- Computational and Information Sciences
- Human Research and Engineering
- Auditory Hazard Assessment Algorithm for Humans (AHAAH)
- Sensors and Electron Devices
- Vehicle Technology
- Weapons and Materials Research
- Army Research Office
- Locations and Facilities
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
CDR Rickie R. Davis, Ph.D.
Hearing Loss Prevention Section
Centers for Disease Control and Protection
Daniel L. Johnson, Ph.D.
Interactive Acoustics, Inc.
Carrick Talmadge, Ph.D.
National Center for Physical Acoustics
University of Mississippi
Michael Holthouser, M.D., M.P.H.
Health at Work, Norton Healthcare
April 2, 2001
- Tina A. Rosenthal, M.H.S., Project Coordinator
- William L. Daniels, Ph.D. Program Manager
- John Lively, B.S., Project Associate
- Scientific Peer Advisory and Review Services
- American Institute of Biological Sciences
107 Carpenter Drive, Suite 100
Sterling, Virginia 20164
Phone: (703) 834-0812 Fax: (703) 834-1160
In December 2000, the US Army Medical Research and Materiel Command (USAMRMC) tasked the American Institute of Biological Sciences (AIBS) to convene an independent scientific peer review of the HRED method for assessing the risk of auditory injury for hearing-protected Soldiers exposed to impulse noise from weapon systems during training. The purpose of the review was to evaluate the scientific validity of the HRED method and to determine if in its current state and using sound pressure level data collected by weapon developers, it is a suitable replacement for the current MIL-STD-1474D. The methodology was developed to evaluate and model the parameters of noise and noise effects on human ears.
The review was held on January 30 and 31, 2001 at US Army Research Laboratory (USARL) in Aberdeen, Maryland. The Terms of Reference, list of reviewers, meeting participants, and agenda, were provided in appendices. The reviewers were provided with background information and a CD-ROM that could be used to run the model prior to the meeting.
Charge to Panel
The reviewers were asked to evaluate the HRED impulse noise auditory injury assessment method focusing on the following eight questions provided to the Panel by the USAMRMC.
- Is the assessment method based on sound scientific principles?
- Does the method adequately protect the noise-exposed population from a well-defined auditory injury?
- Has the method been validated by the existing human exposure data sets?
- Is the accuracy of the method in determining acceptable exposure conditions adequate for use as an occupational exposure standard?
- Will the method remain valid and retain necessary accuracy as the impulse noise characteristics change (e.g., longer and shorter sound pressure wave duration, complex and reverberate environments) and as the hearing protections devices change?
- Is the method’s assessment of the XM 232 MACS reasonable and can this assessment be used in place of a human volunteer noise exposure study of this system?
- Does the method provide clear guidance on the hearing protection devices that will be acceptable under the conditions assessed?
- Is the HRED assessment method, in its current state and using sound pressure data as weapon developers currently collect it, a suitable replacement for MIL-STD-1474D for general application in limiting exposure of hearing-protected Soldiers to impulse noise?
Panel members independently reviewed material provided by the presenters prior to the meeting and attended one and a half days of presentations and discussions on the model. The Panel and AIBS staff compiled this evaluation of the model. The Panel approved the final report prior to its submission to the USAMRMC.
HRED Overview: Mr. Bruce Amrein
Mr. Amrein described the mission of the Human Research and Engineering Directorate (HRED) and how this entity fits into the USARL.
The stated mission of the HRED is twofold.
- Provide the Army and USARL with human factors leadership to ensure that Soldier performance requirements are adequately considered in technology development and system design.
- Conduct broad-based program of scientific research and technology directed toward optimizing Soldier performance and Soldier-machine interactions to maximize battlefield effectiveness.
The development of the method to assess risk of auditory injury as a result of impulse noise has its administrative origins in a Work Package assigned to the Visual and Auditory Processes Branch of HRED; WP-4112, Acoustic Modeling for Improving Soldier Performance. This Work Package directs the Branch to expand and improve on existing ear models to include hearing protection algorithms combined with measurement over a wide range of pressures.
