Meeting Notes

“Informal” HPD meeting held at EPA offices

January 23-24, 2007

Tuesday, January 23

9:30 – 10:15 am

Introduction by Ken Feith – went over presentation that was to be
given at an ISEA meeting in November 2006 

The EPA is required to review periodically its regulations according to
Public Law 95-574 enacted 27 Oct 1972.  This authority cannot be
delegated to another agency.  Under EPA’s responsibility, a recent
study was funded to evaluate noise levels for railroads.  EPA has funded
this sort of effort out of discretionary funds from the general fund
since there is no specific budget for the Office of Noise Abatement and
Control.  Since commencing the work on the EPA regulation, approximately
$500000 has been awarded in interagency agreements to fund NIOSH to
assist in the efforts.  The EPA’s mandate is to provide information to
the public about noise.

The criteria for classifying a device as a hearing protector must be
resolved.  The label on the protector and the description of the meaning
of the label are of primary interest.  Much of the rest of the labeling
information must stay the same.  That is the number rating, the
interpretation of the rating within the NRR label must be changed.  The
manufacturer information and the product identification cannot be
changed.  The goal of the revision is to make the information clear and
simple such that the public can understand its purpose and intent.

EPA has contract with ECR to identify small and large companies that
manufacturer and test hearing protectors.  ECR interviewed several
companies and provided reports to the EPA detailing their findings.  The
final contract report will be submitted as soon as the budget is
resolved.

10:20 – 11:20 am

Bill Murphy – went through inter-lab study slide presentation that was
given at the December 2006 ASA meeting in Hawaii.  There was an extended
discussion on reliability and reproducibility (between labs; between and
within subjects).  Also discussed the variability between Method A and B
results.

The essential findings from the interlab study are as follows: 

Method B produced lower interlab variability.

Method A and B produced comparable results for the earmuffs.

Method A yielded higher rating values than Method B for the earplugs.

Confidence intervals were computed for the various laboratories and
products.

Confidence intervals provide information about the historical
measurements but are not predictive of future measurements.

The confidence intervals were computed using a numerical technique,
bootstrap procedure, that resamples the data to simulate the potential
variability of the population from which the subjects were selected.

Earmuffs, canal caps and custom earplugs did not benefit significantly
from providing training to the subjects.  

Insert earplugs exhibited significant increases in the rating as a
result of training.

Selling insert earplugs without training will result in workers not
receiving the performance that should be attainable.

Selling other products (custom ear plugs) usually include training
because of the expense of the product.

11:30am – 12:45pm

In-depth Method A vs. Method B discussion

Lee Hager – What does EPA want the label to be?  Method A is a fairly
reliable measure, but “wrong.”  Lee suggested that both A and B
numbers appear on the label.

Jeff Birkner – Will be an undue burden on manufacturers if have both A
and B ratings – it should be up to the employer to teach workers to
use HPDs properly.

Rich McKinley – Difference is rating the HPD itself vs. rating the
hearing conservation program at the plant.

Dan Gauger – How well a device behaves depends on how it is worn (Dan
likes Method B).  If HPDs are rated by method B, then manufacturers will
develop HPDs that can be used easily.  Method B is a better correlation
to the real world.

Elliott Berger – Method B still over-estimates field data; this is a
statistical value – some people do well with the device and others do
not.  There is a “consumer” issue (e.g., the person buying an HPD
without getting training).

Brian Meyers – The consumer doesn’t know his noise exposure level. 
The rating should give an appropriate relative rank-order.

Elliott – Referring to his lab vs. field performance bar chart: Method
A doesn’t rank-order or give the absolute performance of the HPD.

Brian – De-rating an HPD is the absolute worst thing to do (most
people appeared to agree).

Rich – The lab-to-lab variability is too high (Elliott mentioned that
his ANSI Working Group is looking at this).  Need to have a training
session for test lab personnel.

Janice – (she asked a question re: product labeling for other products
rated by the EPA)  Ken gave an answer using automobile mileage rating. 
Steven Silverman (EPA General Council) said that there is nothing to
prevent it.  Lee Hager agreed, but added that both the current NRR
rating and fuel efficiency ratings are not realistic.

Rich – Method A is not the optimal capability of the HPD – the
experimenter doesn’t fit the HPD as in ANSI S3.19.

