Annals of Occupational Hygiene Advance Access originally published online on June 16, 2006
Annals of Occupational Hygiene 2006 50(7):679-691; doi:10.1093/annhyg/mel025
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Variability of Real-World Hearing Protector Attenuation Measurements
Department of Environmental and Occupational Health Sciences, University of Washington Box 354695, Seattle, WA 98195-4695, USA
*Author to whom correspondence should be addressed. Tel: +1-206-221-5445; fax: +1-206-616-6240; e-mail: rneitzel{at}u.washington.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The attenuation provided by a hearing protection device (HPD) in the field is usually estimated by applying a derating factor to the laboratory-determined noise reduction rating (NRR) of the HPD. However, attenuation is highly dependent on individual-specific HPD fit. Prediction of an individual's attenuation depends on the accuracy of the measurement system and the variability of attenuation over time (e.g. after HPD refitting). Variability in attenuation and attenuation test systems has not been adequately characterized to allow for such an assessment. This study compared attenuation measurements made with two systems, Real-Ear-at-Threshold (REAT) and Microphone-in-Real-Ear (MIRE), on 20 workers using two earplugs (foam and custom-molded). Workers' perceptions of the earplugs were also evaluated. Individuals' attenuation results were summarized as personal attenuation ratings (PARs, similar to NRRs). Variability in PARs from between-subject, within-subject and within-day (i.e. repeated tests on a subject without earplug refitting) differences was assessed and used to present the lower confidence limit, or uncertainty factor (UF), of an average individual's attenuation. The custom-molded earplug PARs achieved a higher mean percentage of labeled attenuation than did the foam earplug with both test systems. The custom-molded earplugs also had higher overall acceptance among workers. MIRE PAR levels were lower than REAT levels for both earplugs, but the relationship between the two test systems was highly variable. The MIRE system had lower within-day variability than the REAT system. One individual's MIRE results were highly influential; removal of these results greatly reduced the UF for the custom-molded earplug/MIRE combination. UFs ranged from 8.8 to 13.5 dB. These findings highlight the importance of evaluating variability in individual-specific protection results for personal protective equipment like HPDs, rather than relying on single measurements.
Keywords: attenuation measurement • hearing protection • individual fit test • noise exposure • noise-induced hearing loss • variance components
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Occupational noise exposure has been recognized as a causal factor in permanent, irreversible hearing loss for several hundred years. Hearing protection devices (HPDs) are often relied upon to reduce noise exposures to acceptable levels, and, when used in the presence of training (Toivonen et al., 2002), feedback (Zohar et al., 1980) and enforcement (Hager et al., 1982), can provide effective protection to workers (Edwards and Green, 1987; Savell and Toothman, 1987; Bruhl and Ivarsson, 1994). Several factors contribute to the effective protection provided by an HPD. The first is the amount of exposure time during which the HPD is actually worn, which is driven by a wide variety of determinants (Lusk et al., 1994, 1995). Equations are available which allow for estimation of the effective protection achieved by an HPD user given the user's HPD use time during exposure (Arezes and Miguel, 2002). The second factor is the amount of noise attenuation that the HPD provides when worn. All hearing protectors sold in the US are required to be labeled with a noise reduction rating (NRR)—an estimate of the amount of attenuation that a particular hearing protector provides when fitted on trained users by an experimenter in a laboratory setting. The test procedures on which NRR calculations are made, which are specified by the US Environmental Protection Agency (EPA, 1979), are based on a 30-year-old ANSI standard (ANSI, 1974) which ANSI has since rescinded. The amount of protection received by hearing protection users in typical industrial environments is consistently found to be lower—and, in some cases, much lower—than the labeled NRR (Behar, 1985; Casali and Park, 1991; Berger et al., 1996; Stewart, 2000).
The NRR cannot be used as a measure of field attenuation even when various derating schemes are used to account for the differences between laboratory and real-world attenuation (Franks et al., 2000). Although the NRR accounts for interindividual variability in the laboratory measurements, it does not adequately control for real-world variability in attenuation.
Attenuation ratings based on methods that test attenuation on HPDs fit by naïve subjects rather than by experimenters can provide a better estimate of real-world attenuation (Berger et al., 1996). However, measurement of attenuation on an individual in the real world should provide the best estimate of the actual attenuation achieved.
There are two primary techniques available for measuring the real-world attenuation performance of HPDs on individual workers: Real-Ear-at-Threshold (REAT) and Microphone-in-Real-Ear (MIRE) (Casali et al., 1995; Berger et al., 1996). The REAT technique makes psychophysical measurements of attenuation by evaluating audiometric hearing threshold levels on a subject with (occluded) and without (unoccluded) earplugs. The difference between the occluded and unoccluded thresholds is equivalent to the attenuation [termed insertion loss (IL) in dB] provided by the HPD. The MIRE technique makes objective measurements of attenuation through the use of either one microphone (placed in the ear canal during separate measurements with and without an HPD inserted) or two microphones (one placed inside the ear canal underneath an earplug and the other simultaneously placed just outside the ear). In the former approach, the attenuation provided by an HPD is the difference in the noise level in the ear canal with and without the HPD inserted, and is termed IL (in dB), while in the latter approach, the attenuation is the difference between the sound levels measured simultaneously by the internal and external microphones, and is termed noise reduction, NR (also in dB). NR levels differ from IL measurements by factors which represent the transfer function of the open ear (TFOE) (e.g. IL = NR + TFOE) (Berger, 1986), and which are frequency-, ear- and subject-specific (Casali et al., 1995). Individual-specific TFOE factors can only be measured with IL MIRE measurements, although estimated TFOE factors for ears in free-field noise are available (ISO, 2002). Although the MIRE and REAT techniques are both intended to measure the same phenomenon, e.g. attenuation provided by a hearing protector worn by an individual, they produce different estimates of attenuation, as MIRE test values do not account for bone conduction (Casali et al., 1995; Berger, 2005) and are not subject to the effects of physiological noise owing to the occlusion effect (Berger and Kerivan, 1983). It has been suggested that the differences in measurements made with the REAT or MIRE NR methods are small enough that either methodology can be utilized to estimate attenuation (Casali et al., 1995), but this notion has not been evaluated in a real-world setting.