Purpose of this Review: LTC Michael J. Leggieri, Jr.
The HRED Auditory Injury Risk Assessment Method was developed as a response to assertions that the existing MIL-STD-1474D overestimates auditory injury risk from impulse noise severely restricting the training use of new weapon systems, according to LTC Leggieri. It is thought that a new method is needed to replace the overly conservative MIL-STD-1474D whose stringent criteria for use of weapon systems actually endangers Soldiers by preventing them from training in a realistic manner. This review is to help the USAMRMC advise the Army Surgeon General regarding the suitability of the HRED method to replace MIL-STD-1474D as the health risk criteria for the Army’s Health Hazard Assessment (HHA) Program.
Review of Present Practice in Auditory Hazard Assessment: Mr. Felix Sachs
Mr. Felix Sachs explained the role the US Army Center for Health Promotion and Preventative Medicine (CHPPM) has in health hazard assessment in general and specifically, their interest in supporting the HRED method. He provided background on theories and practices of weapons impulse-noise evaluation. He introduced issues that must be addressed to develop Army Medical Department (AMEDD)-approved damage risk criteria (DRC) for impulse noise under conditions of protected exposure that are unambiguous and applicable to various scenarios. He discussed hearing protective devices and measurement of attenuation associated with their use. He presented justifications for re-evaluation of current practices citing examples where information about risk and actual hazard are not consistent.
Overview of Model Development: Dr. Richard Price
In his first presentation, Dr. Richard Price discussed background and approach to modeling noise hazard to the human ear. The basis for modeling impulse noise and its effect on the human ear should incorporate enhanced understanding of the mechanism behind hearing loss and should improve upon A-weighted standards and use of peak pressure and duration (for MIL-STD-1474D). Mammalian cochleae are similar and basic research on the cat and chinchilla have led to important insights. (It should be noted however that the cat, chinchilla, and human outer and middle ears are quite dissimilar, making revalidation of the model in humans crucial.) At high levels, mechanical stress at the level of the hair cell is the primary mechanism of hearing loss. Above a critical level, once hearing loss begins, mechanical damage and threshold shift progress quickly. Aspects of human and animal cochlear physiology (peak clipping and middle ear muscle contraction) protect hearing to some extent. The theoretical model is based on data, conforms to human ear structure, and is adaptable to various scenarios and individuals.
Developing the Mathematical Model: Dr. Joel Kalb
Dr. Joel Kalb presented the mathematical model. Initial choices made in formulating an approach to modeling hearing loss were discussed. He explained why the electroacoustical model and the cat ear were chosen as cornerstones of the model. Concepts important to understanding issues that were considered for modeling development were explained. These included points of entry into the model: ear canal entrance (ECE), free field (FF), or ear drum pressure (EDP) level, and the model of basilar membrane function, the Wentzel-Kramers-Brillouin (WKB) model. The model had to take nonlinearities into account such as, annular ligament/stapes displacement at high frequencies and the middle ear reflex. The mathematical model also considered azimuth calculations.
Formalizing the Model for Use as a Standard: Dr. Richard Price
Dr. Price presented data validating the Auditory Hazard Assessment Algorithm (AHAA). Anesthestized cats’ ears were subjected to a wide range of impulses to confirm the model’s predictive capacity. Compound threshold shift (CTS) seen in cats could be predicted using the model. The output of the model is in auditory damage units (ADUs) which relate to calculated displacements in the inner ear and damage resulting from displacements. ADUs correctly predicted CTS and permanent threshold shift (PTS) in cats. Histological examination also revealed that hair cell loss was predicted by the model.
Demonstration of the Model: Drs. Kalb and Price
The model’s ability to handle various human scenarios was demonstrated. Initial settings, inputs, and assumptions were explained. The model assumes that in the most sensitive range, stapes-to-basilar membrane displacement ratio is the same for cat and man. The model has the ability to account for the middle ear reflex; for this demonstration, it was set to moderately strong. Susceptibility to hearing loss was discussed; the model treats the susceptible ear as though it was a nonsusceptible ear impacted by higher intensity stimuli. Currently, the model estimates that humans vary by about 6 dB in susceptibility. The model allows its user to edit waveforms input such as, baseline, peak pressure, and the user can establish start times. The effect of hearing protective devices can be calculated. The output of the model is in ADUs that relate to calculated displacements in the inner ear and damage resulting from these displacements.
