An Investigation of Unilateral Hearing Loss amongst Professional Flight Crew.

Dr Michael Bagshaw

Head of Occupational and Aviation Medicine, British Airways

April 2001

Professional flight crew undergo periodic audiometry as part of the regular medical examination required to maintain a professional flying licence. It was observed that some flight crew employed by a major international airline appeared to have unilateral hearing loss. This showed characteristics of noise induced hearing loss (NIHL) in some cases, suggesting the possibility of an occupational cause.

A study was undertaken to make sample measurements of the noise levels on the flight decks of the different aircraft types in the airline fleet, including the noise levels produced by communications heard through the flight crew headsets. Measurements were also made with active noise reducing (ANR) headsets during some flights. Additional laboratory measurements were made, utilising the facilities of the Institute of Sound and Vibration Research at the University of Southampton.

Members of the flight crew of commercial aircraft registered in the United Kingdom are required to hold a professional licence issued by the Civil Aviation Authority (CAA). This licence must include a valid Class 1 medical certificate issued by the CAA or by an authorised medical examiner (AME) acting on its behalf. At the time of this study, the standards were laid down by the CAA in conformity with the recommendations of the International Civil Aviation Organisation (ICAO) [1]. These have since been superceded by the Joint Aviation Requirements (JAR-FCL 3) of the European Joint Aviation Authorities (JAA).

The hearing requirements for Class 1 certification are based on periodic pure-tone audiometry, the periodicity depending on age, and is to establish fitness to operate an aircraft safely rather than to screen an individual's hearing.

The CAA standards required that there should be no hearing loss in either ear separately of more than 35 decibels (dB) at any of the frequencies 500, 1000 or 2000 Hz, or more than 50 dB at 3000 Hz [2]. However, a flight crew member with a hearing loss greater than this could be declared fit if a practical hearing test was satisfactory and it was considered that there was no risk to flight safety.

The JAA standards for revalidation or renewal of a Class 1 medical certificate are the same.

Although most structures in the inner ear can be harmed by excessive sound exposure, the sensory receptor cells of the cochlea are the most vulnerable.

Damage to the stereocilia bundle is often the first structural alteration observed. It is also noted that once a large number of hair cells are lost, the nerve fibres to that region also degenerate resulting in an irreversible hearing loss [3].

Beside the well described changes in stereocilia and hair cells, postsynaptic damage at the synaptic pole of the inner hair cells is also prominent after acoustic trauma. It entails a disruption of the dendrite ending of the spiral ganglion neurons below the inner hair cells, leading to synaptic uncoupling [4, 5, 6, 7]. Recently it has been suggested that dendrite damage might be due to excessive release of neurotransmitter from the inner hair cells, which is toxic (excitotoxic) to the structure and function of the spiral ganglion [7].

An alternative explanation by Bohne in 1976 [8] suggests that high ambient noise levels can cause vascular change, the extent of damage depending on the frequency, intensity and duration of the noise. The inner ear is particularly vulnerable to a reduction in blood supply because of its dependence on a single end-artery without cross-circulation.

Decreased capillary flow in the spiral ligament and stria vascularis of the Organ of Corti has been observed following exposure to excessive noise [8]. Quirk et al have demonstrated vascular changes in the cochlea in response to loud noise [9].

An association between smoking and high frequency hearing loss has been reported by a number of workers [10, 11, 12, 13, 14]. Cigarette smoke has been implicated as a direct ototoxin (ie a nicotine effect) [15] or as an inducer of ischaemia through production of carboxyhaemoglobin, vasospasm or promotion of arterio-sclerosis [11, 12, 16]. The observed association between hearing loss and smoking status for ‘former’ smokers implies that there may be cumulative and permanent effect [17].

The primary measure of hearing loss is the hearing threshold level, which is the level of a tone that can just be detected. The greater the hearing threshold level, the greater the degree of hearing loss or partial deafness. An increase in a hearing threshold level that results from exposure to noise is called a threshold shift.

Some threshold shifts are temporary and diminish as the ear recovers after termination of the noise. Frequently repeated exposures can produce temporary threshold shifts that are chronic, although recoverable when the exposures cease. When a threshold shift is a mixture of temporary and permanent components, it is a compound threshold shift [18]. When the temporary components of a compound threshold shift have disappeared, the remaining threshold shift is permanent and persists throughout the remainder of life.

Temporary threshold shifts can vary in magnitude from a change in hearing sensitivity of a few decibels restricted to a narrow region of frequencies, to shifts of such extent and magnitude that the ear is temporarily, for all practical purposes, deaf. After cessation of an exposure, the time for hearing sensitivity to return to near-normal values can vary from a few hours to two or three weeks [18].

Ward has shown [19] that noises with energy concentrations between about 2000 and 6000 Hz produce greater temporary threshold shifts than noises concentrated elsewhere in the audible range. In general, A-weighted sound levels must exceed 60-80 dB before a typical individual will experience temporary threshold shifts even for exposures that last as long as 8 to 24 hours. All other things being equal, the greater the noise level above 60-80 dB and the longer the time of exposure, the greater the temporary threshold shift. However, exposure duration beyond 8 to 24 hours may not produce further increase in the magnitude of the shift [20, 21].

Under certain conditions, contraction of the muscles of the middle ear can offer significant protection from exposure to intense sound. Also, there is less temporary shift when an exposure has frequent interruptions than when an exposure is continuous [18].

People differ in their susceptibility to temporary threshold shifts. Unfortunately, these differences in susceptibility are not uniform across the audible range of frequencies. One individual may be susceptible to noises of low pitch, another to noises of medium pitch, and another to noises of high pitch [18].

Thus the threshold for hearing damage varies between individuals so that although the reference action levels are treated as absolutes in health and safety legislation, they are not such absolutes in biological terms.