One REAT measurement system, FitCheck (Michael and Associates, State College, PA) has been evaluated previously—first in its prototype, audiometer-based form in a field setting (Lempert and Edwards, 1983; Edwards and Green, 1987), and then, approximately two decades later, in a computer-based commercial form in a laboratory setting (Franks et al., 2003). Although the configuration of the FitCheck system changed between the prototype and the commercially available version, the attenuation measurement techniques and results produced by the system have remained consistent over time. A MIRE measurement system, FlashTest (Custom Protect Ear, Surrey, BC), which uses the NR method, has also been developed recently. The current study evaluated these two attenuation measurement systems using both a foam earplug and a custom-molded earplug. The study had three goals. The first was to evaluate the real-world performance of the two attenuation measurement systems. The second was to assess the amount of protection provided by the two different earplugs, and compare this with the labeled attenuation for each protector. The third was to assess whether a single measurement of attenuation could be used to determine the amount of protection a worker will receive from a hearing protector over time.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Overview
All data were collected at a corrugated packaging plant in Washington state that employs
100 production and salaried workers. Subjects were selected from among volunteers in the plant workforce, approached by research staff to participate in the study, and asked to sign an informed consent form prior to participation. All research protocols were reviewed and approved by the University of Washington Institutional Review Board. Subjects were randomly assigned to four test groups of five workers each to facilitate testing.
Hearing protectors
The hearing protectors tested in this study were an expandable foam earplug (E-A-R Classic, 29 dB labeled NRR, Aearo Corporation, Indianapolis, IN) and a custom-molded silicone earplug (dB Blocker Vented, 24 dB labeled NRR, Custom Protect Ear, Surrey, BC). Custom-molded earmold impressions were taken at subject enrollment by a single experienced employee of the custom-molded earplug manufacturer. Subjects received a brief (several minute) training from research personnel on the proper use of each type of hearing protector on the first day they were issued that protector. This training took place in a quiet area of the facility, and included explanation of how to wear the earplug correctly, when and where to wear the earplug in the facility and a demonstration of how to properly insert the protector. Subjects then demonstrated to the researcher their ability to properly fit the protectors. Subjects were directed to lubricate their custom-molded earplugs several times daily during the first 5 days of wear, as per the manufacturer's instructions, with petroleum jelly provided with the earplugs. Compliance with these directions was not monitored, as the purpose of this study was to measure the real-world attenuation. During testing, subjects fit their hearing protectors as they normally would, without instruction or guidance.
Both the foam and custom-molded earplugs tested required a vent that passed through the earplug to allow for MIRE attenuation measurements. This vent allowed a microphone to be inserted into the earplug to measure sound pressure levels (SPLs) inside the ear canal while the plug was worn. The custom-molded earplugs tested include a vented bore into which different types of filters can be inserted for attenuation adjustment. When subjects wore custom-molded earplugs but attenuation was not tested, a green filter (the highest attenuation filter available from the manufacturer of the earplug) was left in the vent. The filter was removed from the vent and a microphone inserted during attenuation tests (Fig. 1). Filters were replaced in the custom-molded earplugs before subjects returned to work following testing.
|
When subjects wore foam earplugs during work periods, an off-the-shelf version of this earplug was worn. During attenuation tests on the foam earplugs E-A-R Classic earplugs fitted with a 1/16 inch inner diameter hollow plastic core running the length of the earplug through the centerline (supplied by Michael and Associates, State College, PA; Fig. 2) were used. The microphone was inserted into this vented bore before testing. After insertion into the earplugs the MIRE microphone tip was flush with the vent opening inside the ear canal. The microphone body, which was very closely matched to the diameter of the earplug vents and lubricated with petroleum jelly for insertion, was assumed to provide a complete acoustic seal of the vent when inserted, and the attenuation of the vented earplug when the vent was blocked was assumed to be identical to that of an unvented foam earplug. Vented foam earplugs were only worn during testing, and were never worn during noise-exposed work periods.
|
Attenuation measurement systems
Two systems were used in this study to make monaural attenuation measurements. The REAT system, FitCheck, consisted of the FitCheck hardware box and a laptop PC running the proprietary FitCheck software. A set of superaural headphones containing speakers delivered the sound stimulus (generated by the PC's sound card) to subjects, who used a switch connected to the FitCheck hardware box to indicate when they detected the sound stimulus. Occluded and unoccluded thresholds were measured via automatic audiometric testing (also known as Bekesy audiometry) with 1/3 octave band pulsed sound stimuli delivered in seven bands centered at 250, 500, 1000, 2000, 4000, 6300 and 8000 Hz. Subjects taking the test depressed and held the switch until the sound stimulus became inaudible, and then released the switch, which caused the sound stimulus to increase in amplitude. Three cycles of increasing to decreasing amplitude were required at each frequency, and amplitude changes occurred in steps of 1.5 dB s–1. The stimulus had rise and decay times of 25 ms, an on time of 250 ms, and an interstimulus interval of 200 ms. The system cannot calculate attenuation at frequencies where subjects have inconsistent thresholds, or for frequencies at which the maximum or minimum output of the system is exceeded. These limitations can result in missing attenuation level data at the affected frequencies. FitCheck system output was a database of attenuation levels (e.g. ILs in dB) for each frequency tested on each subject's individual ears. One FitCheck test, which included occluded and unoccluded thresholds in all seven frequency bands on one ear, took 15–25 min.
The second system, FlashTest, consisted of a laptop PC connected to a Creative Labs SoundBlaster Model S80300 [GenBank] external sound card and an Altec Lansing VS2121 speaker system. Two Knowles FG-3652-P16 over-molded microphones (one for measurements underneath earplugs and the other for simultaneous external measurements) were connected to a powered amplifier which provided 20 dB gain, which was in turn connected to the external sound card. Broadband white noise test stimuli were generated on the PC using the program ATSpec Pro v2.2 (http://www.taquis.com/atspec.htm, last accessed on October 15, 2003) and broadcast through the speaker system, creating a SPL of about 75 dB at the test subjects' position 2 m from the speakers. ATSpec Pro was also used as a sound analyzer and configured to analyze the same 1/3 octave bands measured with FitCheck using a sampling rate of 44 kHz and a sampling time of 30 s. For each FlashTest, the internal microphone was inserted into an earplug that the subject then fit into their ear. The external microphone was mounted in a machined aluminum cylinder that sat on the same shoulder as the test ear. Prior to testing, microphone output was plotted in ATSpec Pro to confirm that the output levels were identical when both microphones were in the same free field of sound. ATSpec Pro converted the internal and external microphone inputs (in Volts) at each frequency band to sound pressure (in Pascals), and the sound pressure values were then exported to an MS Excel file, at which point the difference in sound pressure (e.g. NR) at each frequency band was computed. These individual attenuation values were not corrected for TFOE, as the NR method used did not allow for measurement of TFOE on individual subjects. The frequency-specific attenuation values were converted to SPLs (in dB) and plotted. The resulting graph was visually examined for irregularities and unusual attenuation patterns. A single FlashTest took 3–5 min.