Validation of the Model: Dr. Richard Price
Dr. Price presented data from the US Army Albuquerque Study to further validate the AHAA. In this study, human volunteers were exposed to impulses intended to simulate weapons in controlled environments. Waveforms under earmuffs (some modified to reduce performance) and in the free field were measured. Nonlinearity was seen; those muffs modified to reduce performance became nonlinear (worked better) at higher level impulses. Hazard of exposures (safe or hazardous) predicted by MIL-STD-1474D, A-weighted energy, and AHAA were compared to the actual outcome from the Albuquerque studies. MIL-STD-1474D was shown to significantly overpredict hazard, indicating that this standard may be having an unnecessarily deleterious impact on training and readiness of Army units. The A-weighted energy method successfully predicted hazard in 25% of all cases. This method overpredicted true hazard in most cases; it also underpredicted hazard for two of the rifle impulses. Evaluation by AHAA correctly predicted 96.2% safe and hazardous outcomes for the Albuquerque data. In two cases it overpredicted hazard. Evaluations with AHAAH are highly consistent with hearing loss data in these studies.
From the Model Toward a DRC: Dr. Richard Price
Before moving to a formal standard there are still some issues to be settled. Dr. Price stated that the goal of this program to establish damage risk criteria (DRC) that will avoid threshold shifts of 25 dB or greater in 95% of the exposed population may be called into question. Protection for 95% of the population is customary. The model currently considers anything in excess of 500 ADUs unsafe; 200 ADUs and below are safe, associated with zero threshold shift. While humans typically recover after experiencing occasional shifts of 20-25 dB, daily noise exposure at the 500 ADU limit is not safe. The model will continue to grapple with issues regarding angle of incidence. Use of manikins was discussed. Variables regarding attenuation achieved from hearing protective devices (HPDs) are factored into the present model. However, additional variables can be taken into account to make attenuation estimates more accurate. AHAA could conceivably obtain pressure measurements on a manikin; however, free field pressure and attenuation data will be used to calculate the hazard initially. Decisions about other issues such as when to consider muscles “warned” and “unwarned”, if 6 dB (1 standard deviation) is a reasonable assumption to account for human variability (susceptibility), should be addressed in the future.
Evaluations of the Project
Is the assessment method based on sound scientific principles?
The method is a model-based approach. The model uses physical laws to obtain a set of algorithms which can be used to determine the percentage population that would sustain a permanent threshold shift based on impulsive sound measurement under a variety of exposure conditions. This method is far more scientifically sound than MIL-STD-1474D, the existing standard for impulsive noise exposure.
The model can account for impulse noise measurements under free field conditions, at the ear canal entrance, at the tympanic membrane, at probe tip location, and while using a variety of hearing protection devices. Specific anatomical and physiological aspects of the ear canal model are based on scientific principles including effect of pinnae and external auditory canal as an exponential horn, two-piston model of the tympanic membrane, pars flaccida and pars tensa, an ossicular chain which includes compliant malleo-incudo and incudo-stapedial joint. The model correctly accounts for the nonlinear nature of annular ligament and stapes activity. The cochlea is modeled using the WKB method and dividing the basilar membrane into 23 “bins” or segments where acoustic energy accumulates.
The model’s explanation of head transfer function is scientifically based.
Does the method adequately protect the noise-exposed population from a well-defined auditory injury?
Based on the data presented, the method is able to adequately predict the probability of PTS with 96% accuracy. This assumes that the Soldier properly wears recommended hearing protection.
This model is unable (as is MIL-STD-1474D) to account for Soldiers who are not allowed adequate recovery time (estimated at least 24 hours) between consecutive periods of exposure to high level sounds.
Has the method been validated by the existing human exposure data sets?
