Asymmetric sensorineural hearing thresholds in the non-noise-exposed UK population: a retrospective analysis

Transcription

1 ORIGINAL ARTICLE Asymmetric sensorineural hearing thresholds in the non-noise-exposed UK population: a retrospective analysis Lutman, M.E.* & Coles, R.R.A. *Institute of Sound and Vibration Research, University of Southampton, Southampton, UK, and MRC Institute of Hearing Research, Nottingham, UK Accepted for publication 4 May 2009 Clin. Otolaryngol. 2009, 34, Objectives: To estimate the distribution of inter-aural sensorineural hearing threshold level differences in the non-noise-exposed adult population of the UK. Setting: A two-stage population study carried out in , initially by postal questionnaire, followed up in a proportion of participants by clinical and audiological examination. Participants: Volunteers (n = ) initially selected at random from the electoral registers of four cities, subsequently selected at random from questionnaire respondents stratified by answers to questions about hearing. Main outcomes measure: Inter-aural hearing threshold level differences measured audiometrically, as a function of age and gender. Results: Tables of inter-aural threshold level differences provided as a resource with potential medicolegal, clinical and research applications. Based on the average of the frequencies 0.5, 1, 2 and 4 khz, 1% of the general UK population aged years have an asymmetry of 15 db or more. The prevalence is greater in older than in younger people. Conclusions: Inter-aural threshold differences greater than attributable to measurement error are not uncommon in the adult population, even after screening for conductive hearing loss and substantial noise exposure. They are typically of unknown origin. Asymmetries in the audiogram can lead to diagnostic and procedural uncertainties in several areas of otological and audiological practice. Of course, the matter is usually evident when there is a middle-ear disorder and conductive hearing loss. And sometimes in cases of asymmetric sensorineural hearing loss there is a likely cause: for example, a severe head injury or a nearby explosion or gun blast can result in a hearing loss greater on one side than the other. But where there is no apparent cause, further investigation may be indicated to check on the possibility of a more serious condition, such as a vestibular schwannoma. That possibility underlies the American and British procedural guidelines in hearing aid fitting for deciding when patients should be referred for otological opinion. 1,2 This raises the question of how much asymmetry should be regarded as abnormal. Small asymmetries are Correspondence: M.E. Lutman, Hearing and Balance Centre, Institute of Sound and Vibration Research, University of Southampton, Highfield, Southampton S017 1BJ, UK. Tel.: ; fax: ; mel@isvr.soton.ac.uk à Now retired. bound to occur simply as a result of the imprecision of audiometry, where test retest differences of 5 or 10 db are common. According to Lutman et al. 3, a threshold shift at 4 khz measured with an audiometer using 5 db steps must be at least 15 db to be treated as significant with a probability of error below 5%. On top of measurement variability are probable innate biological differences between the two ears. Consequently, to identify what may be regarded as a pathological difference, it is helpful to have some idea of what is the range of inter-ear differences found in single conventional audiometric measurements, after screening out obvious causes. Similar considerations affect diagnostic assessment in medicolegal work in relation to noise exposure, particularly where noise-induced hearing loss (NIHL) may possibly be complicated by gunshot noise effects. It is well known that those who fire weapons from the right shoulder tend to have greater high-frequency hearing losses in the left ear. One reason for this is that the noise (mainly coming from the muzzle) in such a firing position is a few decibels greater at the left ear than at the right. But the typical left worse than right effect on hearing is far from universally found: often there is little or 316

