Clarification of the AIBS Peer Review of AHAAH

Peer Review on the Human Research and Engineering Directorate (HRED) Method for Assessing the Risk of Auditory Injury for Hearing-Protected Soldiers Exposed to Impulse Noise

April 26, 2001

US Army Medical Research And Materiel Command
ATTN: MCMR-PLC / LTC Karl Friedl
Building 722
Frederick, MD 21702-5012

Dear LTC Friedl:

This letter and attachment are in response to your request for clarification on the report, The American Institute of Biological Sciences Scientific Peer Advisory and Review Services Peer Review on the Human Research and Engineering Directorate (HRED) Method for Assessing the Risk of Auditory Injury for Hearing-Protected Soldiers Exposed to Impulse Noise. The four reviewers who were present at the meeting in Aberdeen on January 30 and 31 were faxed a copy of your 5 April 2001 letter (attached). Three of the four reviewers responded to issues brought out in the letter. Their comments appear below.

I enjoyed working with you and your staff on this Program Review. I hope you find this attempt to clarify the Panel’s findings helpful.

Please let me know if you require further information.


Tina Rosenthal, M.H.S.
Project Coordinator
Scientific Peer Advisory and Review Services

Cc: W. L. Daniels
S. Glisson

Reviewers’ Responses

Reviewer 1 Response

The review board appreciates the gravity of the decision to be made. If the proposed standard provides the desired level of protection, the existing MIL-STD is not only overly restrictive, but is guaranteed in practice to lead to battlefield causalities as a result of inadequate training. Given this consideration, in the opinion of this reviewer, it is in the US Army’s vital interests to adopt an improved standard based on a physical model-based approach. Furthermore, this reviewer believes that the HRED method is substantially ready to be used as the basis for a new standard. Finally, this reviewer believes there is absolutely no question that the proposed model-based approach is vastly superior to that used in the existing ad-hoc standard.

The tough issues that this reviewer has not seen addressed by the Military Operational Medical Research Program (MOMRP) are:

  1. What is the best path for the production of a new standard based around the current HRED model, especially if deficiencies have been noted in the current model? Is it really in the Army’s best interest to reject a model-based approach if deficiencies existed but were readily fixable? Science (physical or military) rarely works well in this black-and-white manner. Science in any form is “self-correcting” and allows for the possibility of errors, but provides feedback mechanisms for correcting any deficiencies.
  2. The current HRED method is in reality a set of partially documented algorithms that have been (principally) implemented in the computer program, MATLAB. In practice, any standard must be more general than any particular implementation of a set of algorithms. A noise-exposure standard should include a complete documentation of algorithms used to compute the acceptable exposure levels in such a manner than any individual (or company, etc.) could implement their own version of the underlying software. Finally, the standard should include a set of benchmark results allowing for the validation of any particular implementation of the HRED program.

Finally, a standard should (of course) include a description as to the correct set of procedures for establishing exposure levels for any given training exercise. Do these exist? If not, they need to be developed and carefully reviewed, again by an independent body of scientists.

As to the particular questions raised by LTC Friedl:

  1. This reviewer felt that the minimum-phase method used by the HRED method is adequate for assessing the effect of particular hearing aid protectors. Free field trials would have been desirable, however.
  2. Dr. Price in his notes on “Using the HRED method” states that “500 ADUs is just safe, meaning that there may be temporary changes in hearing sensitivity of up to 25 dB, but recovery should occur and no permanent hearing loss is expected from the exposure.” It is difficult for this Panel to respond to a specific number without the existence of a detailed proposed standard, especially given the contradictory language that appears in the various documentation provided to us.
  3. No outer/middle-ear models exist which give a correct description above 5 kHz. The Price/Kalb model (based in turn on the Shaw-Stinson 2-piston model) underestimates the amount of energy transmitted through the middle ear above 5 kHz. Note, however, that the ADU is computed from the basilar membrane band with the highest response level. Since the outer/middle ear transmits very readily at lower frequencies, it is unlikely that any impulsive sound would actually elicit a maximum response above 5 kHz. (The worst case “real life” scenario might elicit a maximum response around 4-4.5 kHz, however, underscoring the importance of an adequate model of the external auditory meatus and a middle ear model with a compliant incudo-stapedial joint.) If the MOMRP is concerned about this deficiency of the Price/Kalb model, a simple fix would be to modify the definition of the ADU in order to more heavily weight the cochlear bands above 5 kHz.
  4. See my above comment #2, as well as our report, which summarize what the characteristics of a standard are. Clearly, what we have seen is not a standard (yet). I would suggest that a board of scientists and audiologists be appointed by the Army to draft such a standard based on the HRED method. I do not believe that Drs. Price and Kalb are qualified to develop such a standard (though clearly they should be integrally involved in the development of the final document).