Pilot test
A pilot test was conducted on 20 subjects from January to April 2004. Subjects completed full-shift noise exposure measurements, FitCheck and FlashTest attenuation measurements on the two HPDs, and questionnaires regarding their perceptions of both protectors. A microphone output problem prevented the use or analysis of FlashTest data from the pilot test. Subjects received a new set of custom-molded earplugs during the test. Each subject participated in the pilot test over 10 consecutive work days (5 days for each earplug). Subjects wore each earplug for 4 days to allow them to adjust the protector, and earplug attenuation was tested on the fifth day. The foam earplugs were worn on days 1–5, and the custom-molded earplugs were worn on days 6–10. On the two test days, subjects received one right-ear and one left-ear FitCheck test, then one FlashTest test and completed one protector-specific questionnaire. The questionnaires asked workers about use time, comfort, convenience of insertion, perceived protection and overall perception of the hearing protectors, and included 14 questions with five-point Likert-scale responses. During FitCheck testing, the FlashTest microphones were placed in the vents of the tested earplugs to provide an acoustic seal. Three full-shift dosimetry measurements (Quest Technologies Q-300, Oconomowoc, WI) were made on each subject to assess their noise exposure according to the Washington state Permissible Exposure Limit (PEL) of 50% dose, or 85 dBA (WISHA, 2003). Dosimeters were configured to measure A-weighted decibels (dBA) using a criterion level of 90 dBA, a threshold level of 80 dBA; slow response, and a 5 dB exchange rate, as required by the Washington state regulation, and yielded an 8 h time weighted average (TWA) for each full-shift measurement. Dosimeter microphones were mounted at the top of the subject's shoulder on the same side as the subject's dominant hand. All dosimeters were calibrated pre- and post-shift, and measurements with calibration values outside of the acceptable range specified by the manufacturer were discarded.
Survey
FitCheck and FlashTest attenuation data from the two hearing protectors were collected on 20 subjects from March to September 2005. Seventeen of these subjects participated in the pilot test. All subjects were issued new custom-molded earplugs. Each subject was tested repeatedly with both systems over two consecutive workdays, with tests occurring at approximately the same time on both days. On day 1, each subject received two right-ear FitCheck tests and two right-ear and one left-ear FlashTest tests for each earplug, for a day 1 total of four FitCheck and six FlashTest tests. On day 2, each subject received two right-ear FitCheck tests and two right-ear FlashTest tests for each earplug, for a day 2 total of four FitCheck and four FlashTest tests. On both days, the foam earplugs were tested prior to the custom-molded earplugs. Only right-ear test data are reported here. The testing sequence was designed so that subjects were tested repeatedly for each earplug with both test systems without having to remove and refit their hearing protectors. This was done using the following testing sequence for each earplug: unoccluded and occluded FitCheck tests, occluded FlashTest tests, and then occluded and unoccluded FitCheck tests. The FlashTest MIRE microphones were inserted into each subject's earplugs at the start of each testing sequence to seal the earplug's vent for the FitCheck test. Prior to the custom-molded earplug tests, the researcher removed the filters in the vented bores of each subject's earplugs and inserted the MIRE microphones into the vent; subjects then fit the earplugs for MIRE and REAT testing. For the foam earplugs, researchers inserted the MIRE microphones into the bore of the special vented foam earplugs used for MIRE testing, and the subjects then fit the vented earplugs for MIRE and REAT testing. This testing protocol was selected to allow for the determination of between-subject variation in attenuation, between-day variation within subject (by testing on two different days) and within-day variation (by repeating tests on the same day on the same subject without refitting the earplugs).
Background noise during attenuation measurements
Attenuation testing took place in the quietest areas available in the participating facility, which were a conference room and a quality control room. Because both attenuation test systems are somewhat sensitive to background noise, noise levels in the test rooms were monitored during the pilot test using a SVAN 912AE Type I frequency analyzer. The levels were compliant with the US Occupational Safety and Health Administration requirements (OSHA, 1983) for audiometric testing in a hearing conservation program at 2000, 4000 and 8000 Hz, but exceeded the limits at 500 and 1000 Hz by 13 dB. Background levels did not meet the more stringent requirements of the American National Standards Institute (ANSI, 1999) at any test frequency. It is possible that unoccluded thresholds measured by FitCheck were affected slightly as a result of masking by the levels of background noise present. However, given the attenuation of the FitCheck headphones, which attenuate external sounds by 20 dB or more at and above 125 Hz according to the manufacturer's specifications, and which were fit on subjects by research staff, this is unlikely. FlashTest results should be stable in environments with steady-state background noise, which was the case in the test rooms used.
Data management and analysis
Hearing protector questionnaire responses were assigned numeric values of one to five, and total numeric scores were computed for topically grouped questions. Higher scores indicated more positive worker perceptions of specific HPD parameters. Frequency-specific FitCheck attenuation data were imported into an MS Excel database by date and subject ID. Each pilot test subject had two FitCheck tests (one per earplug). Each subject had a total of four FitCheck and four FlashTest tests for each survey earplug (two tests on each of two consecutive days for each earplug). Some subjects had fewer data points as a result of incomplete FitCheck data. Three FlashTest and one FitCheck measurement from the survey which had negative PAR levels (indicating an amplification of sound by the protector) were not included in the dataset. These measurement results would not be accepted in a real-world attenuation testing program, and would instead be re-tested, so it was reasonable to exclude them from analysis.
