a 40-year exposure period. By comparison, the EPA estimated that 10 years of exposure at an L_EX of 80, 85, and 90 dBA would result in an average 4 kHz NIHL of 4, 9, and 15 dBHL, respectively. At the time the EPA recommended limit was released, the EPA made no distinction in the expected NIHL resulting from the same exposure levels of either occupational or nonoccupational noise.

In 1998, the World Health Organization adopted Community Noise Guidelines designed to protect against, among other things, NIHL (WHO, 1999). The WHO adopted the same primary limit as the EPA—that is, an L_EQ24h of 70 dBA—and reported essentially the same risk of hearing level (HL) for this limit as did EPA in their earlier document. The WHO guidelines also note that nonoccupational exposures have been noted to result in a similar amount of NIHL as occupational exposures of a similar duration and level. The WHO guidelines differ from the EPA recommendations with regards to music exposure. The guidelines suggest that exposure through headphones should not exceed 70 dBA L_EQ24h, or an equivalent 1-h exposure of 85 dBA. They further indicate that maximum levels of 110 dBA L_EQ should be avoided to prevent “acute hearing impairment.” WHO noted that for teenagers and young children the evidence suggests that L_EQ24h exposures <70 dBA are not expected to produce measurable NIHL. WHO also acknowledges that certain groups are especially vulnerable to small amounts of NIHL, i.e., children, those in the process of acquiring language skills, and the elderly (WHO, 1999).

The Swedish National Board of Health and Welfare has recommended limits specific to music which were initially derived from their occupational noise exposure limit (Arbetsmiljöverket, 2005). The limits for adults are a 1-h L_EQ of 110 dBA for no more than five days per week, and a ceiling level of 115 dBA (Ryberg, 2009). Given that these limits were derived from the upper exposure limit in use for occupational noise in Sweden, it is likely that that the limits assume some excess risk of hearing loss; however, no risk estimates were identified in a search.

The setting of occupational and environmental noise exposure limits is inherently a political issue that must consider social, cultural, and medical ramifications of exposure. The majority of nations and regulatory agencies in the world (with the notable exception of the U.S.) have selected an 85 dBA L_EX as their occupational exposure limit, and in doing so have (implicitly or explicitly) accepted an excess risk of a material NIHL (i.e., 25 dBHL or greater on average across the frequencies of 1, 2, 3, and 4 kHz) of 8% after a 40 years working lifetime or, expressed another way, an average NIHL of <5 dBHL across these frequencies in the median individual in a population after a 40 years working lifetime. Conversely, environmental noise standards (represented primarily by recommended limits from the EPA and WHO) specify an L_EQ24h of 70 dBA—equivalent to an L_EX of 75 dBA assuming noise ≤60 dBA for the other 16 h per day. This limit is intended to prevent any measurable hearing loss in any individual over a 40-year exposure lifetime, and includes a 5 dB margin of safety. The EU Directive for occupational noise specifies a lower exposure action value of 80 dBA L_EX, which is equivalent to the EPA and WHO recommendations without the margin of safety, and is presumed to involve nearly zero risk of NIHL over a 40-year working lifetime, though this is not explicitly stated in the directive documentation.

While occupational noise exposure of sufficient intensity has been proven beyond a doubt to cause permanent noise-induced hearing loss, the literature on the risk of permanent NIHL from recreational sound is not definitive. To evaluate the relationship between nature and impacts of exposures to occupational versus nonoccupational noise, we explored three key issues. These were: (1) evaluation of potential differences in the risk of hearing loss from identical occupational and nonoccupational noise exposures; (2) assessment of differences in exposure durations for occupational and nonoccupational noise exposures; and (3) evaluation of differences in exposure levels for occupational and nonoccupational noise exposures. Each of these issues is explored in detail below.

