Measurements of Earplug Attenuation Under Supra-Aural and Circumaural Headphones

International Journal of Audiology 2012; 51: 730–738

Original Article

Measurements of earplug attenuation under supra-aural and circumaural headphones

Jennifer B. Tufts* , Jillian V. Palmer * & Lynne Marshall†

*Department of Communication Sciences, University of Connecticut, Storrs, USA, and †Naval Submarine Medical Research Laboratory, Groton, Connecticut, USA

Abstract Objective: Supra-aural audiometric headphones are generally not recommended for use in measuring the attenuation of earplugs, because contact between the headphone and pinna and/or earplug could alter the attenuation obtained, and because of concerns of non-comparability between modes of excitation from supra-aural headphones and the sound-fi eld procedure required by the standardized method. In this study, we compared measurements of earplug attenuation obtained under Telephonics TDH-50P supra-aural headphones with measurements obtained under circumaural headphones designed expressly for such testing. Design: The attenuation of three types of earplugs (foam, premolded quadruple-fl ange, and custom-molded) was measured in a repeated-measures design. Study sample: The study sample comprised 42 normal-hearing adults (21 females, 21 males). Results: With the foam earplugs, nearly all of the attenuation measurements under the supra-aural headphones fell within 10 dB of the measurements under the circumaural headphones. With the fl ange and custom earplugs, approximately 10% of individuals obtained spuriously high attenuation under the supra-aural headphones. Conclusions: We conclude that standard supra-aural audiometric headphones are suitable for measuring the attenuation provided by foam earplugs. However, supra-aural headphones should not be used to measure the attenuation of fl ange or custom-molded earplugs. The potential exists for substantial over-estimation of attenuation, especially of custom plugs.

Key Words: Hearing protection; HPD; earplugs; attenuation; ﬁ t-testing; supra-aural; circumaural; headphones; TDH

For personal use only. The amount of protection one receives from hearing protection and attenuation needs. Fit-testing can also provide documentation of devices (HPDs) cannot be known with certainty unless the amount HPD adequacy in cases of possible noise-induced threshold shift, for of attenuation is measured while the HPDs are being worn. HPD defense against workers ’ compensation claims, for compliance with packaging is labeled with an attenuation rating, such as the Noise federal or state regulations, and for assessment of overall HPD effec- Reduction Rating (NRR; used in the U.S.A.) or the Single Number tiveness in a hearing conservation program. Methods of HPD ﬁ t- Rating (SNR; used in the EU). However, it has long been known that testing include microphone-in-real-ear (MIRE) procedures, loudness these laboratory-based ratings do not correspond to the attenuation balancing procedures, and real-ear-attenuation-at-threshold (REAT) achieved by typical end-users in real-world environments. The NRR, procedures (Berger et al, 2008; Hager, 2011; Schulz, 2011). MIRE for example, notoriously overestimates the attenuation achieved by and loudness-balancing procedures are described elsewhere in the most end-users, especially with earplugs (Berger, 2000). For this literature and are not discussed in this paper. reason, both the U.S. Occupational Safety and Health Administration The REAT procedure is documented in past and current ANSI

Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12 and the U.S. National Institute for Occupational Safety and Health standards (ANSI 12.6) and is considered the “ gold standard” for recommend that the NRR be derated when it is used to estimate measuring attenuation (Berger, 1986; ANSI S12.6-2008). Briefl y, in the adequacy of hearing protectors in particular noise environments. the REAT procedure, a listener’ s hearing thresholds are measured Derating provides a more realistic estimate of the amount of attenu- in the sound fi eld with and without his/her ears occluded by HPDs. ation that an individual is likely to achieve, but it is not a substitute The difference between the “ open ” and “ occluded ” thresholds at a for individual fi t-testing. particular frequency is the amount of attenuation at that frequency. Individual HPD fi t-testing measures the attenuation achieved Attenuation per frequency can be reported, or the attenuation values by an individual wearing a particular HPD in a particular way at a at the different test frequencies can be used to compute a single- particular time. Fit-testing has several applications, as discussed by number measure of attenuation such as an NRR. When fi eld mea- Berger (2006), Berger et al (2008), Hager (2011), and Schulz (2011). surements are conducted on individuals, the data at the individual It can be used to help train employees to wear their HPDs correctly, test frequencies may be summarized in terms of an analogous single- to help trainers improve the fi t for an employee, and to guide the number value for the individual often called a personal attenuation assignment of HPDs to an individual based on his/her noise exposure rating (PAR). There is no standardized way to calculate a PAR; one

Correspondence: Jennifer Tufts, Department of Communication Sciences, University of Connecticut, 850 Bolton Road, Unit 1085, Storrs, CT 06269-1085, USA. E-mail: jennifer. [email protected]