Ken – 30 years ago the question was asked why an acoustical test
fixture (dummy head) wasn’t used instead of a real person.  Henning
von Gierke said that it wouldn’t mean anything – a person is always
going to be wearing the HPD.

Ken – We are still looking for the “bottom line.”

Ted Madison – It doesn’t matter; current 3M products have many
ratings (e.g., a range of 15-25 dB) – the text on the package
describes what an experienced vs. an inexperienced user might achieve. 
3M’s point is that performance varies; most people only need a few dB
of protection.

Theresa Schulz – What does the person ask when he/she goes to select
an HPD?  Also, there is the issue of enforcement to consider.

Ken – What would happen if the EPA rescinded the regulation?  (Lee
thought it would be a good idea; Dan shook his head “no”)

Neil Davis – Method B could be used for “consumers” – the actual
number rating could be de-emphasized.  The “industrial” population
would require something different.

Ted – The test method is less critical – the message from 3M is that
“performance varies.”  (He will provide EPA with examples of wording
that 3M is putting on their labels to downplay the NRR rating.)

Brad Witt – Agreed with Ted, but said that a number is needed, and
wants the number to mean something.  Brad doesn’t like Method B – it
would be like saying “We (the manufacturer) think you won’t use our
HPD correctly, but that is okay.”

Brad – (Referring to the fact that inexperienced wearers are needed
for Method B testing) Can the regulation be based on an “uniformed
user?”  This type of person doesn’t exist in the real world.  (Dan
responded to this by referring to Figure 4 in his No Ear Left Behind
paper – he again referred to his contention that Method B would
encourage manufacturers to make easier-to-use HPDs.)

Brian – Agreed with Brad’s point about “uninformed users” not
existing in a compliant hearing conservation program.

Ken – Is there a need for a numerical rating?  Need to have a
“fixed” reference – assume a reasonably intelligent user, and the
HPD is used according to the manufacturer’s instructions.

Lee – A “pass/fail” system is preferable.  (Elliott disagreed.)

Elliott – Method B is a “model” – a lab method that represents
the field data – each individual subject is not meant to be matched.

Rich – Using field studies as the primary reference is not a good idea
– there are problems with these studies.

Elliott – (Reviewed his Excel graphs of 1981, 1996, the 2006 method A,
and the 2006 method B inter-lab studies.)

Ken – Recent 6-lab inter-lab study showed variability between labs was
significant.  (He asked for thoughts on using a prescribed script that
each experimenter must use for fitting the HPD.)

1:50 pm – 2:00 pm

Bill Murphy – showed his next-to-last slide from his 2006 ASA
presentation (Questions for Discussion).

When we are looking at a range of test lab data, the question of
combining data across labs must be considered.  If we examine the Method
A data from the interlab study, the test methodology is precisely the
same.  Subjects exhibited lower variability and since the methods were
identical, the results can be combined.  However, the Method A data will
not lend themselves to being pooled.  The EAR Classic earplug tested at
the various labs yielded considerable variability as a consequence of
the experimenter training used in the study.  The AFRL laboratory made
marks on the protector and required subjects insert the protector to the
point that the mark was obscured.  If the results can be combined, then
this lends credibility to the Method B system as being more universally
applicable.

Dan – Data can be pooled with Method A or B if the product is exactly
the same.

Rich – Highest attenuations are reflective of what the device can do.

2:00 pm – 2:05 pm

Ken – The EPA must describe the test procedures (e.g., technical
requirements of the sound field, microphones, etc.).  The rule will be
more flexible this time – probably will not be revised again for a
long time – should be able to accommodate updated regulations.

Neil Davis – OSHA doesn’t have the same flexibility.

Janice – The Administrative Procedures Act applies.

Ken – EPA can adopt an ANSI standard by reference – can amend
sections as necessary.

2:05 pm 

Discussion began regarding the 2nd bullet item on the agenda (Test
procedures and protocols).

Kevin – Human subjects testing is the most common type of attenuation
testing done world-wide.  Test frequencies, signal generation, rise/fall
times, etc. have been pretty well standardized and accepted.

*Summary: Mostly a Method A vs. Method B question for passive muffs,
plugs, and semi-inserts (ANSI S12.6 procedure is adequate).