There is a growing body of literature suggesting a risk of temporary and permanent NIHL (le Clercq et al., 2016; Fligor, 2009; Gholamreza et al., 2016; Jiang et al., 2016; Lewis et al., 2013; Meyer-Bisch, 1996; Le Prell et al., 2012; Putter-Katz et al., 2015) and tinnitus (Guest et al., 2017; Moore et al., 2016) associated with duration and intensity of exposure to music and recreational noise, though there is not complete consensus on this risk (Colon et al., 2016; Twardella et al., 2017; Zhao et al., 2010). Temporary threshold shifts (TTS), measured via audiometry, have been demonstrated in a number of studies that had well-quantified exposure levels (Park, 2003; Potier et al., 2009; Sadhra et al., 2002), and in a larger number of studies with poor exposure assessment or very crude measures of hearing (Gunderson et al., 1997). Risk of permanent hearing loss has been identified in a handful of studies, most of which suffered from inadequate or incomplete exposure assessment (Axelsson and Lindgren, 1981; Mori, 1985; Kahari et al., 2003; Bray et al., 2004; Jansen et al., 2009; Kim et al., 2009; Sulaiman et al., 2013; Schink et al., 2014). Not all studies of permanent hearing loss from music have identified an elevated risk from music exposure for those with typical exposures (Mostafapour et al., 1998; Weichbold et al., 2012). Given the amount of evidence available, it is reasonable to assume that sufficiently high music exposure levels can result in temporary threshold shifts, and that permanent loss is possible given sufficiently high exposure levels and long exposure durations.

The literature comparing effects of exposures to noise and music signals that are of similar energy but differ in temporal pattern and frequency content is limited. It is very important to note that all of the studies below focused on short-term exposures and measures of TTS; there appear to be no chronic music exposure and permanent NIHL data available to answer this question.

Lindgren and Axelsson (1983) examined ten subjects in a study of TTS resulting from exposures to an electronically generated noise spectrum and found that these exposures resulted in TTS severity that exceeded those from musical noise of the same duration and overall A-weighted sound pressure level. Four of the subjects experienced essentially the same TTS from both sources, while six experienced greater TTS from the non-musical exposure than from the musical exposure. This provides some evidence that the content of sound, and the resulting subjective perceptions of exposure may affect the risk of TTS. In a separate study, Axelsson and Lindgren (1981) documented TTS effects among musicians that were less than those of audience members.

Strasser et al. (2003) also studied ten subjects over three energetically equivalent exposures to music and non-music sound over three days. Classical music (2 h exposure, mean 91 dBA) was found to be associated with substantially less TTS (10 dB vs 25 dB) compared to industrial noise of the same duration and mean level, as well as an energetically equivalent industrial level (94 dB for 1 h) and which recovered much faster (100 min vs 800 min). As with the Lindgren and Axelsson study, this study suggests that the content of sound may affect the risk of TTS.

Strasser et al. (1999) examined four energetically similar exposures (94 dB for 1 h): white noise, industrial noise, heavy metal music, and classical music. Industrial noise and heavy metal music were found to induce a similar amount of TTS and to require similar durations of time to recover (i.e., restitution time). However, classical music was found to result in less TTS and shorter restitution times than industrial noise, heavy metal music, or white noise. As with the previous studies, this study highlights potentially different consequences of exposure to classical music than to other types of music and industrial noise.

Mostafapour et al. (1998) prospectively examined hearing loss among 50 university student subjects (mean age 22.1 years). They compared noise exposures (assessed via self-reported participation in a number of occupational and nonoccupational events, as well as firearms use) to observed degree of hearing loss. The authors noted no association between qualitative exposure to any of the assessed sources of noise and presence of a noise notch (determined via pure tone audiometry), and determined that there was low risk of NIHL among the subjects.

Finally, Swanson et al. (1987) exposed 20 male subjects to music and noise of approximately equivalent energy (about 106 dBA) and for the same 10-min duration. Both exposures resulted in a significant post-exposure audiometric TTS at 4 and 6 kHz. TTS was significantly greater from music exposures among subjects who reported disliking the music used in the experiment. This study further supports the notion that subjective factors related to music may influence the risk of hearing loss resulting from music exposure, though it should be noted that audiometric testing involves a cognitive element that could conceivably be negatively influenced/biased by fatigue, loss of motivation, or frustration.