(Received 2 September 2011 ; accepted 18 May 2012 ) ISSN 1499-2027 print/ISSN 1708-8186 online © 2012 British Society of Audiology, International Society of Audiology, and Nordic Audiological Society DOI: 10.3109/14992027.2012.696217 Earplug attenuation under headphones 731

fi ttings of four earplug types (foam, two types of premolded, and Abbreviations custom-molded). Differences between headphone and sound-fi eld HPD Hearing protection device measurements of 10 dB or less were considered acceptable, given NRR Noise reduction rating the variability inherent in hearing threshold level measurement. Of PAR Personal attenuation rating the 32 headphone-sound fi eld comparisons at 500 Hz (four sub- REAT Real-ear-attenuation-at-threshold jects four earplug types two fi ttings), 24 (i.e. 75%) met this SNR Single number rating criterion. In four of the remaining eight cases, the supra-aural head- TDH Telephonics Dynamic Headphone phone attenuation was more than 10 dB higher than the sound-fi eld attenuation. Those points were of concern, because they represented situations in which the supra-aural headphone REAT considerably overestimated the true REAT. They included two improper fi ts of published method is described later in this paper. The PAR is simi- one of the premolded plugs, an improper fi t of the foam plug, and lar in some respects to the well-known NRR. One major difference a proper fi t of the custom earplug. Berger (1984) stated that the between the two metrics is that the PAR represents the attenuation agreement between supra-aural headphone REAT and sound-fi eld achieved by an individual, whereas the NRR is an average across a REAT was better than might have been expected given the concerns group of test subjects, minus two standard deviations. listed above. Nevertheless, for some individuals, REAT testing with While REAT testing in the sound fi eld according to ANSI 12.6- supra-aural headphones may yield large overestimates of earplug 2008 is the gold standard, REAT can also be done under headphones. attenuation. Despite the promising results, the study was too small Testing under headphones is desirable when noise levels in the envi- to draw fi rm conclusions about the suitability of supra-aural head- ronment preclude testing with loudspeakers in a sound fi eld or when phones for use in REAT testing. a sound-fi eld set-up is not available or convenient. Such a scenario In the current study, we compared REAT under supra-aural head- might occur if REAT testing is done onsite as part of a workplace phones to REAT under circumaural headphones. In real-world fi t- hearing conservation program, for example. With headphone REAT, testing settings, REAT with circumaural headphones is accepted as of course, earplugs are the only type of HPD that can be evaluated. a proxy for sound-fi eld REAT. The comparison between supra-aural It should be noted that headphone REAT and sound-fi eld REAT yield and circumaural measurements is therefore a comparison between similar, though not identical, attenuation values. (1) a convenient and accessible headphone REAT paradigm that is Headphone REAT systems are commercially available (Hager, often discredited (supra-aural), and (2) an accepted headphone REAT 2011). The headphones used with these systems are circumaural, paradigm that is less widely available. We used identical threshold- i.e. designed to rest on areas of the skull surrounding the pinna, testing procedures across test conditions to isolate the effect of the with deep earcups to prevent contact with the earplug. Despite the structure and human interface of these two headphones. Our test availability of circumaural headphone REAT systems, their use is subjects were not highly trained and could be considered typical of not widespread at this time. Possible reasons could include the some- the general population of HPD users. For personal use only. what time-consuming nature of the REAT test or the expense of the Overall, the current study addressed the following questions: To fi t-testing system. what extent do measurements of earplug attenuation under supra- Nevertheless, audiologists and occupational hearing conservation- aural headphones differ from measurements under circumaural ists may wish to evaluate the attenuation of earplugs even if a fi t- headphones? Second, given these differences, is it acceptable to use testing system is not available. For example, an audiologist may wish supra-aural headphones to make REAT measurements? to measure the attenuation of custom-molded earplugs dispensed to a musician client, or an occupational hearing conservationist may wish Methods to verify that an employee ’ s earplugs are providing adequate protection in the work environment. In such cases, an obvious alternative Participants is to conduct a REAT test with equipment that is readily available, Forty-two people (21 females, 21 males, mean age 22.3 years, such as an audiometer and a pair of standard supra-aural audiometric s.d. 4.7) participated in the study. This was a convenience sample Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12 headphones. recruited via fl yer and word of mouth. Several of the female par- However, the use of supra-aural headphones in REAT testing has ticipants were graduate students in audiology or speech-language- long been discouraged because of potential measurement artifacts pathology, while most of the male participants were college students (Berger, 1984, 1986). Berger (1984) cites four concerns: (1) the from non-audiology/non-speech-language-pathology backgrounds. headphone may rest on the earplug, either breaking its seal in the All participants had normal hearing (thresholds at 125, 250, 500, ear canal or pushing it deeper into the ear canal; (2) the pressure of 1000, 2000, 3000, 4000, 6000, and 8000 Hz, 25 dB HL in both the headphone on the pinna may distort the pinna and/or ear canal, ears, with no more than a 10-dB difference between ears at each altering the fi t of the earplug; (3) contact between the headphone frequency), normal tympanograms, clear ear canals, and no history and the earplug may cause structural vibration of the earplug; and or current evidence of outer-ear or middle-ear pathology. Experi- (4) acoustical excitation of the earplug under the very limited vol- ence using HPDs was neither an inclusion nor an exclusion criterion. ume between the headphone and the earplug “ is not representative Anecdotally, most of the subjects were inexperienced with respect of sound fi eld excitation” (p. 79). For these reasons, the attenuation to HPD use. measured under supra-aural phones may not be an accurate represen- tation of the attenuation the individual is actually receiving. Berger (1984) collected pilot data comparing measured attenua- Attenuation measurements tion under Telephonics TDH-50 supra-aural headphones to attenu- Earplug attenuation was measured with the Fitcheck2 Insert Hearing ation measured in the sound fi eld according to ANSI S3.19-1974, Protector Attenuation Measurement System (Michael and Associ- a precursor to ANSI S12.6-2008. Monaural REAT measurements ates, Inc.; hereafter called the “ Fitcheck system ” ). This headphone were made on four participants using both “ proper ” and “ improper ” REAT system has been found to be reliable for fi eld-testing earplug 732 J. B. Tufts et al.