2 Distribution of sensorineural hearing asymmetry 317 no difference between the ears, and sometimes the right ear is worse than the left. 4 Complicating factors include the greater noise that sometimes comes from a neighbouring firer on one or other side, the effects of firing in reverberant enclosures or near to a reflecting wall and additional exposures to other sorts of weapon with an uncertain asymmetry of noise field. Yet another possible, but probably infrequent, factor is a material difference in noise susceptibility between right and left ears of the same individual. Assessment of individual cases needs to take account of not only the many uncertainties of noise exposure and its effects, but also the unexplained degrees of asymmetry to be found in non-exposed persons. When gunshot noise effects are a possibility, dogmatic statements linking them to an asymmetry of hearing loss are seldom justifiable. The purpose of this paper is to provide the reference data that are needed to enable balanced and informed judgements to be made as to the possible and most likely causation of an asymmetrical hearing loss. Materials and methods The population study analysed The data analysed are from the UK National Study of Hearing, a large-scale multicentre population study carried out in several phases between 1979 and A description of the study as a whole, together with a mass of statistical data arising from it, is given in book form by Davis. 5 In the main part of the study, questionnaires were sent to people randomly sampled from the electoral registers of four cities: Cardiff, Glasgow, Nottingham and Southampton. From the responses to those questionnaires, stratified samples were invited to attend research clinics for in-depth interview, clinical examination and audiological measurements. The number tested in the age range years was Of these, 2679 gave acceptable audiograms and constituted the database for Davis detailed analysis, and for the particular further analysis reported here. The screening criteria To identify the prevalence of asymmetries in the general population, it is important to exclude from analysis those participants who have characteristics that are both quite common and also quite likely to be associated with audiometric asymmetry. On the other hand, it is desirable to apply as few screening criteria as possible to retain a sizeable number of participants for analysis. The most common cause of hearing loss in the population is the inherent degeneration of structure and function that occurs with ageing, but not just in the elderly. This is combined with a multitude of other factors that may affect hearing and which accumulate with advancing years. Combined, they are best described as age-associated hearing loss (AAHL). However, for the purposes already discussed, AAHL has to be regarded as within a normal population. Nevertheless, the prevalence of asymmetric AAHL needs to be examined as a function of age group, as it transpired that asymmetry was found to be age-dependent. The two main exclusion factors used here to obtain a screened reference group suitable for the purpose of this study are conductive hearing losses and excessive noise exposure. Conductive hearing loss is often asymmetric, but is readily identifiable by clinical and audiological means. It is additional to the unexplained asymmetric sensorineural hearing impairments that this study aims to explore. In the present analysis, any participant with an air bone gap averaged over 0.5, 1 and 2 khz exceeding 10 db in either ear was excluded. A substantial problem with applying an exclusion criterion for noise exposure is to define what should be regarded as excessive and having a substantial potential for causing hearing loss. Even with symmetrical industrial noise exposures, the NIHL sometimes appears to be considerably greater in one ear than the other for no apparent reason. With noise from guns or other explosive sources, asymmetry of exposure and its effects is quite common. The question of how much noise exposure to exclude has been considered in some detail by Lutman and Davis. 6 They studied the 406 persons in the years age group in the National Study of Hearing and applied various levels of screening for past noise exposure, assessed by detailed questioning according to an elaborate protocol. 5,6 They found that their moderately screened group had virtually the same hearing as a highly screened group, and concluded that the former was effectively uncontaminated by noise exposure. The moderate noise screening criteria have therefore been used for the present analysis. The occupational noise exposure criterion used was based on noise immission level (NIL), as defined by Burns and Robinson. 7 The exclusion criterion was an NIL of 97 db(a) or more; NIL of 97 db(a) has the acoustic energy equal to unprotected workday noise exposure of 97 db(a) for 1 year, which is equivalent to 94 db(a) for 2 years, 90 db(a) for 5 years or 80 db(a) for 50 years, the maximum likely duration of employment. That exclusion criterion was also applied to leisure noise exposure. For gunfire noise, the criterion was exposure to more than 100 rounds of rifle or equivalent without ear protection, or to

3 318 M.E. Lutman & R.R.A. Coles an explosion causing a permanent or very marked temporary effect on hearing. Heavy gun rounds were considered to be equivalent to 10 rifle rounds for this purpose. After application of the above conductive hearing loss and noise exposure exclusion criteria, there remained 1231 participants for analysis, 368 male and 863 female. The data from each sampling stratum were weighted according to the prevalence of the stratum in the general population, as estimated from the original questionnaire stage of the study. In this way, each subject was allocated a fractional case weight. The consequence is that statistics calculated from the weighted data (e.g. means, distributions) reflect the characteristics of the general UK adult population aged years, after applying the exclusion criteria on air bone gap and noise exposure. Because these criteria excluded more people in the older age groups, the weighted sample sizes decreased with increasing age to a much greater extent than survival rates would suggest. The weighted sample number totalled Ethical considerations This was a Medical Research Council study, approved inhouse and also by the hospital ethical committees and the Community Health Councils in the districts in which the main study was carried out, following the principles of the Declaration of Helsinki. Results Initial analysis showed little influence on asymmetry of age, gender or test frequency. Therefore, data for all participants were pooled across audiometric test frequency (0.25, 0.5, 1, 2, 3, 4, 6 and 8 khz) to provide an indication of the distribution of left right differences and their overall statistics. The distribution is shown in Fig. 1. The mean value of the left-minus-right threshold differences was +0.4 db, with a standard deviation of 8.9 db. In view of the near-zero mean value, for the remainder of the analysis, asymmetries are presented as unsigned data, simply indicating the absolute magnitude of the asymmetry, irrespective of whether left was greater than right or vice versa. There was, however, considerably more asymmetry with increasing age and audiometric frequency. There was also a tendency for more asymmetry in males than females. The results of further analyses are therefore shown as a function of age and audiometric frequency, separately for males and females. The results of the main analysis are shown in Table 1. In addition to the data for individual audiometric test frequencies, they are shown for four particular frequency Fig. 1. Distribution of differences in hearing thresholds at each frequency (left minus right), pooled across 0.25, 0.5, 1, 2, 3, 4, 6 and 8 khz for all participants and weighted to reflect the UK adult population. The best fit normal distribution curve is also shown: mean difference +0.4 db, standard deviation 8.9 db. Note that negative numbers indicate right ear worse than left ear. averages. s over 1, 2 and 3 khz and over 3, 4 and 6 khz are included because of their relevance in assessment of noise-exposed persons. The average over 0.5, 1, 2 and 4 khz, the four-frequency average, is included as it is often regarded as the best single number representation of the hearing status of an ear. The average across 3, 4, 6 and 8 khz is also included as it gives a good general quantification of the state of hearing at high frequencies, the frequency range most prone to degenerative and pathological effects. To provide a more easily digestible summary, the data for all age groups together are given in Table 2. It should be noted that the numbers in the younger age groups considerably exceed those in the older groups. Therefore the combined figures are influenced strongly by the interaural threshold differences observed in young adults, which are smaller than in the older adult population. On the other hand, they have an advantage in being derived from a considerably larger number of subjects than in the separate age sub-groups. Discussion Strengths and limitations of the study The main strength of the study is its foundation in a whole population survey, so that the prevalence estimates reflect the general population rather than clinic samples that are inevitably influenced by referral patterns. The