Lifetime exposure durations considered in current hearing loss risk models—which are uniformly based on occupational noise exposure data (ISO, 2013)—are far shorter than likely exposure durations for music and recreational sound exposures. The maximum duration of exposure for which hearing loss risk estimates are available is 40 years. This duration of exposure was appropriate in the 1960s and 1970s when early models for predicting hearing loss were created (EPA, 1974), as average life expectancies in high-income countries were approximately 65–70 years (National Institutes of Health, 2011), and noise exposures—primarily occupational in nature—did not start until first employment in a noisy industry. Today, exposures tend to start earlier; children may be exposed from an early age to high levels of music (Jiang et al., 2016), often delivered through headphones (Basjo et al., 2016). Also, life expectancies have lengthened considerably, and now exceed 80 years in many high-income nations (National Institutes of Health, 2011). These changes assume a 40-year lifetime exposure duration for music or high noise extremely conservative; in fact, lifetime exposures of 60 years (age 10 through 70) are more reasonable given current demographic and social trends. One final change that must be considered is the transition from noise being a primarily industrial exposure (with a reasonable expectation of 40 year lifetime exposure and limited contributions from nonoccupational activities) to one that, for many individuals employed in white-collar jobs, results primarily from nonoccupational activities, including listening to music (Neitzel et al., 2012). Given these substantial changes in sources and durations of exposure, the relevance of occupationally derived hearing loss prediction models that have an upper boundary exposure duration of 40 years is highly questionable. The average life expectancy is >40 years in all regions of the world, suggesting that a minimum 40-year expected exposure duration to noise is reasonable. For many regions of the world, life expectancies approach or exceed 70 years, suggesting noise exposure periods that may dramatically exceed traditional exposure duration assumptions in noise standards (NIOSH, 1998; OSHA, 1971; WHO, 1999).

A secondary consideration with regards to exposure duration is the daily exposure. Whereas occupational exposures to noise have historically been restricted to 8 h in many industries (though with exceptions; workers in some industries may work up 12–16 h per day) (Neitzel et al., 2006). However, music exposures among both children and adults have the potential to exceed these durations daily, though it appears that most listen to music for something on the order of one to several hours per day (Jek et al., 2014; Twardella et al., 2017; Zhao et al., 2010). Neitzel et al. noted that among nearly 4500 adult subjects, the vast majority of individuals were estimated to receive most of their exposure through music (Neitzel et al., 2012). Although they used MP3 players and stereos to listen to music only 2.2 h per day on average, music was the primary source of annual noise exposure for 59% of subjects in that study. In a separate analysis of the same cohort in New York City, Lewis et al. (Lewis et al., 2013) estimated that the risk of any NIHL was greater from music than from any of the other sources assessed (e.g., occupational noise, transit noise, etc.) in that study. However, the authors also noted that the greatest impairment expected among the study subjects was associated with individuals with high levels of occupational noise exposure.

A number of studies have evaluated maximum outputs of personal music players and demonstrated possible exposure levels that far exceed any recommended exposure limit (Fligor and Cox, 2004; Kumar et al., 2009; Torre 3rd, 2008). Such studies are useful to establish the potential risk of hearing loss, but do nothing to quantify that risk. Fortunately, there now exists a large and growing body of literature describing the listening levels and estimated L_EX exposures for users of portable music players. A review of this literature has already been conducted by Jiang et al. (Jiang et al., 2016). This review, which involved a systematic review of the published literature and an evaluation of 26 studies that met the authors' inclusion criteria, determined that between 3% and 58% of subjects in each of the individual studies exceeded a daily dose of 85 dBA L_EX from their music listening alone. These exceedances are similar to a number of studies that have evaluated the prevalence of overexposures to occupational noise across various industries (see, for example, Franks, 1988; NIOSH, 1998; Kock et al., 2004; Tak et al., 2009; Neitzel et al., 2011; Neitzel et al., 2012), and highlight the potential similarities between music and occupational exposure levels.