attenuation (Franks et al, 2003). The Fitcheck system consists of proprietary software and hardware including a black box, a handheld response switch, and circumaural headphones. The black box was connected to a Dell Optiplex 960 desktop computer with a Sound- MAX HD Audio sound card. Stimuli for measuring thresholds were generated by the computer sound card and controlled by the Fitcheck system. The stimuli were 1/3-octave-band noises with center frequencies of 125, 250, 500, 1000, 2000, and 4000 Hz1 . The noises pulsed on and off at a rate of two pulses per second. Threshold testing proceeded according to a modifi ed Bekesy procedure, in which the subject controlled the level of the stimulus with the handheld response button. Stimulus level changed at a rate of 1.5 dB per second. Threshold was taken as the midpoint between the average of six peaks and fi ve valleys in the threshold tracing, with the fi rst peak and valley discarded. Threshold values were automatically saved in a Microsoft Access database. A “ run ” consisted of a single set of threshold measurements at the six test frequencies. An “ open run ” Figure 1. Photograph of the two headphone types and the three refers to a run completed while the subject was not wearing earplugs; earplug types used in the study. Headphones: Fitcheck circumaural an “ occluded run” refers to a run completed while the subject was headphone set (L) and Telephonics TDH-50P supra-aural headphones wearing earplugs. All testing was done binaurally. Since signals were housed in MX-41/AR cushions (R). Earplugs: Howard Leight presented to both ears, information as to which ear was controlling Fusion ® (L); E-A-R ™ Amigo ™ (middle); Westone ® #40 (R). the test under each headphone type was lost. It is possible that the same ear controlled the tests under both headphones; alternatively, opposite ears could have controlled each test. The decision to test subject. She described the process of fi tting the earplugs so that the binaurally, rather than monaurally, was made because of concerns participant experienced a proper fi tting (e.g. straightening the ear about cross-hearing in a monaural testing paradigm, and because canal by pulling up and back on the pinna prior to earplug inser- fi eld REAT testing is often done binaurally. It was felt that these tion; obtaining a pneumatic seal in the ear canal with the earplug, concerns outweighed the disadvantage of binaural testing. etc.). At this visit, each subject was assigned quasi-randomly to one Thresholds were measured under two types of headphones, Tele- of three groups, A, B, or C. Each group contained 14 participants phonics TDH-50P supra-aural headphones housed in MX-41/AR (seven females, seven males). cushions, and Fitcheck circumaural headphones. The Fitcheck cir- Attenuation data were collected on the second, third, and fourth cumaural headphones are supplied with the Fitcheck system, and visits. At each visit, identical procedures were followed, but a dif- For personal use only. have deep earcups designed for REAT testing of earplugs. Head- ferent earplug was tested. The order of earplug testing was assigned phone output levels were measured on an IEC-318 fl at-plate coupler to the three groups according to a modifi ed Latin square design. before, halfway through, and following data collection, and remained Earplug test order for Group A was foam-fl ange-custom; for Group stable. All testing took place in a single-walled sound-treated booth B it was fl ange-custom-foam; and for Group C it was custom- (IAC Model 403). foam-fl ange. Three types of earplugs were tested: the E-A-R ™ Amigo ™ , which At each of the three data-collection visits, the subject was allowed is a “ roll-down ” foam plug; the Howard Leight Fusion® , which is a to practice fi tting the earplugs to be tested that day. The subject was premolded quadruple-fl ange plug; and the Westone® #40, a silicone given any written or pictorial instructions on the earplug packag- custom-molded plug. The fl ange plug came in two sizes. Figure 1 ing. Assistance from the investigator was limited to answering ques- shows a photograph of the two headphone types and the three ear- tions from the participant and pointing out gross mis-fi ttings. This plug types used in the study. protocol, combined with the investigator-assisted fi tting on the fi rst Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12 day, was designed to emulate a situation in which a user of earplugs is neither completely naï ve, nor extensively trained, in fi tting Data-collection protocol earplugs. Each participant made four visits to the laboratory. At the fi rst visit, Following fi tting practice, the subject removed the earplugs, and the subject ’ s hearing and middle-ear function were evaluated. Next, entered the sound-treated booth. The circumaural headphones were he or she practiced the REAT test procedure. Five open runs were placed on the subject and an open run was administered. The head- completed, and the last three were checked for variability. If the range phones were removed, and the participant inserted the earplugs in of thresholds at each test frequency was 6 dB, then the participant the presence of a “ fi tting noise. ” This fi tting noise, pink noise at immediately qualifi ed. If not, additional runs were administered 70 dB SPL, was presented in the sound fi eld through three Electro- until the last three runs met the 6-dB requirement, or until a total voice Sx100 loudspeakers situated in three corners of the sound- of 10 practice runs was administered. All subjects qualifi ed within treated booth. The participant was encouraged to take the time 10 practice runs. Next, earmold impressions were taken of both ear necessary to get a good fi t with the earplugs. Once the participant canals by a graduate student in the university audiology program. indicated that he/she was satisfi ed with the fi t, the fi tting noise was These impressions were used to fabricate the custom-molded ear- turned off, and he/she was instructed not to touch or adjust the earplugs. The student investigator was well-versed in making earmold plugs in any way until told to do so. impressions for hearing aids, but had not been trained specifi cally The investigator replaced the circumaural headphones, adminis- for making impressions for custom-molded earplugs2 . Finally, the tered an occluded run, then reseated the headphones and administered student investigator fi t the correct size foam and fl ange plugs on the a second occluded run3 . Following this, the investigator removed Earplug attenuation under headphones 733