The first effect of exposure to excessive noise is a temporary loss of hearing, maximal at 4000 to 6000 Hz, which is sometimes accompanied by a persistent noise in the ear (tinnitus). This temporary threshold shift lasts for a few hours and the rate of hearing recovery is exponential, returning to normal within 12 hours.

The degree of temporary threshold shift and its rate of recovery are not related to the susceptibility of an individual to inner ear noise damage [personal clinical observation]. If the exposure to noise is repeated and prolonged, a permanent threshold shift may occur affecting the same 4000 to 6000 Hz frequencies. As exposure continues or is repeated, what began as a localised audiometric dip becomes broader and lower frequencies become increasingly affected, often accompanied by the development of tinnitus, giving a permanent threshold shift.

The individual may be completely unaware of any defect in hearing until the speech frequencies are affected. Normal conversation frequencies lie between 500 and 3000 Hz, with the vowel frequencies between 500 and 2000 Hz and the consonants between 1000 and 2000 Hz. Speech becomes increasingly difficult to understand, with the first symptom being difficulty of discrimination in a noisy environment, but it may take several years for severe deafness to develop. For this reason, routine periodic audiometry is an important component of hearing conservation programmes in noisy occupational environments, to monitor compliance with hearing protection. It is similarly an important component of the periodic screening medical for flight crew.

Permanent threshold shift is irreversible and the rate at which it proceeds depends upon the sound pressure level of the noise, the duration of continuing exposure and the individual susceptibility.

The frequency content of the offending sound is important because high frequency noise can be more injurious to hearing. It is for this reason that A-weighting is used for measuring noise which may be harmful to hearing. All regulations and standards dealing with noise exposure accept A-weighting as the best physical measure of noise correlated with chronic hearing damage.

The normal ageing process involves a reduction in hearing acuity at higher frequencies known as presbycusis, and when the effects of noise induced hearing loss (NIHL) are added it can become a severe handicap. Presbycusis appears to have a multifactorial origin [22] involving physiological ageing, high ambient noise levels, infection, ototoxic drugs, and trauma on the auditory system. Genetic factors also seem to play a role.

It has already been stated that the inner ear is particularly vulnerable to a reduction in blood supply. It has been shown that progressive degenerative change occurs mainly in small calibre arteries [23]; the larger cochlear, vestibular and labyrinthine arteries appear unaffected even in advanced age.

Presbycusis is manifested by an increase in hearing threshold levels at higher frequencies, initially affecting 8000 Hz, but affecting lower frequencies as age progresses.

The human ear has a remarkable ability to discriminate and ‘hear out’ some sounds from a background of other sounds, although there are limits to this ability [27]. Through a process known as ‘masking’, noise can make a sound inaudible. ‘Partial masking’ can cause the speech from a headset used in a particular ambient noise to seem quieter than the same speech at the same volume setting but without the background noise. This is similar to the effect experienced with a car radio - when leaving a motorway and stopping at a junction the radio can be uncomfortably loud, whereas it was comfortable against the background noise when driving at speed.

The ear can be thought of as analysing sounds through a set of filters which can be tuned to any centre frequency. Below about 500 Hz the bandwidth is approximately constant, and above 500 Hz it is approximately proportional to frequency. Above 500 Hz the critical bandwidth is reasonably approximated by 1/3 octave bands and so these can be used in approximate masking calculations.

The ear summates the sound energy within a critical band, and therefore the detectability of a signal in noise is determined by the total signal energy within a critical band relative to the total noise energy within the same band. The signal-to-noise ratio (S/N) within a critical band at which the signal can just be heard is dependent on frequency; however, to a first approximation a signal will be audible if the 1/3 octave band S/N is -5 dB [27].

At higher intensities the critical bandwidths are increased, and the threshold S/N of tonal signals may rise by 5-10 dB [37]. In addition, low-frequency noise can have a greater masking effect on higher frequency tones than that indicated by the size of the critical bands, an effect known as ‘upward spread of masking’. For noise spectra with pronounced low frequency components (differences between successively higher frequency 1/3 octave bands greater than approximately -10 dB) it would also be necessary to take account of this factor.

Commercial aircraft operated for purposes of public transport carry a minimum flight crew of two pilots. The captain occupies the left hand seat of the flight deck, while the first officer occupies the right hand seat. Many aircraft carry a third crew member, a flight engineer, who usually occupies a position behind the first officer.

Commercial pilots normally begin their career as first officers and may spend a number of years (average 7-10) in the right hand seat before progressing to captaincy in the left seat (average 20-25 years). Pilots who fulfil training or supervisory roles may occupy either seat when acting in that capacity.

Pilots may fly a number of aircraft types during the course of their careers, and can move from being captain of one type to being a first officer in another and vice versa.

The flight deck is the regular place of work for commercial aircrew and a normal duty day may involve up to 12 or 14 hours in this environment. This is in contrast to military aviators who frequently intersperse flying duty with other squadron or regiment activities, as well as spending a proportionately greater ground time on pre- and post-flight activity such as planning, briefing and debriefing.

Another contrast is in the form of protection provided for the crew to counter the physiologically hostile environment in which they work. The military ethos recognises the potential dangers, both military and physiological, and there is an acceptance of the need to provide personal protective equipment which may compromise comfort.

Conversely, there is an expectation that the airliner flight deck will be non-hazardous and provide a comfortable working environment, so it has evolved as a ‘shirt-sleeved’ office environment with personal protective equipment limited to anti-glare spectacles.

As part of this evolution, radio communication is facilitated by the use of light weight headsets which provide no discomfort to wear, but give little or no noise attenuation. A further concession to the ‘shirt-sleeved’ office environment is the avoidance of use of the radio intercommunication system (intercom or interphone) for verbal communication between crew members on the flight deck, using instead direct inter-personal speech.