ISO 1999 (ISO, 2013) explicitly states a specific range of noise exposures (75 to 100 dBA) to which its estimation equations can be applied, and that the noise exposures for which the standard is relevant are:

The systematic review by Jiang et al. indicated that average L_EX exposures across the 26 studies reviewed ranged from 61.6 to 87.2 dBA (Jiang et al., 2016). These exposures range from values that are below the threshold for hearing loss risk consideration using the ISO standard to values that are well within the specified range of these standards. L_EX exposures below the ranges specified in the ISO standard are presumed to present no risk of NIHL even with chronic exposure, while the L_EX exposures within the range of the ISO standard all have an expected risk of NIHL, with that risk depending on the magnitude of the exposure. Given the overlap between L_EX exposures associated with listening to music and L_EX exposures available for hearing loss prediction using the ISO standard, it is reasonable to assume that the hearing loss predictions from these standards are appropriate for application to music exposures.

There are suggestions in the literature that the effects of occupational/industrial noise exposures on TTS may be worse than those of some types of energetically equivalent music. If true, this would indicate that exposure limits based on risk of NIHL from occupational noise may be overprotective for exposures to some types of music. There is also an indication in the literature that contemporary daily listening durations among children and young adults may be substantially greater than those traditionally associated with occupational noise exposures. However, the tremendous variation in types and patterns of music listening warrant the adoption of conservative exposure guidelines that presume exposure to the most harmful types of music (i.e., that music exposures be considered equally as hazardous as noise exposures). The standards currently available for estimating risk of NIHL are assumed to be appropriate for daily exposures of up to 12 h, which appears to be greater than the vast majority of listening durations reported by young adults (SCENIHR, 2008). Furthermore, the average levels of exposure associated with music are consistent with those used to predict noise in ISO 1999 (ISO, 2013). Given these findings, this report therefore recommends that exposure limits developed for occupational noise be considered appropriate for assessing risk from music exposures, though additional exploration of the dose-response relationship between chronic music exposure and hearing loss is needed.

Unless an exposure level associated with zero risk of hearing loss (i.e., 75 dBA L_EX, equivalent to a 70 dBA L_EQ24h) is adopted as a recommended exposure limit, the adoption of any exposure limit is inherently a political compromise that explicitly acknowledges that some exposed individuals will suffer an adverse health outcome—in this case, permanent NIHL. Another factor that must be considered is the possibility that exposure limits intended to protect individuals against NIHL may not be protective against other, arguably more significant health effects (e.g., hypertension and myocardial infarction, which are at this point well-substantiated sequelae associated with community and occupational exposures to noise) (Basner et al., 2014; Münzel et al., 2014; Passchier-Vermeer and Passchier, 2000). The consideration of these health outcomes is beyond the scope of this report, but it is important to note that there is evidence that non-auditory health impacts begin at levels substantially below those associated with NIHL (Basner et al., 2014; Münzel et al., 2014) (e.g., <55 dBA 24-h day-evening-night noise level, L_DEN, equivalent to 60 dBA L_EX). Additionally, the dose-response relationship between noise exposure and noise-induced tinnitus is not well defined. It is known that tinnitus is strongly associated with noise-induced permanent threshold shift, but an upper limit of exposure known to not cause noise-induced tinnitus has not been established (Hoffman and Reed, 2004; Engdahl et al., 2012). Such is a limitation of the science at present. However, there is some evidence that noise exposure can result in “hidden hearing loss” (i.e., damage to the auditory periphery, including cochlear synaptopathy, that does not manifest as elevated audiometric thresholds) (Liberman et al., 2016). While it is premature to attempt to derive a dose-response relationship for such hearing loss given the state of the science on this issue, such loss would be expected to precede any measurable change in hearing threshold levels in the conventional range of test frequencies (i.e., 125–8000 Hz), though the relationship between noise and high frequency audiometric outcomes (e.g., 10 kHz and above) is unclear (Liberman et al., 2016).

One critical consideration in determining an acceptable exposure limit for music or noise is the criterion specifying the amount of hearing loss that would constitute an acceptable “noise-induced hearing loss.” Predicted noise-induced permanent threshold shift for the median population and risk for a material hearing impairment in the speech (ISO, 2013; NIOSH, 1998) are presented in Table III. For comparison, Table IV shows the predicted NIPTS at noise-sensitive frequencies (ISO, 2013) for the 10th and 5th population percentiles (i.e., those most susceptible to NIHL).