Active Noise Reduction/Cancellation Hearing Protectors

Ken – (Mentioned 5 Jan 07 memo from FAA re: not being able to hear
warning signals in the cockpit with ANR/noise cancellation headsets.)

Dan – Bose’s response: The cockpit will sound different with an ANR
headset – the headset itself doesn’t cause a problem – the
signal-to-noise ratio is not affected – there is no technical basis
for this concern.

Dan – Could use MIRE to test ANR performance – combine with REAT
(for passive attenuation) – works for circum-aural and supra-aural
muffs only.

Rich – NATO round-robin testing was done that used that used REAT for
passive attenuation and MIRE for the ANR component.

Rich – Will have to use a test head to measure earplugs.

Rich – An ANSI standard does not exist for the combined REAT + MIRE
procedure (ANSI S12.42 says not to combine REAT and MIRE).  

Elliott – A footnote in the draft ANSI S12.68 standard on rating
systems discusses how to put REAT and MIRE together.

Kevin – Costs will not be significantly greater for the test lab to
perform both REAT and MIRE.

Level Dependent Hearing Protectors

Rich – The Air Force lab uses an ISL test head for level-dependent
testing.

Bill – ISO method could be used for level-dependent testing – the
label would give the maximum level of noise that the HPD could be used
in.

Rich – Could have a separate number on the label for impulse noise
reduction.  Some HPDs have both ANR and level-dependent capability.

Elliott – There is an ISO standard (EN 352-4) that is used for steady
(not impulse) noises – indicates the noise level that would keep the
under-the-muff level to 85 dB.

Ken – (Summary) Are there existing test procedures that could be used?

Kevin – He could do a test with a shock tube and a test head, but
would it be representative of actual use, or would it be useful to the
manufacturers?

Bill – (Discussed indoor and outdoor measurements he had done on a
Billsom Impact 707).

Ken – (Summary) Can run passive (REAT) tests, but attenuation numbers
may be small at low sound levels.  There is a test method/protocol for
impulse noise testing, although it is not in any ANSI standard (tests
have been done by NIOSH, WPAFB, and USAARL).

A range or matrix of noises need to be created to evaluate the
performance of level dependent hearing protectors.  The noises should
include signals with a 1-2 millisecond A-duration.  Signals with short
durations will allow the evaluation of the compression and
level-limiting circuits of the device.  Also, the device needs to be
evaluated for a range of impulse levels.  It would be useful to collect
signals at 150 to 170 dB peak using a shock tube to demonstrate the peak
reduction of the devices.

Action

Draft test methods will be circulated to Dan Gauger and Rich McKinley to
 propose tests of hearing protectors with impulsive noises.

Communication earmuffs

Rich – For communication headsets (i.e., earmuffs that incorporate
other sound-producing devices) the duration of the exposure enters into
the problem.  

If the device/regulation limits the output of the device, then we must
worry about the duration of the exposure while wearing the protectors.

Brian – Some headsets with an external input do not regulate/limit the
volume level of the external device.

Aircraft cockpit levels are flight-line levels must be considered. 
Perhaps a warning label for the device in high noise levels should be
included.

Maximum output that can be produced by the device or that could be
produced as a result of the introduction of a standard signal should be
used to rate the communication output of the protector.  This suggests
that two tests will be needed:  One to evaluate the input output
function of the protector in response to external signals and another
test to evaluate the output of the device due to a signal of standard
levels transmitted electronically to the device (e.g. the input from the
signal.

Wednesday, January 24

9:00 am

Ken – Review of yesterday’s discussions – 2 questions arose: How
can lab-to-lab variability be minimized?  How can poor subject
performance be managed? (Elliott explained that this is what happened at
his lab during the inter-lab study.)  Solutions should be practical, not
necessarily mathematical.  These questions are important since the EPA
is likely to propose periodic testing, and possibly 3rd party testing
(for manufacturers that do their own in-house testing).

Rich – Poor subject performance can be handled with a standing subject
panel.

Elliott – (Disagreed with Rich) If the subject panel gets to be too
experienced, the test results start looking like Method A results.

Ken – When looking for repeatable measurements on what an HPD can
provide, then Method B or field testing may not be appropriate.