The standard operating procedure is for the telephone ear piece to cover the ‘outboard’ ear (the right ear for the occupant of the right hand seat, and the left ear for the occupant of the left seat) to receive radio-telephonic communication (R/T), and to leave the ‘inboard’ ear uncovered to enable direct verbal communication to occur between crew members. The lightweight headphones are fitted with a boom microphone for R/T transmission.

An exception to this procedure is the Concorde supersonic transport aircraft and the ATP turboprop aircraft, both of which are acknowledged to have noisy flight decks. Crew members of these aircraft wear headsets with some basic noise attenuating properties, and use the intercom as a routine. This is accepted by the crew in view of the relatively short flight times of legs flown by these types.

External communication is with air traffic agencies and company operations departments and utilises very high frequency (VHF) and high frequency single side band (HF SSB) radio-telephonic transmission (R/T). As well as the speech communication, there is frequently much background R/T noise such as hiss and ‘static’.

To obtain meaningful measurements of noise, it is necessary to include a frequency or octave band analysis of the noise spectrum, and to derive an equivalent continuous sound level over the sampling period (L_eq).

This will be considered in more detail in the methodology.

Sources of noise can be occupational or non-occupational (social).

Potential non-occupational sources are many, but could include gun shooting, participation in musical bands and orchestras, listening to personal stereo equipment and the use of tools and building equipment. Of these social sources, only gun shooting without the use of hearing protection would be likely to produce the pattern of unilateral noise induced hearing loss observed in the subject group, and this is not a common recreational activity for professional flight crew.

Occupational sources of noise for flight crew can be on or off the flight deck.

Off the flight deck, noise levels on the aircraft parking area (the ramp) are such as to require the use of hearing protection, and this is available for use by flight crew when performing external checks of the aircraft prior to departure. Noise on the ramp would affect both ears and would be unlikely to cause unilateral hearing loss.

On the flight deck, ambient noise is generated from many sources, including aerodynamic, propulsion, avionic and electrical. The frequency spectrum is complex and varies according to the equivalent air speed, power setting and altitude, and the sound pressure level varies at different points on the flight deck.

The widespread practice of listening to radio communications through a headset covering one ear only, leaving the other ear uncovered to allow direct conversation between crew members, was thought to be the most likely source of potentially harmful noise. Sample measurements of the noise levels on the flight deck, including those produced by communications through the headsets, formed the basis of this study.

The simplest means of reducing noise at the ear would be to wear earplugs under the headset. They are light, cheap and easy to maintain. However, it has been shown that wearing foam ear plugs under the headset decreases speech intelligibility dramatically and requires the intercom volume to be maximised to ensure speech understanding [38]. In addition, many flight crew would find them uncomfortable to wear for long periods of time.

Noise attenuating headsets may act passively or actively.

A passive noise attenuating headset is composed of four basic parts - the shell, the seal, the internal damping and the R/T transducer. The attenuation characteristics of the headset depends on the frequencies involved.

Attenuation of low frequencies (below 400 Hz) is controlled by movement of the earshell against the head. Thus the important parameters of the earshell are volume, stiffness of the earshell seals and the fit of the shell on the head. Increasing shell volume increases low frequency attenuation, but increases shell bulk. Doubling the volume improves low frequency attenuation by 6 dB [39].

Attenuation of intermediate frequencies (400-2000 Hz) depends on the transmission loss of noise through the shell walls, and thus the type and mass of shell material are important. The greater the mass, the greater the attenuation.

Above 2000 Hz the noise field inside the shell is complex and it is the damping material therein which provides the attenuation. The installation of the R/T transducer in the shell is critical and improper support of the structure can lead to significant reduction in attenuation of frequencies above 500 Hz [39].

Headset design is necessarily a compromise between weight, bulk and individual acceptability. The best passive noise attenuating headsets can give anything up to 30 dB attenuation at the high frequencies, but only up to 10 dB at the lower frequencies [39]. However, prolonged wear can lead to complaints of discomfort due to applied pressure at the side of the head to ensure an adequate fit of the seals and pressure from the headband due to the weight of the headset.

One passive headset type is manufactured by Peltor and this is already in use by the airline in the ATP during relatively short trips. However, flight crew consider it unacceptable for long-haul flying due to the increasing discomfort when worn for long periods of time.

The principle of active hearing protection was described by Lueg in 1936 [40], but active noise reduction (ANR) systems have only become commercially available on a large scale in recent years.

An ANR headset works by continuously sampling the noise inside the earshell using a miniature microphone. The sampled noise is then phase inverted 180 degrees by an electronic circuit and reintroduced through the earphone speaker. This reduces the noise levels inside the earshell by destructive interference of the acoustic field thus cancelling out the original noise. Reductions of 15 dB(A) or more are possible [39], but limitations in the cancellation mean that ANR systems are more effective in the low frequency range.

Experience of ANR in aviation has been reported from within the military environment, particularly with respect to helicopter operation. Wagstaff et al provide advice for evaluating ANR for use in an aviation operational environment [41] , with specific reference to military rotary operations. Pelausa et al give similar consideration in the context of the Canadian military [42]. There is little published with respect to use of ANR in commercial aircraft.

ANR circuitry can be installed in the ear shell of relatively light headsets. It is possible to design such a headset which provides some passive attenuation as well as active noise reduction. Passive attenuation is more effective at higher frequencies, whereas active attenuation is more effective in the lower frequency range. This can lead to a light and efficient noise attenuating headset which is acceptable to flight crew in the ‘shirt-sleeved’ office environment of the airliner flight deck.