4 Distribution of sensorineural hearing asymmetry 319 Table 1. Percentiles for absolute inter-aural sensorineural hearing threshold level differences in db (male female) in adults not exposed to noise. Numbers are weighted to reflect the adult UK population Age group, percentile Audiometric frequency (khz) , 2, 3 3, 4, 6 0.5, 1, 2, 4 3, 4, 6, years (n = ) 50% % % % % years (n = ) 50% % % % % years (n =62 139) 50% % % % % years (n =46 166) 50% % % % % years (n =37 123) 50% % % % % years (n =36 93) 50% % % % % sample tested, even after exclusion of conductive hearing loss and history of material noise exposure, was large compared with most clinical studies. However, the numbers of cases in the sample with asymmetrical hearing threshold levels remained small due to the low prevalence of asymmetrical hearing in the general population. Therefore, the prevalence estimates are inevitably surrounded by some uncertainty, especially when subdivided by gender and age band. Nonetheless, while the exact prevalence may be uncertain, the general finding of low prevalence is robust. More accurate estimation of prevalence of asymmetry in the general population would entail an enormous study and is probably unnecessary for most purposes. Knowledge that only about 1% of the nonnoise-exposed adult population have asymmetry of 15 db or more, based on the average of the frequencies 0.5, 1, 2 and 4 khz (see Table 2), is sufficient to guide clinical diagnosis and screening policy. Comparison with previous studies Another study that includes data on audiometric asymmetry is that by Robinson. 8 He examined a carefully screened sample of people with industrial NIHL, having presumed

5 320 M.E. Lutman & R.R.A. Coles Table 2. Percentiles for absolute inter-aural sensorineural hearing threshold level differences in db (male female) in adults not exposed to noise for all age groups combined. Numbers are weighted to reflect the adult UK population Percentile Audiometric frequency (khz) , 2, 3 3, 4, 6 0.5, 1, 2, 4 3, 4, 6, years (n = ) 50% % % % % % symmetrical noise exposure. His purpose was to provide statistical tests to uncover causation other than noise. The first of his eight tests concerned the size of the left right hearing threshold level differences. In his sample of 63 cases of NIHL, the overall standard deviation of the interaural differences, which were measured at 0.5, 1, 2, 3, 4 and 6 khz, was 5.6 db. That is considerably smaller than the corresponding value of 8.9 db we found in our effectively non-exposed sample. That in turn suggests that the considerable inter-aural NIHL differences shown in Robinson s Table 2 are likely to be due in many cases to nothing more than noise damage superimposed on the sort of asymmetries to be found in the non-exposed population. There appears to be a continuum in degree of asymmetry in NIHL cases, which runs from minor asymmetry that would usually be recorded as normal, or attributed to measurement uncertainty, to large enough to exceed Robinson s 95th or 98th percentile criterion for likelihood of alternative causation. In the light of this study results, such statistical criteria for asymmetry seem to have rather little value for medicolegal diagnosis, where the purpose is to separate a normal from a pathological extent of asymmetry. Potential applications for the prevalence estimates It is envisaged that the prevalence estimates of asymmetry of hearing in the non-noise-exposed population may be helpful in medicolegal work where hearing asymmetry needs to be evaluated in terms of the relative probability of extrinsic or constitutional pathological causation. Is the asymmetry likely to be due to some asymmetrical noise exposure or some asymmetric pathology? Or is it more likely to be just something to be found in a proportion of non-exposed people of that age group and gender? Detailed inspection of the data, particularly in Table 1, indicates that some sizeable inter-aural differences were encountered. Large differences can be identified in the males even in the youngest age group: 55 db at 4 khz, and 41 db in the average across 3, 4 and 6 khz; also 77 db inter-aural difference at 4 khz in the years group. Why were there such cases in this screened population? Was there something in an individual s history that could be a likely contender for the cause of the asymmetry? In those with the largest asymmetry, detailed re-examination of the original data failed to reveal a Asymmetry criterion Percentage exceeding criterion by age group (years) US * Male Female UK Male Female Table 3. Estimated percentages of nonnoise-exposed UK population sample exceeding asymmetry criteria for further medical investigation of hearing aid candidates *Inter-aural air-conduction hearing threshold level difference of 15 db or more averaged across the frequencies 0.5, 1, 2 and 3 khz. Inter-aural air-conduction hearing threshold level difference of 20 db or more at any of the frequencies 0.5, 1, 2 or 4 khz.