the circumaural headphones and replaced them with the supra-aural . headphones. Two occluded runs were administered, again with the headphones reseated between runs. After these two occluded runs, ttings the subject removed the earplugs. The investigator replaced the supra-aural headphones and administered an open run. A total of six runs was administered, three under the circumaural phones (one

open, two occluded) and three under the supra-aural phones (two 42) and two ﬁ 34.8 (6.6)30.5 (7.0) 40.0 (7.4) 33.9 (12.2) occluded, one open). At no time between the insertion and removal

of the earplugs was the participant allowed to touch or adjust the 7) 33.1 (6.0) 37.0 (7.9) eld corrections applied for earplugs in any way. This series of six runs was repeated for a second

fi tting of the earplugs. Thus, in all, twelve runs were administered at each visit, six for each fi tting of the earplugs. Testing with the circumaural phones always occurred before testing with the supra- aural phones. This was done so that any disruption of the fi t of the earplugs by the supra-aural phones would occur after measurement d across all subjects (n d across all subjects (n under the circumaural phones (the “ gold standard ” in this study) was completed. All study procedures were approved by the Institu- Supra-aural headphones tional Review Boards of the University of Connecticut and the Naval with sound-fi d threshold, Submarine Medical Research Laboratory. Data were pre-processed before analysis. For a given fi tting of an earplug under the supra-aural headphones, attenuation at each test frequency was calculated as the mean of the two consecutive occluded thresholds minus the open threshold measured immediately afterward. For the circumaural phones, attenuation was calculated in the following way: First, the difference between the mean of the two consecutive occluded thresholds and the open threshold measured immediately prior was calculated. This difference was multiplied by a small frequency-dependent constant, and the product was added to another small frequency-dependent constant. These constants were previously derived by the developers of the Fitcheck system. Their purpose is to correct the measured attenuation values slightly so that they match more closely the values that would be obtained in a dif- For personal use only. fuse sound fi eld according to ANSI 12.6-2008. When the Fitcheck system is used in the typical manner, these constants are applied automatically by the software. Occasionally, threshold values were missing in the database. This occurred if, for whatever reason, the minimum or maximum out-

put limits of the Fitcheck attenuators were reached during threshold measurement. In these cases, we calculated attenuation with a single occluded threshold instead of two, or, more rarely, by substituting the open threshold from the other ﬁ tting of the earplug in place of the missing value. In no case were we unable to calculate attenuation. PAR was calculated according to the method described by Michael Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12

(1999). A pink noise measuring 100 dB SPL in each one-third-octave in parentheses are shown headphone types (circumaural and supra-aural). Standard deviations ange, and custom-molded) under two test band was assumed. Each of these test bands was A-weighted and then summed to give an overall A-weighted exposure level. Next, the Circumaural headphones measured attenuation in dB in each of the six bands was subtracted from the A-weighted level in the corresponding band. The resulting differences were summed to give an overall A-weighted level under the HPD. Finally, the PAR was calculated as the difference between 125 250 500 1000 2000 4000 PAR 125 250 500 1000 2000 4000 the overall A-weighted unprotected exposure level and the overall A-weighted protected level.

Results Analyses of variance Table 1 gives the attenuation values and PAR values for each ear-

plug type under the two headphone types, averaged across partici- average Attenuation at six test frequencies (125, 250, 500, 1000, 2000, and 4000 Hz) the personal attenuation rating (PAR), . pants and across the two earplug ﬁ ttings. On average, participants obtained relatively good attenuation with their earplugs. As expected, Frequency (Hz)Earplug PAR FoamFlangeCustom 29.2 (7.6) 27.2 (7.9) 22.6 (9.7) 29.3 (9.7) 25.6 (8.3) 26.1 (12.4) 25.6 (10.0) 26.0 (8.5) 27.6 (13.7) 27.8 (10.2) 31.1 (13.6) 29.0 (8.4) 25.7 (9.6) 32.4 (5.7) 29.0 (9.6) 28.0 (8.3) 30.8 (8.2) 39.8 (6.0) 28.4 (6.3) 37.6 (8.4) 34.6 (11.3) 30.2 (7.7) 31.5 (6.2) 27.1 (7.3) 21.7 (9.8) 27.7 (8.8) 24.4 (8.2) 25.3 (10.1) 28.4 (9.3) 31.2 (7.3) 25.4 (6.5) 31.1 (8.7) 31.9 (6.6) 27.4 (8.7) 31.6 (6. 28.0 (7.6) Table 1 1 Table Within a given subject, attenuation at each frequency was calculated as the mean of two occluded thresholds minus the unocclude calculated as the mean of two was subject, attenuation at each frequency a given Within of three earplug types (foam, premolded quadruple-ﬂ the standard deviations were fairly large, owing to the nature of the the circumaural headphones per manufacturer. 734 J. B. Tufts et al.