The frequency-specific attenuation levels provided by the manufacturer for each earplug were directly compared with the frequency-specific attenuation measurements made in this study for both the FlashTest and FitCheck systems. This comparison was based on the assumption that the microphone inserted into the vented bore of the earplugs during testing provided attenuation equivalent to that of an unvented earplug. If the vent was not completely sealed, the manufacturers' labeled attenuation values would exceed the nominal attenuation of the vented earplugs. Personal attenuation ratings (PARs) were calculated across all seven test frequencies for every FlashTest and FitCheck test. PARs were only calculated for those tests for which data were available at all seven frequencies. The method by which a PAR is calculated is similar, but not identical to, the method used to calculate an NRR. NRRs are based on multiple measurements on a number of individuals in a known sound environment under highly controlled laboratory conditions, with the hearing protector fit by an experimenter, and with two standard deviations subtracted from the mean frequency-specific attenuation values to account for variability in fit (ANSI, 1974). PARs are based on a single measurement on a test subject in a real-world (or laboratory) setting, with hearing protectors fit by subjects and no adjustment for variability (i.e. no standard deviation correction). ‘Adjusted NRRs’ were therefore calculated for each earplug using a modified version of the attenuation computation method in ANSI S3.19-1974 (ANSI, 1974), which excluded two frequencies (125 and 3150 Hz) required by ANSI but not included in the FitCheck system measurements. PAR levels were calculated for each subject using the same modified ANSI method, but without adjusting for inter-subject variability, as done for the NRR. With the frequencies of 125 and 3150 Hz excluded, PAR levels and the adjusted NRR for each earplug were sufficiently similar for direct comparison. The adjusted NRR for the foam earplug was identical to the labeled NRR (29 dB), while the adjusted NRR of 23 dB for the custom-molded earplug was 1 dB lower than the labeled NRR.
All statistical analyses were conducted using Intercooled Stata 9.0 (Stata Corporation, College Station, TX). Descriptive statistics were computed for noise exposure and attenuation data. t-Tests were performed on the mean FitCheck PAR data from the pilot test and the survey to assess differences in mean levels. The percent of the adjusted NRR achieved by the PAR for each attenuation measurement was calculated. t-Tests were also conducted to compare whether the PAR levels for the two earplugs differed from the labeled NRR levels, and whether attenuation levels for each earplug differed significantly by test system. Estimated TFOE correction factors (ISO, 2002) were applied to the group mean FlashTest frequency-specific attenuation levels to determine how inclusion of these factors affected the relationship between the FlashTest and FitCheck PAR levels for the two earplugs.
A random effects analysis of variance was conducted (Stata ‘xtmixed’ command) to partition the total variability in PARs into ‘between-subject’ variability, 'within-subject' (between-day) variability and ‘within-day’ (within subject) variability. The between-subject variability (
bs) represents the differences in attenuation achieved by each subject. The within-subject variability (or between fittings, bf) separates the effect of plug refitting from random errors in the measurement system. The within-day variability (or within fitting, wf) then represents only the changes in the measurement of a PAR made on the same plug, in the same ear, without any refitting. The random effects model was run for each combination of the two test systems and two earplugs being evaluated after exclusion of the four tests with negative PAR levels and tests with incomplete frequency-specific attenuation data. The model was also run for the custom-molded earplug/FlashTest combination after excluding data from Subject 5, who had extremely variable PARs (range >30 dB).
The 95% lower confidence limit of the within subject variability in PAR (i.e. the sum of the between fitting and within fitting variability) was calculated, and is referred to as an ‘uncertainty factor’ (UF). The UF is thus the number of decibels by which one would ‘discount’ an individual's measured PAR in order to obtain the amount of attenuation that individual would receive with 95% and was calculated using the following equation:
 | (1) |
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Full-shift noise measurements
One dosimetry sample did not calibrate correctly post-measurement and was discarded. The 59 remaining successful dosimetry measurements (Table 1) had an average TWA noise level of 86.0 ± 6.1 dBA. Mean levels varied from 81 to 88 dBA by job classification. More than two-thirds of work shifts had levels above 85 dBA, and nearly one-third exceeded 90 dBA.
|
Hearing protection questionnaire data
All subjects reported using both HPDs 100% of their work shift on test days. The custom-molded earplug had the highest average score (out of 5 points possible) for 12 of the 14 questionnaire items. It also had the highest score for comfort (4.2 ± 0.8 out of 5 possible, versus 3.0 ± 0.8 for the foam earplug), perceived protection (4.1 ± 0.9 versus 3.6 ± 1.0) and overall rating (4.3 ± 0.9 versus 3.3 ± 1.0), and tied the foam earplug for the highest score for convenience (3.7 ± 1.0).
Attenuation measurements
Figure 3 shows a graph of the mean and SD frequency-specific attenuation values for the foam earplug. Each frequency has 79 FlashTest measurements and between 73 and 79 FitCheck measurements; the lower number of FitCheck tests are a result of missing data. FitCheck and FlashTest results for this earplug were roughly parallel below 2000 Hz, but diverged at higher frequencies. The FitCheck tests showed 5–10 dB more attenuation than the FlashTest at all frequencies except 4000 and 6300 Hz. The FlashTest and FitCheck results had similar variability below 2000 Hz, and above 4000 Hz FlashTest had much lower variability. Both systems yielded larger standard deviations than the labeled values associated with the NRR. Neither the mean FlashTest nor FitCheck results ever exceeded the labeled attenuation levels.
|
Figure 4 shows a graph of the frequency-specific attenuation for the custom-molded earplug. Each frequency has 78 FlashTest measurements and 76–79 FitCheck measurements. The FlashTest and FitCheck results were relatively dissimilar, with mean FlashTest attenuation steadily increasing from 250 to 6300 Hz, and FitCheck attenuation having relatively uniform (‘flat’) attenuation across all frequencies, with the exception of 2000 and 8000 Hz. FlashTest variability was always lower than that of FitCheck, and substantially lower at 4000 Hz and above. Neither the FlashTest nor FitCheck mean attenuation levels ever exceeded the labeled values, though the FitCheck levels were very close at 2000 Hz and below. The standard deviations for both systems were always larger than those associated with the NRR.