The model’s predictions were in agreement with the results of the data sets analyzed and presented at this briefing. Thus, within the parameters in which the model will be used, the model appears valid.
Is the accuracy of the method in determining acceptable exposure conditions adequate for use as an occupational exposure standard?
The method represents a potentially powerful tool that could be used as an element of a standard. The HRED program could be used to accurately predict exposure under the conditions that are specified in the model (such as proper fit and wearing of HPDs). By analogy, the OSHA Occupational Noise Standard requires hazard identification, noise measurement, hazard control (including training in the use of HPDs), periodic audiometric testing, recordkeeping and reporting requirements, and program evaluation for performance. This model serves some of those functions such as, hazard identification, exposure assessment, and degree of protection afforded by personal protection equipment, and could serve as benchmark for hearing conservation program evaluation.
Will the method remain valid and retain necessary accuracy as the impulse noise characteristics change (e.g., longer and shorter sound pressure wave duration, complex and reverberate environments) and as the hearing protections devices change?
The proposed model should provide a much more realistic and flexible assessment of auditory risk than MIL-STD-1474D. Unlike MIL-STD-1474D, the model does not assume a specific shape of sound impulse. Unlike MIL-STD-1474D, the model does not rely on determining parameters such as duration that are difficult to measure consistently. The use of the model removes subjective interpretation of waveforms that are different from the norm.
Is the method’s assessment of the XM 232 MACS reasonable and can this assessment be used in place of a human volunteer noise exposure study of this system?
The Panel has enough confidence in the HRED model to recommend that it be used to test the XM 232 MACS in place of exposing human volunteers to noise. The HRED method was seen to predict the Albuquerque data very well. In addition this method had a much lower false positive hazard identification (4%) than what would have been obtained by MIL-STD-1474D (62%).
Does the method provide clear guidance on the hearing protection devices that will be acceptable under the conditions assessed?
The HRED program can be used to assess the relative effectiveness of any HPD, or combinations of HPDs, assuming that these are properly worn. It cannot be overstated that training and supervision in the proper use of HPDs are essential elements in the ability of the model to predict hearing risk.
Is the HRED assessment method, in its current state and using sound pressure data as weapon developers currently collect it, a suitable replacement for MIL-STD-1474D for general application in limiting exposure of hearing-protected Soldiers to impulse noise?
While efforts should continue to evaluate or test the model, the model is considered to be a timely and suitable replacement of MIL-STD-1474D in the application of limiting exposure of hearing protected Soldiers to impulse noise. The model should be used with the minimum-phase model of hearing protection in order to use the hearing protection data as weapon developers currently present to derive the number of ADUs. The Panel concurs that 500 ADUs is an adequate daily limit given no temporary threshold shift (TTS). In the event that subjects are exposed to higher levels of impulsive sound, a lack of adequate recovery time may dictate a lower exposure limit for such circumstances. The model user should assume only unwarned ears at this time.
Overall Strengths and Weaknesses of the HRED Model
The Panel found that the HRED method is far more scientifically based than MIL-STD-1474D. The HRED model should provide a flexible assessment of auditory risk remaining valid and retaining accuracy while accommodating changes in impulse noise characteristics and hearing protection devices. The HRED program can be used to assess the relative effectiveness of any HPD, or combinations of HPDs, assuming these are properly worn and correctly modeled.
The HRED method is based on scientific research and can determine auditory injury, as well as, account for impulse noise measurements under free field conditions, at the ear canal entrance, at the tympanic membrane, at probe tip location, and while using a variety of hearing protection devices.
For any waveform, the model should give the same answer, allowing any group, e.g., weapon developers or hearing protection designers, to obtain reliable and consistent answers. The model can be useful as a tool in the pursuit of new, effective arms that produce fewer ADUs.
The HRED method uses well established theories to account for physiological activity and sound processing in the inner ear. The model divides the basilar membrane into 23 sections. This number of cochlear sections (23) appears to be well advised; it agrees with the 26 critical bands of human cochlea, as accepted in the literature. Critical bands relate approximately to the width of the region of maximum excitation on the basilar membrane. The model correctly accounts for the nonlinear nature of annular ligament and stapes activity and provides a plausible explanation for the head transfer function.