In the long term for new generation aircraft, noise reduction around the crew seating positions by active means may be effective, but this is not yet an option. Although noise cancellation of blade passing tones is relatively well established for turboprop aircraft, it is difficult to achieve cancellation with the wideband continuous noise spectra found in turbojets and fan-jets. Theoretical and experimental work is not well developed for these applications.

The periodic pure tone audiometry required as part of the regular medical examination may be performed by any one of a number of authorised medical examiners throughout the UK. There is no standardisation of equipment, technique or training. However, since this study was examining evidence for unilateral hearing loss, each audiogram acted as its own control.

On completion, the audiogram is forwarded by the AME to the Civil Aviation Authority (CAA) where it is stored in the medical records system.

The CAA medical records system contains audiometric records only from 1988 and this determined the starting date for study data. The sample size was determined by the availability of audiograms in the CAA record system from flight crew employed by the airline under study. Although the airline employs almost 3,500 flight crew, it proved possible to extract audiometric records for only a proportion of these.

A sample of 752 CAA records of holders of Class 1 certificates employed by the airline under study, with 2 or more audiograms recorded between 1988 and 1994, was examined for evidence of hearing loss in excess of 25 dB at any frequency. The search was refined to show loss in excess of 25 dB at a frequency of 4000 Hz in either or both ears, this being considered an indicator of early noise-induced hearing loss. The hearing loss was scored only if it occurred in 2 or more audiograms in an attempt to exclude temporary threshold shifts or errors in recording.

A crude prevalence of NIHL in the flight crew population was estimated by determining the proportion of records showing unilateral loss in excess of 25 dB at 4000 Hz.

Having been alerted to the possibility of an occupational hearing hazard, the study was directed at seeking the source and a solution, rather than establishing a truly accurate prevalence which was impossible within the constraints of available data.

Measurements were made during scheduled commercial flights on the aircraft types and routes shown in Table 1.

Table 1
Aircraft types and Routes Measured

During each flight the noise levels were recorded from a headset connected in parallel with the captain's or first officer's. The standard Sennheiser HME/HMD 410 non-noise attenuating headset was used, except in Concorde, where the standard headset is the Racal Astrolite, and in the ATP where a Peltor passive noise attenuating headset is used.

A number of production and prototype noise attenuating headsets were tested in addition during the flight in the Boeing 747-436, and a further prototype active noise reducing (ANR) headset was tested on the Concorde return flight.

Measurements were carried out in the laboratory test facilities of the Institute of Sound and Vibration Research at the University of Southampton to determine the noise reduction characteristics of the standard HME/HMD 410 headset, and of the available ANR headsets.

There are currently no regulations specifically addressing noise levels in aircraft. The UK Noise at Work Regulations 1989 [50] do not apply to aircraft, but there is no technical or scientific reason why they should not be used for guidance. The Noise at Work Regulations are the UK implementation of a European Directive [51], and similar regulations are in force in each member state of the European Union to minimise risk to hearing. The provisions of the regulations and the noise action levels specified are discussed below.

The Noise at Work Regulations address the general problem of noise in the workplace and the risk to hearing. This noise is usually generated by machinery and equipment and is normally measured with a sound level meter. Care is taken when measuring these levels to place the sound level meter so that the meter operator and other individuals present do not unduly influence the readings by causing reflections or sound shadows. The reading is thus representative of an undisturbed sound field. The Noise at Work Regulations 1989 specify maximum sound exposures in terms of undisturbed field levels.

There is at present no specific criterion or method for assessing noise levels from headphones and relating these levels to the risk of hearing damage. To estimate the risk the headset output sound levels must be accurately measured and converted to their equivalent undisturbed field levels. Although there is no standard procedure at present, a method [52] is available based on the use of a manikin, or artificial head and torso.

An acoustic manikin is designed to be geometrically and acoustically representative of a median human. It has external ears moulded in a flexible rubber-like material. It also has ear canals with internal microphones to measure sound levels at the eardrum positions. Any headset, ear insert, hearing aid, stethoscope or other device worn by a human can be worn in the same way by the manikin. Noise levels measured at the eardrum microphone of the manikin are representative of noise levels which would be present at a median human eardrum under the same conditions. For this study, the Institute of Sound and Vibration Research made available a manikin known as 'Kemar' (the Knowles Electronic Manikin for Acoustic Research) [53]. Other similar manikins are available, for example the Head and Torso Simulator manufactured by Bruel & Kjaer.

An alternative to the use of a manikin would be the placing of miniature microphones at the entrance to the ear canal of a human subject wearing headphones. This method can be satisfactory, but the microphone leads could obstruct the flight deck, could restrict the wearer's movement, and would take time to fit while the crew were preparing for flight. The manikin method was selected to avoid interference with the flight crew and flight deck operation.

Because the manikin was to be used for headset measurements it was also used to measure the ambient flight deck noise. Using the manikin ensured that the measuring equipment was effectively self-contained with no trailing leads and no need to find attachments for a microphone on an extension lead. This in turn allowed equipment to be set up quickly with minimal disruption to the flight deck routine. A hand held sound level meter was used in flight for 'spot' measurements throughout the flight deck and to check for differences in the sound levels at the different crew positions.

A schematic diagram of the equipment used to record noise on the flight decks is shown as Figure 2(a).

In each aircraft, the ‘Kemar’ manikin was strapped into the observer’s seat behind the captain. The manikin normally wore one of the spare headsets carried on board which was placed over the left ear, leaving the right ear uncovered. The headset earphones were usually connected in parallel with the captain’s, using a splitter connector, so that the manikin and the captain "heard" the same communications through the same type of headset at the same volume. Occasionally the manikin’s headset was paralleled to the first officer’s instead.