study protocol and the inexperience of the population, as described Table 2. The degrees of freedom (df), F-values, and p-values for previously. Interestingly, under the circumaural headphones but not all main effects and interactions in the ANOVA of the reduced model the supra-aural phones, the across-subject standard deviations of the for attenuation (see Results). Signifi cant main effects and interactions custom earplugs were noticeably higher than those of the foam and are in bold. Corrected df values are shown if Mauchly ’ s test of fl ange earplugs for test frequencies of 125, 250, and 500 Hz. In sphericity was signifi cant. addition, the standard deviations of the custom plugs were mark- edly higher under the circumaural headphones than under the supra- Main effect/interaction df F p aural headphones for test frequencies up to and including 1000 Hz. Headphone 1, 40 2.09 .16 An examination of the custom earplug data frequency by frequency Sex 1, 40 12.06 .00 showed some data points ranging below 10 – 15 dB under the cir- Headphone sex 1, 40 2.04 .16 cumaural headphones, but no such data points under the supra-aural Earplug 2, 80 3.22 .05 headphones. The upper end of the range was similar across head- Earplug sex 2, 80 1.36 .26 phones. It seems likely that the custom earplugs naturally showed Frequency 2.77, 110.80 102.65 .00 greater variability in attenuation in the lower frequencies than either Frequency sex 5, 200 3.29 .00 the foam or fl ange earplugs, and that the supra-aural headphones Headphone earplug 2, 80 .98 .38 altered the fi t of the low-attenuating custom plugs in particular. This Headphone earplug sex 2, 80 1.79 .17 conjecture is examined in more detail in the Discussion. Headphone frequency 3.22, 128.75 3.86 .01 A repeated-measures analysis of variance (ANOVA) was con- Headphone frequency sex 5, 200 1.45 .21 Earplug frequency 5.06, 202.53 10.79 .00 ducted on the dependent measure of attenuation. The signifi cance Earplug frequency sex 10, 400 1.53 .13 level for this and subsequent analyses was 0.05. Within-subjects Headphone earplug frequency 7.54, 301.62 2.80 .01 variables included headphone type (two levels), earplug type (three Headphone earplug frequency sex 10, 400 1.64 .09 levels), fi tting (two levels), and frequency (six levels). Between- subjects variables were sex (two levels) and earplug testing order (three levels). The Huynh-Feldt epsilon was used to correct the prob- abilities for the main effect of frequency and its interactions, due to The attenuation curves of each earplug type under each headphone the violation of the sphericity assumption (Huynh & Feldt, 1976). were examined. The patterns were similar: generally speaking, The main effect of the fi tting variable and its interactions fl ange earplugs provided the least attenuation of all three earplug were examined. Only the three-way interaction of earplug fre- types; custom earplugs provided greater attenuation in the lower frequency fi tting (F(7.146, 257.246) 2.251; p .03) was signifi cant. quencies, and foam earplugs gave greater attenuation in the higher For each earplug frequency combination, the difference between frequencies. For all frequencies and earplug types, differences in the means for fi tting 1 and fi tting 2 was calculated. Of these 18 measured attenuation between the two headphones ranged from differences, two were approximately 3 dB, two were approximately For personal use only. 0– 3.6 dB, with most differences being less than 2 dB. In 10 out of 2 dB, and 14 were either 0 or 1 dB. No pattern was discernible in 18 cases, including the six cases in which the difference was greater these differences. Consequently, values for the two fi ttings were aver- than 2 dB, attenuation measured under the supra-aural headphones aged in the subsequent analysis. was greater than that measured under the circumaural headphones. The main effect of testing order and its interactions were also Of course, headphone comparisons for individual subjects showed examined. Only the two-way interaction of earplug testing order greater differences than those described here. Results for individuals was signifi cant (F(4, 72) 3.171; p .019). The set of three ear- are presented later in the paper. plug means was examined for each of the three testing orders (i.e. A similar analysis was conducted on the dependent measure of foam-fl ange-custom, fl ange-custom-foam, custom-foam-fl ange). No PAR. In the full model, the within-subjects variables were headphone learning effects were evident, nor was any other clear pattern evident. type and earplug type; the between-subjects variables were sex and Thus, testing order was removed in subsequent analysis. testing order. Once again, the main effect of the fi tting variable and A second ANOVA was conducted on attenuation (i.e. attenuation Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12 its interactions were examined. Only the three-way interaction of averaged across the two earplug fi ttings), using the reduced model. headphone earplug fi tting was signifi cant (F(2, 72) 4.219; For this analysis, the within-subjects variables were headphone type, earplug type, and frequency; the between-subjects variable was sex. Table 2 lists the degrees of freedom, F-values, and p-values for all main effects and interactions. Signifi cant main effects and interac- Table 3. The degrees of freedom (df), F-values, and p-values for tions are in bold. all main effects and interactions in the ANOVA of the reduced model The signifi cance of the main effect of frequency was expected, for PAR (see Results). Signifi cant main effects and interactions are since it is well-known that earplug attenuation generally increases in bold. with frequency. The attenuation curve of the fl ange plugs was somewhat fl atter compared to the curves of the foam and custom Main effect/interaction df F p plugs, and this is refl ected in the signifi cant interaction of earplug type frequency. The main effect of sex and its interaction with Headphone 1, 40 3.16 .08 frequency were signifi cant. For all six frequencies, women obtained Sex 1, 40 12.89 .00 greater attenuation than men, with differences of 6– 7 dB in the lower Headphone sex 1, 40 .24 .62 frequencies and 3 – 4 dB in the higher frequencies. Earplug 2, 80 3.91 .02 2, 80 .86 .43 The main effect of headphone was not signifi cant; however, the Earplug sex Headphone earplug 2, 80 2.56 .08 two-way interaction of headphone frequency was signifi cant, as Headphone earplug sex 2, 80 .89 .42 was the three-way interaction of headphone earplug frequency. Earplug attenuation under headphones 735