Given this clarification, I agree that the model “in its current state, without modification or development” is ready to be used in a new MIL-STD intended to replace MIL-STD 1474D.

I agree with the summary statement at the end of the letter.

Reviewer 2 Response

  • Question a.: Although the minimum phase model of hearing protectors was not used to validate the Albuquerque data it is the only way to model an unknown hearing protector. By extending the model of the ear “out” through the hearing protector, the model can be more valuable. In addition, the model can be used to determine the best protector for a particular weapon impulse. As long as the model has been validated against the Albuquerque data, the addition of the minimum phase model of the hearing protector should also be valid.
  • Question b.: The fewer the ADUs the better. Zero would be ideal. A Soldier may be exposed to less than 500 ADUs as long as no TTS is detected. If the Army would like to go with a more conservative 200 ADU limit that will significantly impact their ability to train. We currently do not know how often Soldiers can safely be exposed to these levels.
  • Question c.: Since hearing protectors are generally high frequency devices, I think the model should work for all spectra. Dr. Talmadge could comment better on this.
  • Question d.: I believe there should be a distinction between the MODEL and the IMPLEMENTATION of the model. I believe the reviewers think that the model is fine. It is a logical extension of research carried out on the ear in the past 20 or 30 years. I believe the implementation has problems and it is the responsibility of Dr. Kalb (and Dr. Price) to determine why the software refuses to work on any computer but theirs. Currently the model is not documented well enough so independent researchers could develop their own implementation of the model. As demonstrated at the meeting, the implementation works fine. As tested on my office computer, it does not. I believe this would require a day or two of Dr. Kalb’s time to determine the missing components.

I believe the paragraph LTC Friedl has extracted is a fair representation of the review panel’s conclusions.

Reviewer 3 Response

  • Finding a.: (validation procedure) This is a good point. I thought that Drs. Price and Kalb made the point that using the minimum phase approach allowed them to relate the under the muff data to the free-field data. Then using the Albuquerque data, they showed that this approach, if in error, would be overprotective. This is why they felt that this approach could be extended to any type of hearing protector. Check with them. If this is not the case, then maybe the committee needs to revisit this issue.
  • Finding b.: 500 versus 200 ADUs. Clearly we suggested 500 ADUs. This was approximately equivalent to an 8-hour Leq of 85 dBA. About 8 dB of mean TTS will occur from such an exposure. The 200 ADU would be closer to an 8-hour Leq of 75 to 78 dBA, i.e., no TTS. Certainly the Army could lower the allowable amount of ADUs if the desire was to be more protective. Besides, my feeling is that the problem of misuse of the protection devices is enough of a problem to warrant monitoring TTS in the field in any case.
  • Finding c.: High frequency limit. The comment from the committee did not come from me. I think that it is fine to mention this as a caution in case the model is used to evaluate impulses generated right in or next to the ear, such as Dr. Price primer cap studies. But for practical impulses in which hearing protection is worn, this restriction can be safely ignored. This includes all small arms fire. Remember, any protector will easily attenuate the high frequencies.
  • Finding d.: My feeling is that if the model works, use it. Certainly it would be nice if it were more professionally described. But does it need to be? If the model can be shown to be much better than our current procedure, why wait?

Question: Do I agree with the Army’s interpretation as stated?

Yes. The key to me, however, is that whatever protection used is “worn and properly fit.” The one item that keeps being missed about the Albuquerque study is that we did not let a subject become exposed unless his hearing protection was worn and properly fit.