Elliott – In his experience, people attending his seminars aren’t
necessarily informed about how the current NRR doesn’t represent
actual use (and these might be plant safety people that you would think
should know better, not just uninformed users).

Ken – Would a 2-number system help?  EPA doesn’t want OSHA to
de-rate whatever is on the new label.  What about the “learning
effect?” – aren’t all subjects being trained with each subsequent
test?

Elliott & Ted – (Both don’t want to do both method A and method B
testing.)

Neil – Doesn’t like Method A – OSHA probably would want to de-rate
it, if a single number based on Method A was used.  (He didn’t
understand that you could have a 2-number rating system with either
method A or method B.)

Rich – Using field studies as the “gold standard” is not a good
idea.  (He gave the example of noise floor issues – the noise floor is
difficult to meet in the lab, let alone what happens in field studies.)

Dan – The correlation is 0.9 for method B to field data.

Ted – A 2-number system is better for 3M – it would make method A or
B less important.  Should make the new rating number usable with
A-weighted sound level measurements.

10:00 am

Elliott – Made a comparison chart of Method A vs. Method B:

	Method A	Method B

Within lab variability	X

	Between lab variability

X

Cost	X

	Speed	X

	Realism

X

Rank order

X

Explainable	X

	De-rating

X

Dual number labeling	X	X

Device’s inherent performance	X

	International applicability	X	X



Ken – EPA is encouraged to adopt international standards if it makes
sense.  Reciprocity is a problem due to inherent mistrust.  Problem with
companies exporting HPDs to the US that have not actually been tested
(but have an “NRR” on the label).

10:50 am

Elliott – Discussed his previous work on looking at rating schemes
(see the Berger and Gauger summary report dated January 22, 2007).  Page
4 has recommendations for NRSA – the picture of the label on page 6 is
slightly different than what was in his original 75-page paper.  ANSI
S12.68 will let you use either method A or method B.  The spreadsheet on
page 11 will accompany the new S12.68 standard.

Dan – Showed his suggestion for graphic display of data – he
suggested that the bar chart should replace the 2 numbers on the label. 
He also mentioned that his bar graph approach would fit into a system
for rating ANR devices.

Ken – How would this fit with rating a level-dependent HPD?

Bill – It could be used to represent the passive component.

Catrice – She was concerned that an uninformed user would need more
information that what is on the bar chart – use a “noise
thermometer” to convey a variety of activities that are considered to
be hazardous noise exposures.  These noises depicted on the bar chart or
thermometer could aid the user in understanding the level of protection
needed. 

Dan – Showed his suggestion for bar chart labels that are compatible
with ANR devices (see his handout with PowerPoint slides).  He stated
that it is important to distinguish between different usage categories
(e.g., industrial vs. aviation noise).  Rich agreed for a
military-specific environment.

12:30 pm

Ken – Do we have to differentiate between potential uses of HPDs? 
Should there be different labels for “consumer” and “industrial”
buyers?

1:40 pm

Ken – What are the essential elements for primary and secondary
labels?  What should be on “bulk” packaging vs. an individual
package?  The concept of NRR will be preserved – probably won’t go
to a NRS (noise reduction statistic) since it is a different concept –
NRR might stand for Noise Reduction Range in the new regulation.

Rich – Regarding “industrial” vs. “home” use – some noise
levels at home might actually be higher than occupational noises –
maybe this separation should not be based on the assumption that lower
noise levels are experienced at “home.”

Brian – Some “professionals” get their supplies at the Contractor
Desk in Home Depot – these retail stores are not just for
“consumers.”

Ken – Manufacturers of small batches/units may be exempt from certain
parts of the regulation (e.g., small automobile manufacturers have set
this precedent).

Rich – How do we handle labeling of double hearing protection?

Brian – Aearo used to package double protection (plugs + muffs), but
discontinued due to lack of sales.

Elliott – Might want to consider scaling back on what is on the
secondary label, since (even on an earplug dispenser box) there might
not be enough room to fit all of the information (e.g., octave-band
data, etc.).  Maybe just give a website or a phone number to call for
the additional information.

Ted – Maybe should require an “intended use” statement on the
package.

Brad – Didn’t agree with Ted – the user might think that the
protector is not good for other noise situations (other than what the
“intended use” states).