p .019). For each headphone earplug combination, the difference the degrees of freedom, F-values, and p-values for all main effects and between the means for fi tting 1 and fi tting 2 was calculated. For all interactions. Signifi cant main effects and interactions are in bold. but one of the differences (Fitcheck headphones foam earplugs), The main effect of sex was signifi cant. Collapsed across all other the mean of the fi rst fi tting was greater than that of the second fi t- variables, the average PARs for women and men were 31.9 and 26.3 dB, ting. However, all of the differences were less than 2 dB. Due to the respectively. The main effect of earplug type was also signifi cant. Listed small magnitude of the differences, values for the two fi ttings were in order of increasing magnitude, the average PARs were 27.2, 29.7, and averaged in subsequent analysis. 30.4 dB for fl ange, foam, and custom earplugs, respectively. Neither the The main effect of testing order and its interactions were also main effect of headphone type nor its interactions were signifi cant. examined. Only the two-way interaction of earplug testing order was signifi cant (F(4, 72) 3.143; p .019). The set of three earplug means was examined for each of the three testing orders (i.e. Scatterplots foam-fl ange-custom, fl ange-custom-foam, custom-foam-fl ange). No Data for individual subjects were examined with the help of scatter- learning effects were evident, nor was any other clear pattern evident. plots. Each scatterplot shows the attenuation measured under supra- Thus, testing order was removed in subsequent analysis. aural headphones as a function of the attenuation measured under A second ANOVA was conducted on PAR, using the reduced model. circumaural headphones for a given earplug type. Scatterplots for the For this analysis, the within-subjects variables were headphone type test frequencies of 500 Hz and 2000 Hz are shown in Figures 2, a– c and earplug type; the between-subjects variable was sex. Table 3 lists and 3, a– c, respectively. Scatterplots of the PAR are shown in Figure 4,

(a) Foam earplugs (b) Flange earplugs 50 50

40 40

30 30

20 20

10 10 For personal use only. 0 0 Attenuation (dB) - Supra-aural headphones Attenuation (dB) - Supra-aural headphones 0 1020304050 0 1020304050 Attenuation (dB) - Circumaural headphones Attenuation (dB) - Circumaural headphones

40 Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12

0 Attenuation (dB) - Supra-aural headphones

0 1020304050 Attenuation (dB) - Circumaural headphones

Figure 2. Scatterplots of the attenuation at 500 Hz obtained under supra-aural and circumaural headphones. Each point is the mean of the attenuation values for two ﬁ ttings of the earplug (see Table 1). Dashed lines on either side of the diagonal indicate the limits within which results for the two headphones agreed within 10 dB. Values for men (n 21) and women (n 21) are shown as closed and open circles, respectively. 736 J. B. Tufts et al.

(a)Foam earplugs (b) Flange earplugs 50 50

40 40

30 30

20 20

10 10

0 0 Attenuation (dB) - Supra-aural headphones Attenuation (dB) - Supra-aural headphones 0 1020304050 0 1020304050 Attenuation (dB) - Circumaural headphones Attenuation (dB) - Circumaural headphones