Calibrated tape recordings were made of the Kemar ‘eardrum’ microphone outputs during each flight. Two sound level meters were used to amplify the microphone signals and provide switched gain controls and a portable Digital Audio Tape (DAT) recorder recorded the line outputs from the sound level meters. Both sound level meters were set to linear frequency weighting so that any further weighting correction back to undisturbed field could be made during laboratory analysis of the tape, and the A-weighting applied at that time.

In addition to the manikin and recording equipment, a self-contained hand-held sound level meter (B&K Type 2336) was taken on each flight. During cruise, when noise levels were steady, this meter was used to give a direct read out of sound level to act as an independent cross-check of background recorded using the manikin method. It was also used to measure sound levels at each seating position on the flight deck so that variation from seat to seat could be assessed. This was to allow the sound levels at each crew member’s seat for the whole flight to be estimated from the measurements at Kemar’s seat.

Each component of the tape recording chain was calibrated before each trip.

To provide protection against hypoxia whilst maintaining a ‘shirt-sleeved’ environment, the aircraft cabin and flight deck are pressurised. However, for structural reasons the cabin altitude is not maintained at sea level, but climbs to a maximum of 8000 feet. At this altitude, the pressure falls to _ of sea-level atmospheric pressure, a value of 570 mmHg in the standard atmosphere. As the aircraft climbs to cruising altitude, the cabin altitude increases at a maximum rate of 500 feet per minute. The cabin altitude remains constant for the duration of the flight and then descends at a maximum of 300 feet per minute to ground level at the end of the flight.

Theoretically, changes in ambient atmospheric pressure can slightly affect calibrators, and so calibrations were avoided in the aircraft during flight. In fact, within the IEC 942 specification range of atmospheric pressure between 650 and 1080 hPa, the maximum deviation for the calibrator used in this study is 0.03 dB at 20 deg C [54]. Thus in practice there would have been negligible change in the readings.

The microphones used on the flight deck are of the open noise-cancelling type, which is open at the front and back so that sound can reach both sides of the diaphragm. The diaphragm moves in response to the instantaneous pressure differences between the front and rear ports, and the output voltage is proportional to the sound pressure gradient rather than the sound pressure. Ambient pressure acts directly on the diaphragm, with pressure equalisation from the other side. As the ambient air pressure decreases, so too does the stiffness of the vibrating system, with the result that the sensitivity of the microphone increases. However, for microphones with normal side or back vented static pressure equalisation, i.e. those in aviation use, the effect of varying altitude is very small. A rate of climb of 500 m/s at ground level affects sensitivity by less than 1 dB and for all practical purposes can be ignored [55]. Similar argument applies to the telephone diaphragm in the headsets, and it is unnecessary to change volume settings as the aircraft climbs and descends [personal observation].

A schematic diagram of the analysis equipment is shown in figure 2(b) above.

Considerations of medical confidentiality constrained the availability of audiometric information from the CAA and complicated interpretation of serial audiograms. It proved impossible to determine age or sex for the subjects studied, although the age distribution of the pilot population employed on 1 January 1998 by the airline under study was as follows (this does not include the flight engineers):

The number of individuals exhibiting a hearing loss greater than 25 dB at 4000 Hz, in either or both ears, at 4000 Hz plus 6000 Hz in both ears, and at 6000 Hz in both ears, was determined by inspection of the audiometric records. There were insufficient records showing unilateral loss at 6000 Hz to allow differentiation between ears.

Prevalence was determined by calculating the proportion of total audiograms examined meeting the criteria specified, expressed as a percentage.

The following table shows the result.

The preponderance of left sided hearing loss may well be indicative of the proportionally greater working time spent in the left hand seat during an average career.

There has been no case of a crew member employed by this airline becoming unfit to hold a Class 1 medical certificate as a result of NIHL. However, there have been cases of crew members becoming unfit as a result of tinnitus, associated with bilateral hearing loss, and taking early retirement on the grounds of ill health.

Figure 3 shows the variation in noise levels throughout a representative flight. This example is the Boeing 767 flight from Vancouver to Heathrow.

Figure 4 shows two noise spectra averaged over a flight, again using the flight from Vancouver as an example. Ambient noise levels experienced on the flight deck are shown by the hatched histogram; noise levels with the additional contribution from the headset are shown by the unfilled histogram. As before, these spectra are undisturbed field equivalents measured at the observer’s position.

Notes: 1. All noise levels are rounded to the nearest decibel

2. The Sennheiser HMD/HME 410 headset is normally used in all aircraft except the BAe ATP, where Peltor noise-excluding headsets are used, and Concorde where Racal Astrolite are used.

3. A variety of headsets were used during the Boeing 747-436 flight; the headset noise level is the average level for the whole flight.

4. A prototype headset was worn during much of the flight in Concorde G- BOAC, the Captain stated that he would have preferred to set a higher headset output had this been available.

5. Access to the ATP flight deck was not possible during flight and therefore noise levels at the Captain and First Officer’s positions could not be checked.

Ambient noise levels in the captain’s or first officer’s seats were usually slightly higher than those measured at the observer’s position. The relative levels at each seat were checked with a sound level meter during each flight and adjustments have been made to the levels measured at the observer’s seat to give the ambient noise levels experienced by the crew members. The measurements were A-weighted and taken at one point on the flight deck. The relative A-weighted level at different seats remained constant throughout the flight.

No adjustments are necessary for the noise levels from the headset which are not dependent upon the seating position of the wearer. These noise levels together with the duration of the noise will give the flight crew’s noise exposure during the flight.

Inspection of Table 2 shows that the sound levels from headsets are not related to the ambient noise on the flight deck in any straightforward manner. Individuals’ hearing differences and/or their personal preferences appear to be very large factors in setting volume controls; there are large differences among the crew members. The higher headset outputs are distributed through the fleet and not confined to the ‘quieter’ flight decks.