The assumption that the Q of the cochlea varies from base to apex is supported by data. Another important assumption of the HRED model is that the cochlea (basilar membrane) action becomes linear at pressure levels that exceed threshold. The model assumes the basilar membrane is acting nonlinearly at levels which the ear can be damaged. Using this assumption, the WKB is a good approximation.
The model was validated by human exposure data obtained in Albuquerque, New Mexico. The HRED method has been shown that it can predict the probability of permanent threshold shift with 96% accuracy assuming the proper use of hearing protection. It is felt that the HRED model can be used to test potential health hazards associated with impulse noise levels in excess of 140 dB (e.g., XM 232 MACS fire). Therefore, it will be able to protect the noise-exposed population from auditory injury.
The HRED method can contribute to development of an occupational exposure standard. Elements that can be obtained using HRED methodology include hazard identification, exposure assessment, and appraisal of protection afforded by personal protection equipment.
The model is not able to account for the temporal pattern of exposure of Soldiers who are not allowed adequate recovery time (estimated at least 24 hours) between consecutive periods of exposure to high level sounds. This is also lacking in MIL-STD-1474D or any other damage risk criterion.
The concept of “warned” responses was met with skepticism. Reliance on a warned condition may be underprotective for many individuals in all situations and for all individuals for many conditions. The benefit of the acoustic reflex has not been established for the general population, although further research may provide evidence for the warned response. It is possible that a visually stimulated startle response could evoke a middle ear reflex since the eyelids and middle ear reflex are both innervated by the same facial nerve.
The middle ear model is only accurate to 5 kHz. An improved middle-ear model (and outer ear model) could increase the frequency range over which this model is accurate. It is possible that at this time the model could underestimate the energy transmitted to the cochlea at higher frequencies. Because the military intends to use the HRED model for impulsive sounds, mostly residing in lower frequency ranges, this may not be a major failing. However, the model is being considered for use as an ANSI standard. The model should not be adopted for use with this standard if its ability to estimate energy in the ear canal at high frequencies is compromised. If the HRED model cannot confirm its integrity at high frequencies, then a disclaimer should be written.
The model states that ADUs are closely correlated with basilar-membrane peak amplitudes and resultant hearing threshold shifts. At present, it appears that ADUs and hair cell damage (IHC, OHC) are more closely related. Documentation demonstrating the theory that ADUs and basilar-membrane peak amplitudes are interrelated is lacking. In lieu of providing more documentation on this, HRED developers may choose to explain and analyze ADUs in terms of hair cell damage.
Currently, the HRED model does not account for direct conduction of impulse noise into the skull, bypassing hearing protection. If the Soldier is directly coupled to the noise source, the HRED method may not produce reasonable results. Further research may be needed to determine acoustic losses between shoulder mounted weapons and the ear.
The HRED model needs to be better characterized and verified. In its present state individual versions of the model cannot be generated or made operational by anyone other than its designers. It cannot be duplicated in another system. It should be described in such a way that a qualified worker in the field could build a predictive tool using another development system (S+, VisualBasic, Pascal, etc.) that would produce the same quantitative output (ADUs). Outside of Dr. Kalb’s development environment there have been a number of people unable to run the model. A more stable form of the program needs to be developed and supported.
Several issues in noise research remain unresolved, such as, the time needed for repair and recovery following trauma to the ear. Also, not known is whether the ear sustains damage at levels below those causing TTS. Data indicate that some form of temporary hearing loss precedes PTS. However, some researchers believe that TTS and PTS may be produced through different auditory mechanisms. If this were the case, the Albuquerque results could not verify the model by themselves. However, it was noted that in the cat-based model, the model did agree with the cat PTS data. The question of whether a single noise exposure makes the ear more vulnerable to later exposures is an area under current investigation by Liberman and others. Liberman indicates that noise exposure to the ear not only damages sensory cells but fibers that support the basilar membrane. It is not known whether these fibers recover.