For the purposes of establishing a recommended exposure limit, a quantitative definition of hearing loss is needed unless a 70 dBA L_EQ24h exposure limit is adopted, in which case sufficient literature exists to establish zero risk of NIHL at any audiometric test frequency in any individual over a 40-year exposure duration (EPA, 1974; WHO, 1999). WHO has previously stated that hearing loss in excess of 10 dBHL average audiometric hearing threshold at 2 and 4 kHz in both ears may have an impact on speech comprehension, and that losses in excess of 30 dBHL average at 2 and 4 kHz in both ears result in a noticeable social hearing handicap (WHO, 1999). However, there are other definitions of NIHL in use around the world. For example, the American Academy of Otolaryngology–Head and Neck Surgery considers the audiometric frequencies of 0.5, 1, 2, and 3 kHz (Ward, 1983), while NIOSH considers 1, 2, 3, and 4 kHz (NIOSH, 1998). Many of the current definitions of “hearing impairment” consider hearing loss sufficient to warrant being fitted with hearing aids as the threshold for a “material impairment.”

There may be individuals for whom a “zero risk of NIHL” (i.e., exposure limit of 70 dBA L_EQ24h) is appropriate. These include: young children or those not expected to have the autonomy to make informed personal health decisions; persons with pre-existing hearing loss (NIHL or from another cause) or pre-existing tinnitus; persons who have a family history of NIHL, or in whom there is reason to believe increased susceptibility to NIHL or noise-induced tinnitus (e.g., persons treated with ototoxic medications; and persons exposed to chemicals that might potentiate the deleterious effects of noise). In others, it may be appropriate to provide a more relaxed exposure limit, respecting individuals' autonomy, while providing informed understanding of the risks. A more relaxed exposure limit might be set at 83 dBA L_EX (equivalent to a 78 dBA L_EQ24h).

When considering the risk of NIHL over a lifetime, it is important to note that there is already evidence of NIHL in young people first entering the workforce. Seixas et al. (2004) noted unexpectedly elevated baseline hearing threshold levels among young construction workers (mean age 27.2 +/− 7.0 years) compared to medical student controls of the same mean age. This was attributed to the construction subjects already having worked an average of 2 years in construction prior to entry into the apprenticeship program from which they were recruited into the study, but also to previous exposure to nonoccupational noise (e.g., motorcycles, music, firearms, etc.) that exceeded that of the student controls. Seixas et al. (2012) followed a subset of these subjects prospectively for 10 years, and found that NIHL progression exceeded that predicted by ISO 1999 (ISO, 2013), even when baseline hearing levels were considered.

Rabinowitz et al. (2006) examined hearing threshold levels among young adults (mean age 22.2 +/− 2.1 years) upon entry to the industrial workforce of a single multinational metals manufacturing company over a 20 year period (1985–2004). The authors did not identify an increasing trend of high frequency hearing loss over that time period, but did find that about 20% of the young workers studied in each year of the study period featured audiometric notches consistent with NIHL. As with the Seixas et al. cohort (Seixas et al., 2012), some of these young workers may have had previous occupational noise exposure, and, also consistent with Seixas et al., Rabinowitz et al. (2007) observed greater rates of hearing loss among industrial workers with worse hearing at baseline.

Collectively, these studies suggest that a non-trivial fraction of workers entering the workforce as young adults, or early in their working career, may have already experienced NIHL. While some of this NIHL may have resulted from previous occupational exposures in the cohort studied by Seixas et al. (2004), a substantial fraction of workers evaluated by both Seixas et al. and Rabinowitz et al. (2006) reported noisy nonoccupational hobbies. Given that contemporary workers are likely to have exposures to occupational noise as well as nonoccupational noise (including music), and that young workers have already been noted to have hearing impairment at the beginning of their employment, the need for nonoccupational exposures limits that are truly protective (i.e., eliminate the risk of NIHL) is apparent.

Based on the available evidence, we recommend three possible alternatives for an exposure limit for recreational sound.

