The efficacy of earplugs at a major hazard facility

At a major hazard facility, cases of occupational noise-induced hearing loss have occurred despite the use of hearing protection devices. Preliminary measurements of personal attenuation ratings (PAR) suggested that earplugs may not achieve the attenuation implied by their Australian SLC80 Class designation in field-use conditions. We examined the relationship between PAR and the SLC80 classification for earplugs, as a good understanding of the attenuation provided by earplugs under normal field-use conditions, rather than in laboratory settings, is vital to their effective use as a control measure for noise exposure. A cross-sectional study was performed with 65 volunteers. Participants were recruited from Operations and Maintenance Technicians at the major hazard facility. The participants had their PARs checked with different earplug types using the 3 M™ E-A-Rfit™ system. We examined the PARs measured and also assigned a ‘pass’ or ‘fail’ rating depending on whether the earplug achieved 22 dB attenuation. 22 dB attenuation for 80% of users is the minimum to meet the definition for Australian SLC80 Class 4 hearing protection. None of the earplug types achieved 22 dB attenuation for 80% of users when tested in field-use conditions. There were statistically significant differences in the frequency of achieving a ‘Pass’, and in PAR, depending on earplug type. Roll-down foam earplugs may provide superior attenuation compared to pre-moulded earplugs. Earplugs are unlikely to achieve the attenuation found in laboratory conditions during field-use. Personalised selection of hearing protection devices based on fit-testing results should be encouraged.


Introduction
Occupational noise-induced hearing loss is a significant health problem both in Australia, and globally. Between July 2002 and June 2007, there were approximately 16,500 approved worker's compensation claims for permanent impairment from occupational noise-induced hearing loss in Australia [1]. A survey by Safe Work Australia in 2010 found that 28% to 32% of workers were exposed to noise over 85 dB at work, that only 41% of exposed workers reported receiving training on how to prevent hearing damage, and that there appeared to be a reliance on hearing protection devices (HPDs) to reduce noise exposure [2]. A 2020 study found that under current occupational noise exposure levels in Australia, over 110,000 workers would develop occupational noise-induced hearing loss over 10 years of exposure, with a projected loss of $AUD 26.8 billion due to well-being and productivity loss [3].
A 2005 study found that globally, 16% of disabling hearing loss in adults is due to occupational noise, ranging from 7 to 21% by region. This represented a burden of over 4 million disability-adjusted life years (DALYs) [4]. In 2018, hearing loss as a whole is the fourth highest cause of disability worldwide, and has an estimated annual cost of $750 billion [5]. The World Health Organization (WHO) estimated that there are 1.5 billion people with some degree of hearing loss globally [6], with an estimated 466 million people of those suffering disabling levels of hearing loss, and this figure is expected to rise to 630 million by 2030 and 900 million by 2050 [5]. High levels of occupational noise continues to be a worldwide problem [6,7]. For example, occupational hearing loss is still the most common work-related illness in the United States [8], and a 2015 study reported that occupational noise-induced hearing loss was still increasing in some countries in the European Union [9].
Hearing protection devices, which include earplugs and earmuffs, represent a mature technology that has been widely used since the 1950s [10]. It is well understood that correct insertion technique is vital to the efficacy of earplugs in noise attenuation [11,12]. Earplug attenuation is also affected by a number of other factors including the shape and size of the user's ear canal [13,14], the user's dexterity, and the shape of the earplug [15]. Despite our knowledge of these factors, multiple studies have found that earplugs rarely achieve their laboratory-assigned attenuation ratings [16][17][18][19]. A good understanding of the attenuation provided by earplugs under normal field-use conditions, rather than in laboratory settings, is vital to their effective use as a control measure for noise exposure.
Our study examined the objective attenuation achieved by 65 users with four structurally different types of earplugs. In Australia, hearing protection devices are divided into classes based on the attenuation they are found to provide when tested using the SLC80 method (Sound Level Conversion valid for 80 per cent of the wearers). The specific details of the testing method are described in the Australia/New Zealand Standard, AS/NZS 1270:2002 [20]. We sought to compare our measurements of earplug personalised attenuation ratings (PARs) with their labelled SCL80 classes. At the time of writing, the authors were unable to find any prior published studies examining the relationship between PAR and the Australian SLC80 Class ratings, and believe we may be the first to do so.