A further observation is that volume gain levels from a headset appeared to be set higher on the return sector of a long-haul duty, which could be associated with the individual being tired on the return from a tour of duty and having a need for greater subjective sensory input.

Figure 5 shows the variation in noise level during the flight from Heathrow to New York in a Boeing 747-436. During this flight several different headsets were tried by the captain and also worn by the Kemar manikin.

The letters A - F on the figure show when each headset was in use, as follows:

A: Sennheiser Noisegard HDC 451, ANR switched on

B: Bose, ANR switched on

C: Sennheiser HMR 25, passive noise reduction only

D: Telex Airman, ANR switched on

E: Sennheiser prototype ANR headset based on the HME 25, ANR switched on

F: Sennheiser HME 410 KA, the standard headset with no noise reduction

G: Sennheiser Noisegard HDC 451 (as A above), ANR switched off for Kemar, but switched on for the captain.

Subjective preference for the Sennheiser ANR headset was expressed by all flight crew acting as subjects for the study. This headset is light and comfortable to wear, and provides a degree of passive attenuation in the higher frequency range. It was considered acceptable for use in the ‘shirt-sleeved’ flight deck environment for long duty periods.

This headset was also acceptable to the flight crew union.

All the passive headsets considered were rejected on the grounds of bulk, weight and/or discomfort when worn for extended periods on inter-continental routes. Passive headsets with large ear shells, such as those manufactured by Peltor and David Clark, were rejected subjectively at the outset of the study and were not investigated.

Current noise limits in industry in the UK are set by the Noise at Work Regulations 1989 [50]. These specifically exclude the crews of aircraft, hovercraft and sea-going ships from their scope of application (Regulation 3). The exclusions are a legal technicality. A reasonable and prudent employer would take note of them and apply them in aircraft despite the exclusion.

The UK Noise at Work Regulations are based on EC Directive 86/188/EEC [51]. Similar regulations apply throughout the European Union.

The Noise at Work Regulations specify that every employer shall reduce the risk of damage to the hearing of his employees from exposure to noise to the lowest level reasonably practicable (Regulation 6). In addition the regulations specify various Action Levels.

The Second Action Level is a personal daily exposure of 90 dB(A). At or above this Action Level the employer must provide ear protection and employees are obliged to wear it. The regulations also state that, above the Second Action Level, the reduction in noise exposure should be achieved by means other than hearing protectors where reasonable practicable.

There is a third action level, known as the Peak Action Level. The Peak Action Level is an instantaneous level of 200 Pascals (140 dB) and must never be exceeded without protection, no matter how short the exposure may be. The Peak Action Level is unlikely to be reached in normal aircraft operations.

It is important at this stage to distinguish between noise levels and noise exposures. A noise exposure is dependent upon the noise level but is also dependent upon time or duration. The noise exposure is numerically equal to the average noise level for a standard 8-hour shift. For example, a person working for a total of eight hours a day in a steady noise level of 75 dB(A) will have a noise exposure of 75 dB(A). If the individual remains in the noise level for a shorter time the exposure is reduced; for example, four hours a day in a steady 75 dB(A) noise level, will give an exposure of 72 dB(A). Conversely, working longer hours will increase the noise exposure; for example 16 hours a day in a steady 75 dB(A) noise level will give an exposure of 78 dB(A).

In guidance notes published by the Health and Safety Executive [56] it is stated that there is "a quantifiable risk" of hearing damage from noise exposures between 85 and 90 dB(A), and a "residual though small" risk below 85 dB(A).

Robinson, Lawton and Rice [57] have studied what little evidence is available on occupational hearing loss from low-level noise. In a report produced for the Health and Safety Executive they state that there is a negligible effect on hearing from daily noise exposures of 75 dB(A). Above 75 dB(A) but below 85 dB(A) "long-term exposure to noise has some effect but the amount of noise-induced threshold shift is so small as to be practically undetectable in individual cases and only measurable in a statistical sense. Moreover it is so small as to be overshadowed by the loss of hearing associated with advancing age, whether due to natural causes or the insults of daily living".

An employer has a duty of care to employees. The duty is based on what the reasonable and prudent employer knows or ought to know. Where there is developing knowledge the employer must keep up to date, and where the employer has greater than average knowledge of the risks, the employer may be obliged to take more than the average or standard precautions.

If a noise-induced hearing loss can be established and be shown to be partly or wholly attributable to the noise at the workplace, and the employer ought to have anticipated that this hearing loss might occur, and could have taken measures to reduce the noise that were reasonably practicable and cost effective, then the employer is liable even though no regulations may have been broken. In this particular case, regulation 6 of the Noise at Work Regulations 1989 [50] would have been breached in other, non-exempt, occupational settings.

Recent judgements in the UK, for example, Cropper v Ford and Hulse v ICI have awarded damages for hearing impairment in cases where noise exposures were below action levels specified in the Health and Safety Executive’s Code of Practice for the exposure of employed persons to noise [59]. This was a document which, from 1972 until the Noise at Work Regulations were published, set the maximum daily noise exposure to 90 dB(A), and was the accepted standard of the time.

On this basis, taking into account the Noise at Work Regulations, the comment that a small though residual risk is still present at noise exposures of 85 dB(A) and the fact that there have been proposals to further reduce the action levels, it is clear that the airline should ensure that noise exposures will not exceed 85 dB(A), and would be advised to reduce noise exposures to nearer to 80 dB(A) wherever reasonably practicable.

If a flight crew member could show that he or she has suffered noise induced hearing loss as a direct result of exposure to an occupational hazard, the employer could be liable to pay substantial damages. A number of flight crew have retired prematurely from the airline as a result of tinnitus. So far, none have been able to establish to the satisfaction of a court that the tinnitus is a result of NIHL arising from occupational exposure as a commercial flight crew member. Nonetheless, the airline has a duty to minimise the risk.