Ken – (Showed a newspaper with a Bose headset advertisement.)

Dan – For those types of devices, he doesn’t want any label – he
would put “not for use as a hearing protector” on the package.

Elliott – Didn’t agree with Dan’s wording – it shouldn’t be
exempted from labeling, considering the definition of hearing protector.

Ken – Mentioned the development of the NRR Calculator by NIOSH – all
data would be maintained by EPA (legal aspects of having data
transferred to the EPA need to be worked out).

Elliott – Regarding “consumer” ANR devices – maybe have a
footnote or an icon on the label instead of a number rating – still
would have a label, but it would just direct the user to another source
of information.

Rich – The same could be done for impulse noise ratings.

Dan – Could have different icons to indicate that a particular HPD is
good for certain types of noises.

Ken – Agreed – would have to come up with a catalog of symbols.

Elliott – There are 3 main categories: 1. Industrial/recreational
(continuous) noises.  2. Impulse noise.  3. Low-frequency dominated
noise.  Could use icons in place of the rating bar chart for these
special cases.

Ken – (Summary) A minimum of 2-number rating system is preferred.  Use
a bar chart to represent data.  Icons indicating where a particular
product is more suitable to be used.

Elliott – Don’t make it mandatory for instructions and additional
information to be included in the box of HPDs – make it like an MSDS,
where you can get the additional information upon request from the
manufacturer.

Ken – What has to be included in the package?  NRR?

Brad – Need to have the minimum usage instructions.

Brian – For private labeling of his products, he may not want to list
Aearo as the manufacturer.

Ken – In those cases, the private labeler would have to maintain
adequate records (i.e., original test data, etc.).

Brian – Also doesn’t want specific colors to be mandatory (too
expensive).

3:30 pm

Ken – Asked the group if the NRR Calculator would be useful to send
records to the EPA.  (Seemed to have agreement that it was a good idea.)

Dan – Wants NRR Calculator to collect MIRE data for each ear
separately.

Rich – Made additional suggestions regarding impulse noise.

Elliott – Skeptical about reliability/repeatability of impulse noise
measurements across different labs, since there is no ANSI standard for
this.  He noted that the new EPA regulation will have to be very
detailed.

Ken – The current motorcycle noise standard describes test conditions,
etc., in detail.

Brad – Regarding additional wording of how to apply the NRR – where
will it be required?

Ken – It depends on the product – it may not be required – the
1/3-octave-band data may be useful for an Industrial Hygienist, but not
for the average HPD user.

4:00 pm

Ken – Any last thoughts?

Brian – Aearo would have 34-52 weeks of testing to do, depending on
which method (A or B) is selected.  They have 67 devices, and 203
packages would need to be changed.  He estimated that this would take a
total of 18 months to complete.

Ken – Bill Batye should have asked for this information (Brian
indicated that Bill Batye had asked for it and that Aearo provided it to
him).  There will be at least a 1-year transition period – maybe 2-3
years.

Neil – Will there be a multi-language requirement?

Ken – None of the current regulations are multi-lingual.

Elliott – We never got around to talking about confidence intervals…

Bill Murphy – Went over his slides relating to the Bootstrap method
calculation (to determine the standard error).  The subject of
lab-to-lab variability came up again, and was discussed some more.

Ken – The question is: When will a manufacturer be required to
re-label the HPD?

Dan – Showed his graphs of scatter plots for the recent inter-lab
study.

Ken – If we have 3rd party testing, where would we get a larger error?

4:30 pm to ?

Additional discussion (?)

ACTION:  If we know of small operators in the hearing protection device
(HPD) industry, send information to EPA.

ACTION:  Bill Murphy will draft test methods for hearing protectors with
impulsive noises and will be circulate to Ken Feith, Dan Gauger and Rich
McKinley. 

ACTION:   Ted Madison will send examples of how 3M tries to convey
benefits without NRR (downplaying NRR).

Question:  Without NRR, how would manufacturers sell their products?  

Answer:  Emphasizing other features and showing how the HPD can solve
various issues.  There was general consensus that manufacturers would
prefer less emphasis on NRR even if it is a “noise reduction range”.

Question: How can information be conveyed (benefits) on the outside
package label and allow comparisons?