10 For personal use only.

0 Attenuation (dB) - Supra-aural headphones 0 1020304050 Attenuation (dB) - Circumaural headphones

Figure 3. Scatterplots of the attenuation at 2000 Hz obtained under supra-aural and circumaural headphones. Each point is the mean of the attenuation values for two ﬁ ttings of the earplug (see Table 1). Dashed lines on either side of the diagonal indicate the limits within which results for the two headphones agreed within 10 dB. Values for men (n 21) and women (n 21) are shown as closed and open circles, respectively. Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12

a – c. Each point on a scatterplot represents an individual subject. As across all test frequencies, only three points fell in this region (two the scatterplots show, most subjects obtained adequate to good atten- at 125 Hz and one at 4000 Hz). For the fl ange and custom earplugs, uation, regardless of earplug type. A few subjects obtained relatively 23 and 33 points fell in this region, respectively. Considering the low PARs ( 15 dB). From these data, we cannot tell whether those fl ange and custom earplugs together, the test frequencies of 250 subjects required more extensive training, or better-fi tting earmolds Hz and 500 Hz each fl agged the most points; we chose to show (in the case of custom earplugs), or whether their ear canals simply the scatterplots for 500 Hz. The test frequency of 2000 Hz fl agged were not suited for those HPDs. the fewest points. Interested readers who wish to view all of the The diagonal in each scatterplot indicates exact correspondence scatterplots may contact the fi rst author. between the two headphones. Dashed lines on either side of the Several general observations follow. First, the vast majority of diagonal indicate the limits within which results for the two head- points fell within the dashed lines, indicating relatively good cor- phones agreed within 10 dB. Following Berger (1984), these limits respondence of measurements between the two headphone types. were chosen taking into consideration variability in measurements Second, the points for the foam earplugs were more tightly clus- of hearing threshold level. tered around the diagonal than the points for either the fl ange or If avoiding underprotection is important, then those points falling custom earplugs. Third, more points fell above the upper dashed above the upper dashed line are of greatest concern. Those points line than below the lower dashed line. This is especially apparent represent individuals for whom testing with supra-aural headphones for the custom earplugs at 500 Hz. Fourth, women tended to obtain produced results likely to be spuriously high. For the foam earplugs, higher attenuation than men across earplug types and headphones. Earplug attenuation under headphones 737

Foam earplugs Flange earplugs (a) (b) 50 50

40 40

30 30

20 20

10 10 PAR (dB) - Supra-aural heaphones PAR (dB) - Supra-aural headphones 0 0

01020304050 0 1020304050 PAR (dB) - Circumaural headphones PAR (dB) - Circumaural headphones

Custom earplugs (c) 50

20 For personal use only. 10

PAR (dB) - Supra-aural headphones 0 0 1020304050 PAR (dB) - Circumaural headphones

Figure 4. Scatterplots of the personal attenuation ratings (PARs) obtained under supra-aural and circumaural headphones. Each PAR is the mean of the PARs for two ﬁ ttings of the earplug (see Table 1). Dashed lines on either side of the diagonal indicate the limits within which results for the two headphones agreed within 10 dB. Values for men (n 21) and women (n 21) are shown as closed and open circles,

Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12 respectively.

This result may have occurred because many of the women were particular fi t-testing method that is employed must provide accurate, graduate students in audiology or speech-language-pathology, and usable results. thus may have had greater motivation or experience. In this study, we compared measurements of earplug attenuation obtained under supra-aural audiometric headphones to measurements obtained under circumaural headphones designed expressly for such Discussion testing (the “ gold standard ” in our study). When results were aver- Any aspect of the earplug fi tting process, either singly or in com- aged across participants, agreement between the two headphones bination with other aspects, can result in a poor-fi tting earplug: an was good. Of the 18 headphone comparisons (3 earplug types 6 inexperienced fi tter, poor manufacturing quality control, incorrect test frequencies), the averages for 12 of the comparisons were less fi tting technique, or ear-canal geometry that is incompatible with than 2 dB apart. the style of earplug. Individual fi t-testing can show when, despite Comparisons between the two headphones at the individual level the efforts of an experienced fi tter and a motivated wearer, a par- were more illuminating. For each earplug type, there were 252 such ticular earplug does not fi t well or does not provide the appropriate comparisons (42 participants 6 test frequencies). In most cases, amount of attenuation. Fit-testing data can also reveal any systematic results for the two headphones agreed within 10 dB. Spuriously high weaknesses in a fi tting program, which might be attributable to fi tter attenuation was measured under the supra-aural headphones in 3, inexperience or to manufacturing quality control. In any case, the 23, and 33 cases for the foam, fl ange, and custom plugs, respec- issue can (hopefully) be remedied if it is discovered. Clearly, the tively. When attenuation at the six test frequencies was combined 738 J. B. Tufts et al.