Methods
This cross-sectional study was approved by the Monash University Human Research Ethics Committee under Project ID number 21527.
We examined four structurally different types of Class 4 earplugs and checked their attenuation using a commercially available microphone-in-real-ear (MIRE) system that produces objective test results. As per AS/NZS 1270:2002 (Table 1), Class 4 earplugs should provide at least 22 dB attenuation to 80% of users [20]. We recorded the PAR achieved by participants with each type of earplug. We also calculated the percentage who achieved 22 dB of attenuation (deemed a 'pass' for the test).
Sixty-five volunteers were recruited by email, telephone call, in person at their worksites, and opportunistically at clinic visits from the Operations Technicians and Maintenance Technicians workgroups at the facility in question.
The participants work in 12-h shifts, and typically spend about 4 h a day in areas where the noise levels are 85 dBA or greater. Noise mapping demonstrated that noise levels at the facility ranged from less than 79 dBA to over 97 dBA. Long term average exposures of the workers ranged from 66 to 89 dBA. The noise is steady state without significant impact, peak or impulsive sources.
All the workers had previously received training on correct HPD use at the commencement of employment from a supervisor using standardised instructions, and then annually through computerised training modules that include an instruction video produced by 3 M™. For their daily work, the workers are able to select from a range of products that include earplugs and earmuffs. The 3 M™ Classic™ earplugs used in this study are among the options that workers can use. Earplugs structurally similar (bullet-shaped roll-down foam, cone-shaped pre-moulded, and flanged pre-moulded types) to the ones tested in this study are also available to the workers for daily use, but they are produced by different manufacturers.
The facility that the participants were recruited from manufactures polyethylene and other polymers for household goods and industries such as food and beverage, construction, mining and energy, agriculture, and water conservation. The E-A-RFit™ PAR testing equipment and disposable test probes for the study were supplied by the facility. The researchers were not paid for their time. There were no other costs to account for.
The inclusion criteria were age 18 years or older, and employment as an Operations Technician or Maintenance Technician. The study was open to workers of both genders, but due to the overwhelming majority of the study population being male, there were no female participants.
Exclusion criteria included anatomic or medical conditions that prevented insertion of earplugs (e.g.: severe otitis externa, congenital or anatomical deformity preventing earplug insertion). Individuals with conditions that affected the ear canal but did not prevent insertion of earplugs, such as mild to moderate exostoses, were not excluded from the study. Participants were given a consent form and the test procedure was explained. An explanatory statement was available if participants desired more information. Participants had their personal attenuation rating (PAR) measured using the commercially available 3 M™ (The 3 M Company, Maplewood, Minnesota) E-A-RFit™ system ( Fig. 1) [21] and a selection of four structurally different probed earplugs. The E-A-RFit™ system is a device with a speaker as the source of a standardised broadband noise. It has been found to be a satisfactory alternative to Real Ear Attenuation at Threshold (REAT) testing methods [22].
The earplugs selected were the following: The chosen earplugs for this study were all Class 4 to avoid the introduction of an additional variable in this study, and also because Class 4 had the greatest number of structurally distinct earplugs for us to examine.
The research participants were seated approximately 30 cm front of the E-A-RFit™ system, wearing probed earplugs that were connected to microphones. The microphones were used to detect the sound pressure level reaching the inside of the ear canal past the earplugs, and just outside the earplugs close to the auricles. The microphones were clipped to either the participant's own eyewear, or to a pair of safety glasses provided for the test. The difference in sound pressure levels between the microphones for outside and inside the earplugs is a measure of the attenuation provided. The system uses a Microphone in Real Ear (MIRE) objective measurement approach, which does not depend on the participant's subjective sense of hearing [18]. After measuring the attenuation provided, the E-A-RFit™ tool calculates a Personal Attenuation Rating (PAR), a summary measure which can be directly subtracted from the measurements      [23] of A-weighted noise sources when calculating a user's protected noise exposure [24]. The system was calibrated every time it was turned on, before testing commenced.
Each participant was examined once in a single session, using the four different types of test earplugs. The sessions were conducted in quiet office environments. Participants were given instructions on how to correctly roll down or grip the relevant earplugs, reach over with the other hand to open the ear canal, and insert the earplugs. The researcher physically demonstrated the correct insertion technique to reinforce the instructions. Where roll-down earplugs were used, approximately 1 min was allowed to pass for the earplugs to expand. After insertion, the earplugs were visually examined by the researcher to see that they appeared to sit within the ear canal correctly without crumpling or creasing. If the earplugs were incorrectly inserted, or if they protruded past the tragus, the participant was encouraged to attempt a better or deeper insertion of the earplug. Participants were allowed up to three attempts at achieving an optimal insertion. They were not disqualified if the earplugs were visible past the tragus despite their best efforts. They would have been disqualified if the earplugs appeared to otherwise be incorrectly inserted despite three attempts, or unable to be correctly inserted. However, this scenario did not occur with any participant. The participant was then asked to face the speaker, and the broadband sound was then played and the system calculated the worker's binaural PAR. The binaural PAR was used because it is a summary measure that accounts for the PAR measured for each ear, and is best for predicting overall hearing protection for the individual [24]. The results were automatically recorded by the system software. Attenuation of 22 dB or more was categorised as a 'pass' for that test, as 22 dB is the minimum acceptable attenuation for Class 4 hearing protection. The participants were tested with the 3 M™ Yellow Neons™, Push-Ins™, Ultrafit™, and Classic™ single-use probed earplugs, in that order. This order was chosen to help avoid the occurrence of a practice effect from one type of foam roll-down earplug being tested immediately following another.
After testing, the participants' personal results were shown to them and they were given advice about the type of earplug that had worked best for them. Participants who did not achieve a 'pass' with any type of earplug were encouraged to use earmuffs in future.
We compared the proportions of each earplug type that achieved the minimum attenuation for their class rating; i.e. 22 dB for Class 4. We also compared the PAR of the different types of earplugs using logistic and linear regression models which handle clustered or repeated measurements. All four types of earplug were tested on each participant. Repeated measurements were analysed using logistic regression for protection pass/fail and linear regression for PAR, as initial descriptive analyses suggested that the PAR values for each earplug were normally distributed. Logistic and linear regression were undertaken with the clustered (i.e. repeated measures) version of the robust Huber/White sandwich estimator [25][26][27]. All 95% confidence intervals and p values took repeated measurements into account (i.e., there were 260 measurements from 65 participants).
Type of earplug was entered into the regression models. Following on from a 2018 study [15], we also compared the two foam roll-down plugs (Yellow Neons™ and Classic™) as a group against the two pre-moulded plugs (Push-ins™ and Ultrafit™) as a group, using a planned comparison. Two other planned comparisons compared: 1. Yellow Neons™ versus Classic™ and 2. Push-ins™ versus Ultrafit™.