|
Table 2 provides a summary of right-ear PAR levels and percent of adjusted NRR for both earplugs and both test systems. Tests in which the FitCheck data were incomplete are excluded. Average PARs were lower than the adjusted NRR for each plug/system combination, except for the custom-molded plug tested with the FitCheck system. For this set of tests, the mean PAR was significantly higher than the adjusted NRR (27.5 ± 8.3 versus 23 dB, P < 0.001). For both earplugs, the FitCheck produced a higher average PAR than did the FlashTest (two sample t-test, P < 0.001 for both foam and custom-molded earplugs). The custom-molded earplugs achieved greater attenuation than did the foam earplug (two-sample t-test, P = 0.003 and P < 0.001 for FlashTest and FitCheck, respectively) with both test systems. The overall variability associated with each plug and test system was similar, with SDs from 6.5 to 8.3 dB. Note that these estimates of variability are the sum of the between-subject, between fit and within fit sources of variability.
|
Analysis of attenuation variability
The results from the random effects analysis of variance are shown in Table 3. Mean attenuation and 95% confidence intervals around the mean are presented for each of four test conditions (one for each test system/earplug combination). Between-subject, within-subject and within-day variability are presented by model as standard deviations, along with their 95% confidence intervals. The between- and within-subject variability values presented are relatively unstable (e.g. have wide 95% confidence intervals) owing to the small number of repeated measurements on each individual. However, they represent the best estimate possible from the available data. UFs are also presented by model.
|
For the custom-molded earplug, the largest source of variability in the FitCheck tests came from between-subjects differences, and there was essentially no within-subject variability. However, for the foam earplug and FitCheck test, the largest source of variability was from within-day differences, and the smallest variability was associated with between-subjects differences. Within-day differences contributed greatly to FitCheck variability for both earplugs, indicating that this is an important source of variability for this system. Comparison of the two systems indicated that FlashTest had somewhat less within-day variability than did FitCheck, although the reliability of these estimates is again limited by large confidence intervals.
The estimates of variability for the FlashTest results for the custom-molded earplug varied widely depending on whether one subject's data were included or excluded in the analysis. If subject 5's data were included, the primary source of variability for FlashTest came from within-subject differences, and between-subject differences contributed little to the overall variability. If the data were excluded, however, the primary source of variability was from between-subject differences, and the contributions of within-day and within-subject variability were similar. The primary source of variability for FlashTest measurements on the foam earplug came from within-subject differences (e.g. earplug refitting), followed by within-day differences.
The PAR UFs were large (10 decibels or more). The FlashTest/custom-molded earplug combination test results varied widely depending on whether one subject's data were included (13.5 dB UF) or excluded (8.8 dB UF). The FlashTest system/custom-molded earplug combination without subject 5's data had the lowest UF, while the FlashTest/foam earplug combination had the highest UF. In order to be 95% confident that a subject's under-protector exposure would be <85 dBA, their expected workplace exposure minus their individual measured attenuation would have to be between 71 and 76 dBA (depending on the test system and hearing protector used) to account for the uncertainty in measured attenuation.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Evaluation of the attenuation provided by hearing protectors worn by noise-exposed workers is a critical step in understanding and reducing overexposures to noise and ultimately preventing noise-induced hearing loss. The data presented here represent a unique opportunity to assess the attenuation offered by two hearing protectors measured using two test systems utilizing very different approaches to attenuation measurement. Sources of variability in the measured attenuation levels, including between-subject, within-subject and within-day differences, were assessed. This approach does not appear to have been taken in any previous study of hearing protection, and allowed for evaluation of the relative sources of variability for each earplug and test system. It also allowed for calculation of UFs which can be used to estimate the amount of attenuation an individual worker can achieve with 95% confidence. The UFs measured here are large, and highlight the fact that a single measurement of attenuation on an individual worker is highly imprecise.
The first issue evaluated in this study was the performance of the two attenuation measurement systems. Taking the adjusted NRR as the standard for comparison, the FitCheck system produced the best results, with the highest percentage of the adjusted NRR measured in this study (120 and 70% for the custom-molded and foam earplugs, respectively). The FlashTest system measured lower percentages of the adjusted NRR for two earplugs (78% for the custom-molded and 51% for the foam earplugs). Both systems occasionally produced negative PAR levels, a result implying that the hearing protector worn amplified rather than attenuated sound. Both systems would benefit from a real-time automated screening function that would reject measurements with a negative PAR and force worker re-training and re-testing. The FlashTest MIRE measurement system was found to have lower within-day variability than the FitCheck REAT system, as assessed by repeated tests on the same subject on the same day without refitting of their hearing protector. The FlashTest system underestimated the attenuation measured by the FitCheck system, but the variability in the attenuation levels measured by the two systems was too great to allow for the development of correction factors for estimating approximate REAT attenuation values from the MIRE tests, despite the fact that such estimation may be possible in a laboratory setting (Casali et al., 1995). The FitCheck attenuation values at the lower frequencies may have been influenced by masking from background noise during testing; this situation would have artificially increased the differences between the two test systems. The FlashTest did provide a conservative estimate of the FitCheck attenuation achieved by users, so FlashTest PARs can be viewed as the minimum attenuation that would be measured on a worker. Application of estimated free-field TFOE factors (ISO, 2002) (which range from 0 dB at 250 and 500 Hz up to 7.5 dB at 4000 Hz) to the group mean frequency-specific FlashTest attenuation levels increased the mean FlashTest PAR level for both tested earplugs. The custom-molded earplug mean PAR increased 1.4 dB to 19.3 ± 7.6 dB, while the foam earplug mean PAR increased 1.6 dB to 16.5 ± 6.7 dB. The higher TFOE-corrected mean FlashTest PAR levels reduced the difference between the mean PAR levels measured by the two test systems, but the FitCheck values remained statistically significantly higher (P < 0.01) for both earplugs even after TFOE correction. Also, correcting for TFOE did not uniformly increase agreement between frequency-specific FlashTest and FitCheck attenuation values, and in some cases (particularly at 4000 and 6300 kHz) resulted in even greater differences. Estimated TFOE factors could only be applied to the group mean frequency-specific MIRE NR attenuation levels, and not to individual results.
The primary source of variability among FlashTest attenuation levels was earplug refitting or between-subject differences (depending on whether attenuation data from one subject were included or not), while the main source of variability for the FitCheck system was within-day differences. The FlashTest measurement procedure had greater speed and efficiency than did FitCheck, taking a fraction of the time to administer and producing complete data for every test.