into a single PAR, the number of cases of spuriously high attenuation more conservative recommendations regarding limitations of dropped proportionately to 0, 4, and 5, respectively. These numbers supra-aural headphone use. are negligible in the case of the foam earplugs, but represent about 10% of the individuals tested with fl ange and custom plugs. Acknowledgements Spuriously high attenuation may have been caused by the earplugs getting pushed further into the ear canals by the supra-aural head- The authors thank two anonymous reviewers for their many help- phones. Anecdotally, participants noted few discernible changes in ful comments and suggestions. We acknowledge Angeli Mohanani, the fi t of the foam earplugs under the supra-aural headphones. In Sneha Parikh, and Kara Swan for their assistance with data collection. contrast, several participants remarked about the discomfort they Results from this study were presented at the 2012 conference of the experienced while being tested with the fl ange and custom earplugs National Hearing Conservation Association in New Orleans, USA. under the supra-aural headphones; several felt that the headphones were pushing the earplugs in more deeply. Declaration of interest: This work was supported by the Offi ce of Large discrepancies between the headphones were noted particu- Naval Research through the Naval Submarine Medical Research larly with the custom earplugs. We retested two individuals (one man, Laboratory work unit 50814. The study protocol was approved by one woman) who had very large discrepancies in PARs with their the Naval Submarine Medical Research Laboratory Institutional custom earplugs (see Figure 4, c). Even with additional instruction, Review Board in compliance with all applicable Federal regulations results were essentially unchanged. The female participant was inter- governing the protection of human subjects. The views expressed in viewed after the retest. The canal portions of her custom earplugs this article are those of the authors and do not necessarily refl ect the were atypically short. With the earplugs worn correctly, she was still offi cial policy or position of the Department of the Navy, Department able to understand quiet speech fairly well; when she pressed on the of Defense, nor the U.S. Government. concha portion of each earplug, she was no longer able to under- The fi rst author is considered a part-time employee of the U.S. stand the speech at all. In her case, the result of a simple “ press test ” Government. This work was prepared as part of her offi cial duties. in the presence of speech signaled that REAT testing with supra- Title 17 U.S.C. § 105 provides that ‘ Copyright protection under this aural headphones was contraindicated (and that the poorly-fi tting title is not available for any work of the United States Government. ’ earmolds should be remade). Another situation might not be so obvi- Title 17 U.S.C. § 101 defi nes a U.S. Government work as a work ous. We are unaware of simple tests that can accurately and reliably prepared by an employee of the U.S. Government as part of that per- identify those fl ange and custom earplug fi ttings for which testing son ’ s offi cial duties. The authors report no declarations of interest. under supra-aural headphones would be contraindicated. Therefore, at this time we do not recommend the use of supra-aural headphones when testing fl ange and custom earplugs. References American National Standards Institute. 2008. American national standards methods for measuring the real-ear attenuation of hearing protectors.

For personal use only. Conclusions S12.6-2008. New York: ANSI. We conclude that standard supra-aural audiometric headphones are suit- Berger E.H. 1984. Assessment of the performance of hearing protectors for hear- able for measuring the attenuation provided by foam earplugs. However, ing conservation purposes. Noise Vibr Control Worldwide, 15, 75– 81. supra-aural headphones should not be used to measure the attenuation Berger E.H. 1986. Review and tutorial: Methods of measuring the attenuation of fl ange or custom-molded earplugs. The potential exists for substantial of hearing protection devices. J Acoust Soc Am , 79, 1655 – 1687. over-estimation of attenuation, especially of custom plugs. Berger E.H. 1989. Exploring procedures for fi eld testing the fi t of earplugs. Proceedings, 1989 Industrial Hearing Conservation Conference, Off. Eng. Serv., University of Kentucky, Lexington, USA, 7 – 10. Notes Berger E.H. 2006. Introducing F-MIRE testing: Background and concepts. E • A • R 1. Supra-aural headphones are used clinically, most often to obtain Technical Report 06-29/HP, Indianapolis, USA: Aearo Technologies. pure-tone thresholds. The pure tones are presented via audiom- Berger E.H. 2000. Hearing protection devices. In: E.H. Berger, L.H. Royster,

Int J Audiol Downloaded from informahealthcare.com by University of Connecticut on 10/23/12 eter. If supra-aural headphones are used in a REAT paradigm, J.D. Royster, D.P. Driscoll & M. Layne (eds.). The Noise Manual: Fifth Edition . Fairfax, USA: American Industrial Hygiene Association, ch. 10. most likely they will be connected to an audiometer, not a REAT Berger E.H., Voix J. & Hager L.D. 2008. Methods of fi t testing hearing pro- system. In that case, the operator may have a choice to present tectors, with representative fi eld test data. In: 9th International Congress either pure-tone or narrowband-noise signals. The choice is moot on Noise as a Public Health Problem (ICBEN): ICBEN 2008 Proceed- (see Berger, 1989). In this study, we chose to use one-third- ings . Foxwoods, USA. octave-band noise stimuli for both headphone types (rather than Franks J.R., Murphy W.J., Harris D.A., Johnson J.L. & Shaw P.B. 2003. Alter- one-third-octave-band noise for the circumaural phones and pure native fi eld methods for measuring hearing protector performance. AIHA tones for the supra-aural phones) in order to increase internal Journal , 64, 501 – 509. validity. Hager L.D. 2011. Fit-testing hearing protectors: An idea whose time has 2. Such training is currently required in at least one hearing come. Noise and Health , 13, 147 – 151. conservation program of the U.S. Navy. Huynh H. & Feldt L.S. 1976. Estimation of the box corrections for degrees of freedom from sample data in randomized block and split-plot designs. 3. The “ double-measurement ” of occluded thresholds is not part of J Educ Stat , 1, 69 – 82. the standardized laboratory REAT protocol, nor would it be done Michael K. 1999. Measurement of insert-type hearing protector attenuation in a fi eld setting. Here, we double-measured the thresholds in on the end-user: A practical alternative to relying on the NRR. Spectrum , order to improve the precision of the REAT measurement, which 16, 13 – 17. is subject to considerable variability. Thus, differences between Schulz T.Y. 2011. Individual fi t-testing of earplugs: A review of uses. Noise conditions were more likely to become signifi cant, leading to and Health , 13, 152 – 162.