• A limit that is known to eliminate the risk of NIHL in any exposed individual over the longest exposure duration that can currently be modelled. This exposure limit is a 70 dBA L_eq24h, equivalent to a 75 dBA L_EX, and includes a margin of safety to account for vulnerable/susceptible individuals, and would be appropriate for individuals with reduced autonomy regarding health risks, individuals with existing hearing loss, or individuals at increased risk of hearing loss.

• A limit that will result in a very small fraction of exposed individuals encountering a material hearing impairment, but which is protective against a substantial hearing loss for virtually all exposed individuals. This exposure limit is a 75 dBA L_EQ24h, equivalent to an 80 dBA L_EX (i.e., the current European Union occupational noise lower exposure action value). This 80 dBA L_EX exposure limit is intended to nearly eliminate the risk of measurable NIHL following a 40-year working lifetime. Conversely, however, the standard may not be sufficiently protective. Lifetime exposures to music and noise may reasonably be expected to exceed a cumulative duration greater than 40 years, the loss from which cannot be accurately estimated using existing predictive standards (EPA, 1974; WHO, 1999), though it is worth noting that these models predict that the greatest rate of change in NIPTS will occur in the first ten years of exposure, and that exposures beyond that time contribute much less to overall NIPTS. Nevertheless, despite these uncertainties, adoption of an 80 dBA L_EX limit for exposure to music likely represents an optimal trade-off between being sufficiently protective and being onerous and/or technically or socially infeasible.

• A liberal recommended exposure limit, applicable to individuals willing to tolerate modest risk for a small degree of NIPTS, but still sufficiently protective of the vast majority of people exposed to nonoccupational music exposures. An appropriate level for such a limit might be 83 dBA 8-h L_EX (i.e., 78 dBA L_EQ24h; 92 dBA 1-h L_EQ).

With this review we have sought to determine an appropriate exposure limit to limit the risk of hearing loss risk due to recreational sound. In doing so, we have addressed three specific questions. With regard to the first question, what criteria and mechanisms are used to determine occupational and environmental noise exposure limits, we note that most occupational noise exposure limits are set at an 85 dBA L_EX, and that the primary nonoccupational exposure limit is a 70 dBA L_EQ24h. The 80 dBA L_EX lower exposure action value of the European Union (equivalent to a 75 dBA L_EQ24h) represents a balance between the common occupational and nonoccupational limits. With regard to question two, whether there are differences in the nature and risk of hearing loss from occupational noise versus recreational sound, we found that standards for estimating risk of hearing loss from occupational noise are appropriate for estimating the risk associated with recreational sound. Finally, with regard to the third question—what is the most appropriate exposure limit for recreational sound given the available evidence from occupational and recreational exposures—we recommend a 70 dBA L_eq24h exposure limit for individuals who are vulnerable, susceptible, have reduced autonomy regarding health risks. For others, a limit of 75 dBA L_EQ24h will nearly eliminate the risk of measurable NIHL following a 40-year exposure period. This latter limit for exposure likely represents an optimal trade-off between being sufficiently protective and being technically and socially feasible.

Our review focused on hearing loss risk from occupational noise and recreational sound. However, as we have noted, there are other important auditory and non-auditory health risks associated with noise exposure. In particular, there is no recommended exposure limit based on dose-response relationship between noise exposure and tinnitus; thus, our recommendations may, or may not, be appropriate for mitigating risk for noise-induced tinnitus. However, based on limited evidence (Guest et al., 2017; Moore et al., 2016), it appears that noise-induced tinnitus may precede measurable NIHL, suggesting that more stringent exposure limits may be necessary to protect against noise-induced tinnitus. Furthermore, our recommendations do not specifically consider the impact of additional decrease of hearing sensitivity in those individuals who have pre-existing hearing loss (from noise, or from other causes). Both of these areas warrant further research.

The authors are indebted to the following individuals for their review and input: Ian Wiggins, University of Nottingham; Jeremie Voix, École de Technologie Supérieure; Warwick Williams, National Acoustic Laboratories; and Shelly Chadha, World Health Organization. This work was initially created as a report for the World Health Organization Make Listening Safe Campaign.