When the current REAT and MIRE measurement system results are compared with previous research, some differences are apparent. Casali et al. (1995) found much better agreement between laboratory REAT and TFOE-corrected measurements on earmuffs than those found here. However, the differences measured were still statistically significant, and were as great as 6 dB, depending on the test frequency. Unlike the current study, MIRE measurements made by Casali et al. exceeded REAT measurements at most frequencies. The greater disparity between MIRE and REAT measurements found here may be due at least in part to background noise effects on REAT measurements and the fact that the current study tested only earplugs, which have been repeatedly demonstrated to have greater variation in attenuation than earmuffs (Casali and Park, 1991; Berger et al., 1996; Berger, 2000). Berger has noted that REAT measurements tend to over-predict attenuation at low frequencies due to physiological masking (Berger et al., 1996), which could partially explain the difference between the MIRE and REAT measurements in the current study. Berger also noted that SDs from real-world REAT and MIRE measurements are usually similar (rather than the intuitive expectation that objective MIRE measures would have lower variability than psychophysical REAT measurements). However, in the current study, FlashTest MIRE measurements generally had lower frequency-specific SDs, with the SDs at 4000 and 6300 Hz being markedly lower. The lower variability in the FlashTest system seen in the current study is largely due to relatively low between-subject variability. By comparison, FitCheck system variability was probably higher due to the subjective nature of psychophysical tests and the inclusion of study subjects with various hearing threshold levels, which affects REAT tests, but not MIRE measurements. Hearing protector training has been shown to reduce variability measured for both MIRE and REAT techniques (Toivonen et al., 2002), probably through improvement of workers' ability to properly fit HPDs.
Previous attenuation research conducted using the FitCheck system or its prototypes has identified issues similar to those found here. Franks et al. (2003) found that the FitCheck system measured statistically significantly higher attenuation than three other REAT testing procedures, and that the basic operation of the FitCheck system may make results from the system difficult to compare with other attenuation test methods. Franks et al. did conclude that FitCheck use was feasible for field attenuation measurements, though the system demonstrated larger frequency-specific SDs than any of the other systems tested.
The second issue evaluated in this study was the attenuation performance of the two types of earplugs. The custom-molded earplugs achieved the highest mean percentage of the adjusted NRR attenuation and the lowest variability with both test systems. FitCheck measurements on the custom-molded earplugs achieved attenuation which, on average, exceeded the adjusted NRR for that earplug, while the FlashTest measurements (with a mean of 78% of the adjusted NRR) did not. Neither the mean attenuation for the FitCheck nor FlashTest measurements of the foam earplug ever exceeded more than 70% of the adjusted NRR, and the mean frequency-specific measured attenuation from both types of earplugs never exceeded the labeled attenuation of the protectors. The custom-molded earplug had higher overall acceptance among the tested workers, and was also rated higher in terms of comfort and perceived protection. Comfort (Hsu et al., 2004) and usability are critical issues in hearing protection selection, and the custom-molded earplug was consistently rated higher in these categories by the workers tested. The custom-molded and foam earplugs were rated similarly in terms of convenience; this result may have been different had the tested workers been doing work which regularly involved inserting earplugs with dirty hands. Also, the custom-molded earplug results in this study may have been different had the subjects' earmold impressions been taken by multiple (or less-experienced) employees of the custom-molded earplug manufacturer.
The earplug attenuation results of the current study generally agree with prior research. Edwards and Green (1987) and Behar (1985) both tested real-world attenuation in workers from several facilities who wore the same foam earplug tested in the current study. Subjects in both studies were found to have mean and standard deviation frequency-specific attenuation levels that were similar to those found here at test frequencies at and below 2000 Hz and at 8000 Hz. The current study found lower mean levels and larger standard deviations at 4000 and 6300 Hz. Edwards et al. also computed mean A-weighted NRs (essentially identical to a PAR), and found average levels in the range of 21.0–26.9 dB, findings that generally agree with the FitCheck PARs found here.
The third issue evaluated in this study was whether it is possible to determine whether an individual person will be adequately protected from a given noise exposure based on the results of an individual attenuation test. The repeated measurements approach used in this study allowed for determination of the variability in measured attenuation levels due to within-subject (due to earplug refitting) and within-day differences, as well as between-subject differences. Between-subject differences are not relevant in estimating individual PAR levels, but must be computed to parse out the contribution of the other sources of variability. Ability to evaluate the contribution of these sources of variability is a major strength of the current study, and no analysis of this nature regarding hearing protection appears to have been published previously. Using a statistical model to exploit the repeated measurements on each subject (across test systems and plug types) UFs were calculated which can be subtracted from an individual's PAR to estimate the minimum attenuation that person will achieve with 95% confidence. If a single PAR measurement is obtained, the UFs may be applied to the PAR measured on an individual subject from a single attenuation measurement. The UFs when all available data were considered ranged from 10.0 to 13.5 dB across the four earplug/test system combinations. The removal of one subject's FlashTest/custom-molded earplug results reduced the PAR UF for that system/earplug combination by nearly 5 dB (from 13.5 to 8.8 dB). This large variation driven by one subject's results is of concern, but is not surprising given the small number of workers assessed and instability of the modeled estimates. In practical terms, the 8.8 dB UF can only be applied to screened individual PAR results—that is, results which have been screened for large variability in PAR levels. Subjects could be considered screened when they have received several attenuation tests, with earplug refitting between each test, and have not displayed large (i.e. >2–3 dB) differences between tests. The less protective 8.8 dB UF is appropriate for these individuals, given their minimal variability in attenuation due to earplug refitting. If test results are not screened in this fashion, or if subjects demonstrate high variability or incompetence in fitting their earplug, use of the larger UF (13.5 dB) for this system would be necessary. To the degree that the measured group of blue-collar workers is similar to other groups of noise-exposed workers, these UFs are valid and may be generalized.