Table 2, above, gives the noise levels averaged over each flight. The noise exposures experienced by the crew will depend on these levels and on the duration of the flight and also on any other noise to which they are exposed. When carrying out external inspection of the aircraft prior to flight, hearing protection should be worn.

Concentrating on noise exposures during flight, for a total flight time of 8 hours in a day, the noise exposure will be numerically equal to the average noise level. If the total flight time is less, the noise exposure will be numerically less than the average noise level, but if the flight time is more than eight hours the noise exposure will be numerically more than the average noise level. Table 3 gives a conversion from noise level to noise exposure based on the exposure time.

The value of N can be calculated more precisely from

N = 10 Log₁₀ (Duration in hours/8 hours).

Examples:

If the average noise level over an 8 hour flight is 80 dB(A), and there is no other major source of noise during the working day, the noise exposure will be 80 dB(A).

If the average noise level over a 4hour flight is 80 dB(A), and the crew are engaged on quiet duties for the rest of their working day, the noise exposure will be 77 dB(A).

If the average noise level over a 10 hour flight is 85 dB(A), the noise exposure will be 86 dB(A).

Most of the long-haul flights in this study were of the order of 7 to 10 hours, so the noise exposures will be roughly numerically equal to the averaged noise level. The short-haul flights in this study were return flights and the length of time between boarding at Heathrow and disembarking on return was of the order of 5-6 hours, though not all in flight. In such cases the noise exposure would be obtained by subtracting 1 or 2 from the noise level value in dB(A).

Using the above approximation suggests that noise exposures for the day from the ambient noise on the flight deck would be between 70 and 80 dB(A). These levels are below the current action levels and do not in themselves pose a foreseeable risk of long-term hearing loss. However, the noise levels approaching 80 dB(A) are sufficiently high that they might be expected to give rise to non-auditory effects such as increasing fatigue. The effect of noise on performance is a complex subject, but the effect of noise in the workplace may be seen not in terms of reduced performance but as a reduced ability to react to additional demands and increased fatigue after completion of tasks [27]. It would be desirable to reduce the level of noise from the flight deck reaching the ear, at least in the noisier aircraft, although flight crew would need to adapt to the change in subliminal auditory cues which play a part in spatial perception and situational awareness.

It was accepted by the airline management that the noise exposures experienced from the headsets during most flights needed to be reduced, but without compromising the intelligibility of radio and interphone communications.

The standard HME/HMD 410 headset provided negligible noise attenuation and crew perceived that there was no benefit in covering both ears. With the ANR headsets and the Racal Astrolite and Peltor headsets the noise reduction at the ear is more obvious and crew may perceive a worthwhile noise reduction by covering both ears. Headsets with a worthwhile noise reduction are therefore more likely to be worn binaurally. The effects of reduction in headset output attributable to wearing a headset over both ears and the reduction attributable to a higher noise attenuation are therefore difficult to distinguish.

Whilst the ambient noise on the flight deck is not in itself likely to cause damage, this noise does determine the minimum output level from the headset for satisfactory communication above the noise. For the sound level of the signal transmitted at the headset to be intelligible to the wearer, it has to be of sufficient energy to provide an adequate signal-to-noise ratio when compared with the ambient background noise. Personal preference then comes into play as the crew member sets the volume control to give a headset output above the minimum.

The ambient noise also, through the process known as ‘partial masking’, causes the speech from the headset to seem quieter than the same speech at the same volume control setting but without the background noise.

It is desirable to reduce the level of flight deck noise reaching the ear in its own right, but also necessary to reduce it to allow headset output to be reduced. Crew thus required encouragement to reduce their headset volume settings to take advantage of the reduced noise.

As well as showing the highest level of attenuation at the ear of the headsets studied, ANR headsets proved subjectively acceptable to the flight crew. ANR is acknowledged to be an expensive option, but acceptability and hence compliance are important components of the cost/benefit analysis.

Because the high output levels from headsets are likely to be distributed throughout the fleet and not confined to the noisier aircraft, and because of the need to reduce exposures as far as is reasonably practicable, ANR headsets were introduced on all aircraft types where the HME/HMD 410 headset was currently used.

While it may be possible, through the use of ANR headsets, to reduce the levels of flight deck noise reaching the ear by 10 dB or more, the crew are unlikely to reduce the outputs from their headsets to the same degree, and as a result, their individual noise exposures are unlikely to be reduced to the same extent.

To gain an indication of the likely reduction in noise exposure in practice, samples of continuous speech from each of the headsets tested in the Boeing 747-436 were re-analysed. Inevitably the samples varied in quality throughout the flight and included VHF from UK Air Traffic Control at the start, HF from Shanwick and Gander Radio and from other aircraft during the middle of the flight, and VHF from Boston Centre and the US Air Traffic Controllers at the end, with a mixture of male and female voices. These variables in communication were not balanced over the different headsets - it would have been impracticable to keep changing headsets in flight - so HF with a male voice through one headset may have to be compared with VHF with a female voice through another headset. At one stage a ‘Mayday’ distress message was overheard and the recorded headset output increased in level as the captain turned up the volume. There is no single ‘best’ volume setting; the volume chosen will depend on circumstances, on message content and the relevance and interest to the crew and consequently the volume will be varied during a flight even if the background noise is steady.

Nevertheless the headset outputs resulting from the captain’s chosen volume control settings were measured for each headset using a 16-second sample or shorter samples to make up a 16-second average of continuous speech. The samples were taken where possible from the later part of the time period during which each headset was worn, so the captain had time to acclimatise to each headset. The measured headset speech outputs are given in Table 5 together with the background flight deck noise measured at Kemar’s ear.