Results
All participants produced complete data, and none were rejected for any reason. As discussed earlier, both male and female subjects were invited to participate, but only male subjects did so.
Participants ranged from 24 to 69 years of age, with a median age of 54 years, a mean age of 49.6 years, and a standard deviation of 13.3 years.
The PAR results obtained ranged from 0 to 35 dB. None of the four types of earplugs achieved the attenuation implied by their class rating when tested in field-use conditions; i.e. at least 22 dB PAR for 80% of users. All earplugs considered together achieved 22 dB PAR for only 62 of the 260 measurements, which was 23.85% of users [95% confidence interval (CI) 17.19% to 30.51%]. Table 2 below shows the breakdown of the results by earplug type.
There was a statistically significant overall difference in the frequency of achieving a 'Pass' between different earplug types (Wald likelihood chi-square = 27.22, df = 3, p < 0.001). Performance of earplug groups (foam roll-down and pre-moulded) is given in Table 3.
The foam roll-down earplugs performed significantly better than pre-moulded earplugs [odds ratio (OR) 5.44, 95% CI 2.64 to 11.21lp < 0.001]. Within the foam category, Clas-sic™ performed significantly better than Yellow Neons™ (OR 1.81, 95% CI 1.04 to 3.17, p = 0.037), whereas within the pre-moulded category there was no statistically significant difference between Push-Ins™ and UltraFit™ earplugs (OR 1.68, 95% CI 0.78 to 3.62, p = 0.182). Table 4 shows the mean and 95% confidence interval by earplug type for PAR, measured in decibels. Table 5 shows the mean and 95% confidence interval for broad earplug category for PAR, measured in decibels.