6 Distribution of sensorineural hearing asymmetry 321 potential cause of the asymmetry. Information examined included family history of deafness, head injury, ear injury, administration of ototoxic medication, general health factors including hypertension and history of infectious diseases. We conclude from this, based on our representative screened sample, that many hearing asymmetries in the general population can be expected to be of unknown causation. Another important consideration of audiometric asymmetry occurs when a person consults an audiologist with a view to obtaining a hearing aid. Those with asymmetric sensorineural hearing loss should be referred for a medical opinion. The present data have therefore been analysed in terms of asymmetry criteria used in the Unites States (US) and in the UK for such medical referral. The results are shown in Table 3. Note the substantial proportion of cases that would be referred using the UK criterion, where referral is triggered by asymmetry of 20 db at any of the frequencies 0.5, 1, 2 or 4 khz. A much smaller proportion requires medical referral using the US criterion, which is based on asymmetry of 15 db in the average of the frequencies 0.5, 1, 2 and 3 khz. In fact, so few cases in the present sample meet the US criterion that the estimates in Table 3 are unreliable. The difference in consequence between the UK and US asymmetry referral rules is remarkable, considering that the purpose of those rules is essentially the same: to detect auditory nerve disorders of clinical importance. The ideal criterion may lie somewhere between the two rules, one being too stringent and the other too lax. The study by Obholzer et al. 9 is relevant to this issue. For detection of vestibular schwannoma, they found the following sensitivity and specificity values: 97% and 37% by the UK rules and 92% and 62% by the US rules. Changing to a criterion difference of 15 db or more at two adjacent octave frequencies retained the 97% sensitivity of the UK rules, but improved the specificity to 47%. Ana1ysis of the year olds in this study showed that 7.4% of males and 5.4% of females would exceed the Obholzer 15 db rule: values not surprisingly lying between the UK and US values shown in Table 3. On the basis of the Obholzer study, further supported by this study, it would seem that both the British and American asymmetry referral rules might usefully be revised by making the former less strict, and the latter more strict. However, the Obholzer et al. study was based on only 36 cases of vestibular schwannoma and 100 nontumour MRI cases. Moreover, the revised referral criteria were adjusted after analysis of the data and the sensitivity and specificity estimates may therefore not apply to fresh samples. There would seem to be a need for a further study using a much larger series of tumour cases and also a larger number of negative MRI cases, perhaps relating the results also to the National Study of Hearing data reported in the present paper. Conflict of interest None to declare. References 1 American Academy of Otolaryngology Head and Neck Surgery (1994) Academy responses to the FDA request for comments on hearing aid regulations. AAO HNS Bull. 13, Reeves D., Mason L., Prosser H. et al. (1994) Direct Referral Systems for Hearing Aid Provision. HMSO, London 3 Lutman M.E., Cane M.A. & Smith P.A. (1989) Comparison of manual and computer-controlled self-recorded audiometric methods for serial monitoring of hearing. Br. J. Audiol. 23, Cox H.J. & Ford G.R. (1995) Hearing loss associated with weapons noise exposure: when to investigate an asymmetrical loss. J. Laryngol. Otol. 109, Davis A.C. (1995) Hearing in Adults. The Prevalence and Distribution of Hearing Impairment and Reported Hearing Disability in the MRC Institute of Hearing Research s National Study of Hearing. Whurr, London 6 Lutman M.E. & Davis A.C. (1994) The distribution of hearing threshold levels in the general population aged 18 to 30 years. Audiology 33, Burns W. & Robinson D.W. (1970) Hearing and Noise in Industry. HMSO, London 8 Robinson D.W. (1985) The audiogram in hearing loss due to noise: a probability test to uncover other causation. Ann. Occup. Hyg. 29, Obholzer R.J., Rea P.A. & Harcourt J.P. (2004) Magnetic resonance imaging screening for vestibular schwannoma: analysis of published protocols. J. Laryngol. Otol. 118,