Seventeen subjects participated in both the pilot test and survey, and 14 of these had valid right-ear FitCheck PARs for both the custom-molded and foam earplugs from both data collection periods. The FitCheck attenuation results for these subjects were compared using a t-test. The mean custom-molded earplug PAR was significantly higher (27.5 ± 8.3 dB versus 19.8 ± 10.1 dB, P < 0.05) in the survey compared with the pilot test, and the difference between the mean foam earplug PAR levels from the survey and the higher levels from the pilot test was of borderline significance (20.4 ± 6.5 dB versus 24.1 ± 11.5 dB, P = 0.05). This REAT test–retest difference has been noted by other authors, as well (Edwards and Green, 1987). One possible explanation for the difference is the ‘learning effect’, whereby the performance of subjects in REAT-type tests improves over time as subjects become more familiar with the test protocol. However, a true learning effect should have resulted in an increase in PAR levels for both earplugs between the pilot test and the survey, rather than an increase in attenuation for one and a decrease for the other. Lempert and Edwards (1983) found that earplug attenuation measurements taken with a predecessor of FitCheck had statistically significant differences over time, with the results of the first of 5 days of testing being significantly different than those of subsequent tests. The authors attributed this difference to diminishing motivation to achieve maximum attenuation among study subjects over repeated tests. The difference in the current study is probably owing to inter-survey variability, which adds to the uncertainty with which one judges the results of any single measurement of attenuation.
The variability analysis in the current study is relevant to personal protective equipment other than hearing protection. Respirators, for example, are commonly labeled with assigned protection factors (APFs) indicating concentrations of airborne contaminants from which their users can be expected to be protected. These factors are similar in concept to the NRR. As with hearing protectors, individual tests of respirator fit are more desirable than general protection factors, and a number of test systems are commercially available (Janssen et al., 2002) which are intended to measure a protection factor for an individual user based on a single measurement. Respirator fit test studies have demonstrated that the results of field measurements of protection are dependent on test methodology (Xu et al., 1991; Janssen et al., 2002) and show greater variability than laboratory measurements (Lee et al., 2005)—much like the hearing protection literature. One study that did find good agreement between two different types of respirator fit evaluation (Zhuang et al., 2003) did not account for within-subject variability, and another that found good predictive properties in several types of respirator fit test did not completely control for within-subject differences in all tests (Coffey et al., 1998). Some authors have advocated for repeated tests of respirator fit to reduce measurement error and improve confidence in the protection a respirator provides to an individual (Crutchfield et al., 1999; Campbell et al., 2001; Campbell et al., 2005), and for fewer or shorter tests for subjects who consistently demonstrate good fit (Sreenath et al., 2001). Campbell et al. (2001) explored the use of factors which summarize the ability of particular types of respirator to be fit well by workers, an approach which bears some resemblance to the current NIOSH practice of variably derating different hearing protectors (NIOSH, 1998) based partly on ability of end users to achieve good protection. Nicas and Neuhaus (2004) analyzed within- and between-subject variability in fit among several studies of several different types of respirators, and found that within-subject differences were generally a more important source of variability than between-subject differences. The authors utilized a 95% CI approach similar in some respects to that of the current study, and suggested that the protection factor assigned to a respirator should insure that both individual users and populations of users have acceptable protection factor distributions.
The large size of the attenuation UFs computed here suggests that, rather than relying on a single measurement, assessment of attenuation or other protection factors should utilize multiple measurements. For some types of systems, including REAT systems like FitCheck, the time required for repeated measurements would be great. However, other systems, such as the FlashTest MIRE system, require minimal time for repeated measurements, which is a major strength for any test system. Adoption of a repeated measures approach for hearing protectors and other protective equipment would provide a higher degree of confidence in the protection afforded to an individual. Such confidence is the ultimate goal of individual protection assessment.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The noise exposure levels measured in this study indicate that the participating facility needs to maintain its current hearing conservation program, and that use of hearing protectors will be necessary in the facility unless and until effective noise controls are implemented. Reliance on individual measures of attenuation achieved with hearing protectors, rather than on the labeled NRR of the protectors, is desirable but must account for HPD refitting and measurement system variability. Both of the tested earplugs appear to provide adequate protection for use in the facility after the application of UFs that account for variability in attenuation due to earplug refitting and within-day differences. However, to use the least protective UF presented here (i.e. to reduce by the least amount an individual's measured PAR), repeated attenuation measurements must be made to insure that the individual's measured attenuation is stable across protector refittings. Both the FitCheck and FlashTest systems may be used to make single measurements of individual's attenuation, although the FlashTest will produce more conservative estimates of attenuation than will the FitCheck. Despite this limitation, the MIRE approach used by FlashTest dramatically reduces the speed required for a test, allowing for administration of multiple tests per subject. MIRE tests do require modification (e.g. insertion of a vented bore) of a set of the earplugs worn by each tested worker to allow for attenuation testing, or use of a miniaturized microphone inserted between the earplug body and ear canal wall during measurements. These modifications may introduce some additional error into attenuation measurements. Nevertheless, administration of repeated tests on each subject will result in more reliable estimates of an individual worker's attenuation, and increased confidence that noise-induced hearing loss will be prevented.
Received January 25, 2006; in final form April 4, 2006
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
ANSI S3.19-1974. (1974) American National Standard: method for the measurement of real-ear protection of hearing protectors and physical attenuation of ear muffs. (American National Standards Institute, Inc, New York, NY).
ANSI S3.1-1999. (1999) American National Standard: maximum permissible ambient noise levels for audiometric test rooms. (American National Standards Institute, Inc, New York, NY).
Arezes P and Miguel A. (2002) Hearing protectors acceptability in noisy environments. Ann Occup Hyg 46:531–6.
Behar A. (1985) Field evaluation of hearing protectors. Noise Control Eng J 24:13–8.
Berger E. (2005) Preferred methods to measuring hearing protection attenuation. (International Institute of Noise Control Engineering, Brazil)10 In: 2005 congress and exposition on noise control engineering; August 7–10, 2005. Rio de Janeiro.
Berger EH. (1986) Methods of measuring the attenuation of hearing protection devices. J Acoust Soc Am 79:1655–87.[CrossRef][Web of Science][Medline]
Berger EH. (2000) Chapter 10: Hearing protection devices. In Berger EH, Royster LH, Royster JD, Driscoll DP, Layne ML (Eds.). The noise manual 5th edn. (American Industrial Hygiene Association, US) pp. 379–454.