Discussion
The earplugs we tested did not achieve the attenuation implied by their class ratings. Foam roll-down plugs as a group (Yellow Neons™ and Classic™) appeared to be superior to premoulded plugs as a group (Push-ins™ and Ultrafit™), which is consistent with the findings of a 2018 study [15]. Within the groups, differences were less clear. Classic™ had significantly higher 'pass' results than Yellow Neons™, but the two types of earplug did not differ significantly on mean PAR. The opposite pattern was observed in the case of pre-moulded earplugs. Ultrafit™ did not differ from Push-ins™ on passes, whereas Ultrafit™ exhibited significantly higher mean PAR than Push-ins™.
The percentages of earplugs achieving the minimum attenuation for their class were low, which is consistent with the   1 3 findings of previous studies demonstrating that laboratory measurements are not reliable for predicting the attenuation provided to users in field conditions [16][17][18][19]. Studies similar in methodology to ours have demonstrated similarly disappointing results in the range of attenuation, mean PAR, and percentage of results achieving acceptable attenuation for earplugs [28][29][30]. This includes studies based on the American Noise Reduction Ratio (NRR) system, which demonstrates that the discrepancy in earplug attenuation measurements between laboratory ratings and field-users is not isolated to one type of earplug rating system. One of the reasons for the discrepancy might be that laboratory testing may involve training that differs from what end-users typically undergo in field conditions. Training and fitting technique is crucial to the attenuation provided by earplugs and it has been found to have a larger effect than the laboratory rating for the earplug [12]. Users of earplugs may need to consider that it may not be sufficient to rely on SLC80 ratings or classes when selecting which HPDs to use, and consider the use of personalised fit-testing to supplement their selection. It has been found that repeatedly training earplug users, measuring their attenuation, and fitting them with different earplug types to find the best fit results in significant improvement in earplug attenuation [28][29][30]. Training can even significantly increase the proportion of earplug users who are able to achieve acceptable levels of attenuation [31]. However, training alone is unlikely to be sufficient to address the issue of insufficient attenuation in workers, and it is essential that that alternative protection methods or HPD types are available to earplug users who are unable to achieve adequate attenuation [32].
In Australia, subjects selected for the SLC80 rating process for HPDs must have had no prior experience or training with HPDs [20]. This is likely to be in contrast to the majority of HPD-users in field conditions. In our study, for example, the participants used HPDs every day at work. While this observation might at first suggest that participants should have performed better than rating subjects, it could also mean that HPD-naïve rating subjects pay closer attention to the training they receive as they are hearing it for the first time, and are more careful in following instructions as closely as possible. Typical HPD-users, on the other hand, may instead default to pre-conceived notions or habits, which may have a negative effect on the earplug fitting.
Rating subjects may also differ from end-users in other meaningful ways. The gender ratio of subjects for SLC80 rating must be 50/50 ± 10% [20], whereas the participants of our study were 100% male due to the vast majority of the study population being male, and the gender ratios of rating subjects may likewise be different to the population of HPD-users. The attenuation of earplugs could differ between the two groups, because there are anatomical differences in the external auditory canal between males and females [33].
The education level and socioeconomic status may also differ between rating subjects and HPD-users, as the latter may possibly be more likely to be blue-collar workers. Socioeconomic status is inversely related to cognitive ability [34], and thus may affect literacy and comprehension of instructions and subsequently, earplug fitting. Unpredictable physiological differences such as differences in manual dexterity could also be at play. Future studies should take such factors into account.
An advantage of our study design is the objectivity of the MIRE method [35]. Because the earplug attenuation is measured by a microphone with no input from the person being tested, issues like temporary threshold shifts, hearing loss, and concentration did not play a role in the measurements. Our study is also likely to be highly relevant to the majority of frequent users of HPDs, as the participants were drawn from a population that uses HPDs daily. The population was fairly homogenous, being made up of mostly Caucasian males aged from their 20s to late 60s, and our sample was reflective of this. This helps to reduce the effect of confounders such as age, gender, and ethnicity. Recruiting only Operations and Maintenance Technicians also helps to standardise the education level of the participants.
Limitations of our study include the fact that the MIRE method does not account for bone conduction [36], unlike the REAT method. It is therefore possible that MIRE methods may overestimate the attenuation provided by HPDs. There may have been an order effect as one order of earplugs was used throughout testing, and future studies may instead employ order randomisation. Our study may also be less generalisable to the public, which may be of importance when considering the attenuation of HPD-naïve users, visitors to facilities where HPDs are required, infrequent users of HPDs, and so on. It may be hard to predict how their PARs may differ from our sample, and they may be less protected or even overprotected by comparison. Indeed, the highest measured PAR in our study was 35 dB, which is well above the maximum attenuation of 25 dB for a Class 4 earplug. This could easily have produced attenuated exposures below 70 dB, resulting in overprotection [37]. We also limited our study to Class 4 earplugs manufactured by 3 M™. This was due to the selection of probed earplugs available for testing with the E-A-Rfit™ system. It is compatible only with 3 M probed earplugs, and Class 4 offered a good variety of different earplug structures. It is uncertain how findings may differ in earplugs of different Class ratings or manufacturers.
Further work could expand on our study by incorporating different Classes of HPDs, including the consideration of overprotection scenarios (whereby HPDs provide more attenuation than desired or necessary), relating PAR measurements to ear canal structures and/or hearing threshold shifts, and examining the PAR of each ear separately rather than as a binaural measure.