The headset outputs are plotted against background levels under the headset in Figure 19.

Bearing in mind the variety of speech sources heard, the outputs from the headsets chosen by the captain were remarkably consistent. Whether this captain’s controlling of headset output was typical is impossible to determine from the data at hand. However this behaviour, a partial rather than a full compensation for a change in noise level, is plausible, and there are other similar circumstances where people only half compensate for a change in noise level. For example, talkers conversing in noise, when not otherwise instructed tend to raise their vocal effort by about 5 dB for each 10 dB increase in noise over quite a wide range of moderate noise levels, the ‘Lombard effect’ [60].

Lombard noted in 1911 that a speaker changes his voice level similarly when the ambient noise level increases, on the one hand, and when the level at which he hears his own voice (his sidetone) decreases, on the other. That is, listening and speaking are quite different in their sensory dynamics and are quite separate temporally. Lane and Tranel showed [60] that the speaker tries to maintain a speech-to-noise ratio favourable for communication. They also showed that perturbations in the timing and spectrum of sidetone also lead the speaker to compensate for the apparent deterioration in his intelligibility, although in fact the speaker has no need to listen to himself when speaking.

The best indication at present is that a reduction in noise level from the flight deck reaching the ear of 10 dB will result in a crew member reducing headset volumes by about 5 dB, and pro rata for other noise reductions, as expected from Lombard [60].

But another important factor which may lead to a choice of lower headset outputs is the wearing of the headset on both ears rather than one ear. The indications from the tests in the Boeing 747-436 are that noise levels including communications as low as 77 dB(A) averaged over a flight can be achieved under a headset with ANR worn binaurally in a relatively noisy flight deck, as shown in Table 2.

Data in Table 4 derived from observation of a single subject when several different headsets were tried by the captain and also worn by the Kemar manikin, show that the best noise reduction at the ear was given by the Bose and Sennheiser ANR headsets.

Table 4
Reduction in noise from the flight deck of each of the headsets tested in flight in the Boeing 747-436

The combined effect of supplying ANR headsets and requiring headsets to be worn over both ears during flight has allowed flight crew to reduce their headset outputs by 10 dB or more. The reduction in ambient noise levels at the ear will encourage crew members to set lower volumes as the speech from the headphones will seem louder. But lower volumes are not guaranteed. The large individual differences in the headset outputs set by the different crew members show that the volume chosen by individuals is unpredictable. Some of the higher headset outputs chosen by pilots were encountered on the quieter aircraft, and vice versa. Crew may also adjust their volume controls according to the importance of or interest in the messages they hear and if there are other distractions on the flight deck.

An communication programme was implemented to advise flight crew about the effects of noise on hearing and give guidance that their headsets should be at the minimum volume acceptable for satisfactory communications. They were instructed that, when using an ANR headset, it should be worn on both ears so that excessive volume is not required. However, on some aircraft types it proved technically difficult to use intercom rather than open speech during certain phases of flight such as take-off. In these cases, flight crew were instructed to leave one ear uncovered during this phase of flight, but to wear the ANR headset binaurally at all other times.

The ANR headset reduces the level of warning sounds as well as the background noise on the flight deck, but because it reduces both by the same amount, the signal-to-noise ratio is not degraded and the detectability of the warning should not be impaired. Similarly the speech of visitors to the flight deck will be reduced in level by the headset to the same degree as the background noise; so if for example, a member of the cabin crew visits the flight deck he or she will be as audible and intelligible as before. However, people automatically speak more quietly if the ambient noise is reduced. For a person wearing an ANR headset the noise appears to have been reduced and the person will lower his or her voice. For this reason the flight crew will need to use the interphone, and the level of their sidetone will need to be set low enough to encourage them to speak normally when using the interphone or radio.

Flight crew were advised of the reduction in auditory cues which subliminally contribute to perceptual and situational awareness. Subjects participating in this study quickly adapted to the change in auditory information and made more use of aircraft and engine instrumentation. There was unanimous agreement that this did not present any operational or flight safety hazard, and has been the case following introduction of the ANR headsets.

A follow-up study is planned to begin in April 2001, two years after the introduction of the headsets, when noise and communication levels on selected flight decks will be measured in a similar way.

Ambient noise levels on the flight decks of the airline’s aircraft were found to be between 70 dB(A) and 79 dB(A) averaged over the flight. These levels in themselves would pose no foreseeable risk of hearing damage.

Because there are large individual differences in the preferred headset volumes among the flight crew population, the high noise exposures may occur on any of the aircraft types, not just those with the noisier flight decks. Accordingly, headset volumes needed to be reduced throughout the fleet.

The better examples of noise reducing headsets, worn to cover both ears, would allow flight crew to reduce their headset volumes. To obtain full advantage of these headsets crews needed to be informed and advised on the best use, including wearing headsets to cover both ears and to use lower headset volumes whenever possible.

Reducing noise level at the ear may have additional beneficial effects such as reducing fatigue after long flights. Because high listening volumes may occur on any type of aircraft, not just the noisier ones, noise reducing headsets needed to be introduced in all aircraft types where the HME 410 or HMD 410 headset is currently used.

Although passive attenuating headsets would have provided adequate protection, they were subjectively not acceptable in the ‘shirt-sleeved’ environment of the long-haul flight deck, and it was necessary to introduce ANR headsets.

I am pleased to acknowledge the expert advice and technical assistance given by Dr Michael C Lower of the Institute of Sound and Vibration Research, University of Southampton.

I also acknowledge the co-operation of the avionics engineers, flight technical managers and the participating flight crew without whom this study could not have been performed.

The data in this paper are extracted from a dissertation submitted for membership of the Faculty of Occupational Medicine. Figures have been imported from the dissertation and not renumbered. The reference list has been retained for the sake of completeness.