Berger EH and Kerivan JE. (1983) Influence of physiological noise and the occlusion effect on the measurement of real-ear attenuation at threshold. J Acoust Soc Am 74:81–94.[CrossRef][Web of Science][Medline]
Berger EH, Franks JR, Lindgren F. (1996) International review of field studies of hearing protector attenuation. In Axlesson A, Borchgrevink H, Hamernik RP, Hellstrom P, Henderson D, Salvi RJ (Eds.). Scientific basis of noise-induced hearing loss(Thieme Medical Publishers, Inc., New York, NY) pp. 361–77.
Bruhl P and Ivarsson A. (1994) Noise-exposed male sheet-metal workers using hearing protectors. A longitudinal study of hearing threshold shifts covering fifteen years. Scand Audiol 23:123–8.[Web of Science][Medline]
Campbell DL, Coffey CC, Lenhart SW. (2001) Respiratory protection as a function of respirator fitting characteristics and fit-test accuracy. AIHAJ 62:36–44.[Web of Science][Medline]
Campbell DL, Coffey CC, Jensen PA, et al. (2005) Reducing respirator fit test errors: a multi-donning approach. J Occup Environ Hyg 2:391–9.[CrossRef][Web of Science][Medline]
Casali J and Park M. (1991) Laboratory versus field attenuation of selected hearing protectors. J Sound Vib 25:28–38.
Casali J, Mauney D, Burks JA. (1995) Physical vs. psychophysical measurement of hearing protector attenuation—a.k.a. MIRE vs. REAT. J Sound Vib 29:20–7.
Coffey CC, Campbell DL, Myers WR, et al. (1998) Comparison of six respirator fit-test methods with an actual measurement of exposure in a simulated health care environment: Part II—method comparison testing. Am Ind Hyg Assoc J 59:862–70.[Web of Science][Medline]
Crutchfield CD, Fairbank EO, Greenstein SL. (1999) Effect of test exercises and mask donning on measured respirator fit. Appl Occup Environ Hyg 14:827–37.[CrossRef][Medline]
Edwards R and Green W. (1987) Effect of an improved hearing conservation program on earplug performance in the workplace. Noise Control Eng J 28:55–65.
EPA. (1979) Noise labeling requirements for hearing protectors. Fed Regist 44(190)(Environmental Protection Agency, Washington, DC) pp. 130–56 147.
Franks JR, Murphy WJ, Johnson JL, et al. (2000) Four earplugs in search of a rating system. Ear Hear 21:218–26.[CrossRef][Web of Science][Medline]
Franks JR, Murphy WJ, Harris DA, et al. (2003) Alternative field methods for measuring hearing protector performance. AIHA J (Fairfax, Va) 64:501–9.
Hager WL, Hoyle ER, Hermann ER. (1982) Efficacy of enforcement in an industrial hearing conservation program. Am Ind Hyg Assoc J 43:455–65.[Web of Science][Medline]
Hsu Y, Huang C, Yo C, et al. (2004) Comfort evaluation of hearing protection. Ind Ergo 33:543–51.[CrossRef]
ISO. (2002) ISO Standard 11904-1:2002(E), Acoustics—determination of sound immission from sound sources placed close to the ear. Part 1: technique using a microphone in a real ear (MIRE technique). (International Organization for Standardization, Geneva, Switzerland).
Janssen LL, Luinenburg DM, Mullins HE, et al. (2002) Comparison of three commercially available fit-test methods. AIHA J (Fairfax, Va) 63:762–7.
Lee SA, Grinshpun SA, Adhikari A, et al. (2005) Laboratory and field evaluation of a new personal sampling system for assessing the protection provided by the N95 filtering facepiece respirators against particles. Ann Occup Hyg 49:245–57.
Lempert BL and Edwards RG. (1983) Field investigations of noise reduction afforded by insert-type hearing protectors. Am Ind Hyg Assoc J 44:894–902.[Web of Science][Medline]
Lusk SL, Ronis DL, Kerr MJ, et al. (1994) Test of the health promotion model as a causal model of workers' use of hearing protection. Nurs Res 43:151–7.[Web of Science][Medline]
Lusk SL, Ronis DL, Kerr MJ. (1995) Predictors of hearing protection use among workers: implications for training programs. Hum Factors 37:635–40.[Web of Science][Medline]
Nicas M and Neuhaus J. (2004) Variability in respiratory protection and the assigned protection factor. J Occup Environ Hyg 1:99–109.[CrossRef][Web of Science][Medline]
NIOSH. (1998) Criteria for a recommended standard: occupational noise exposure, revised criteria 1998. Cincinnati, OH: US Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health. June 1998. 98–126 Report No.: DHHS (NIOSH).
OSHA. (1983) Occupational noise exposure: hearing conservation amendment; final rule. (US Department of Labor, Occupational Safety and Health Administration, Washington, DC).
Savell JF and Toothman EH. (1987) Group mean hearing threshold changes in a noise-exposed industrial population using personal hearing protectors. Am Ind Hyg Assoc J 48:23–7.[Web of Science][Medline]
Sreenath A, Weed J, Church T. (2001) A modified protocol for quantitative fit testing using the PortaCount. Appl Occup Environ Hyg 16:979–88.[CrossRef][Medline]
Stewart A. (2000) Chapter 6: Program overview and administration. In Berger EH, Royster LH, Royster JD, Driscoll DP, Layne ML (Eds.). The noise manual 5th edn. (American Industrial Hygiene Association, US) pp. 150–64.
Toivonen M, Paakkonen R, Savolainen S, et al. (2002) Noise attenuation and proper insertion of earplugs into ear canals. Ann Occup Hyg 46:527–30.
WISHA. (2003) Hearing loss prevention (noise), WAC 296-817. (Washington Department of Labor and Industries, Washington Industrial Safety and Health Administration, Olympia, WA).
Xu M, Han D, Hangal S, et al. (1991) Respirator fit and protection through determination of air and particle leakage. Ann Occup Hyg 35:13–24.
Zhuang Z, Coffey CC, Jensen PA, et al. (2003) Correlation between quantitative fit factors and workplace protection factors measured in actual workplace environments at a steel foundry. AIHA J (Fairfax, Va) 64:730–8.
Zohar D, Cohen A, Azar N. (1980) Promoting increased use of ear protectors in noise through information feedback. Hum Factors 22:69–79.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||