Conclusion
Earplugs are unlikely to achieve the attenuation found in laboratory conditions during field-use. Roll-down foam earplugs may provide superior attenuation overall compared to pre-moulded earplugs. However, our raw data revealed that for a minority of individuals, their personal best PAR was achieved with pre-moulded earplugs. The range of PAR measurements was also high; from 0 to 35 dB. The attenuation offered by earplugs to individuals is thus highly variable. Individual fit-testing of HPDs in conjunction with training has been shown to improve PAR results [28]. Fit-testing is a valuable addition to hearing protection programs [38], and personalised selection of hearing protection devices based on fit-testing results should be encouraged.
Author contributions KHL-Study design, data collection, article write-up. GB-Study design, proofreading. DM-Statistical analysis, statistical write-up.

Funding
The major hazard facility in question loaned their E-A-Rfit device for the study and provided the consumables (test plugs). There was no other funding required or received.

Data availability
The de-identified raw data is available on request from the corresponding author.
Code availability Not applicable.

Declarations
Conflict of interest Dr Kah Heng Lee provides occupational medical services to the major hazard facility in the study.
Ethical approval This cross-sectional study was approved by the Monash University Human Research Ethics Committee under Project ID number 21527.
Informed consent Written informed consent to participate in the study with disclosure that the research could potentially be published was obtained from the study subjects.