Feasibility of community-based hearing screening using transient evoked otoacoustic emissions

Public Health (1998) 112, 147±152 ß R.I.P.H.H. 1998 http://www.stockton-press.co.uk/ph

Feasibility of community-based hearing screening using transient evoked otoacoustic emissions B McPherson1, J Kei1, V Smyth1, S Latham2 and J Loscher 2 1 2

Department of Speech Pathology and Audiology, The University of Queensland, Brisbane, Australia; and Community Child Health Service, Queensland Health (Brisbane North Region), Australia This study investigated the feasibility of obtaining transient evoked otoacoustic emissions for hearing screening purposes from infants and children at seven child health clinics. Factors affecting the outcomes of the community hearing screening program were examined. The subject group comprised 2305 children aged two weeks to 11 y 11 mon. Many children were attending the clinics for initial immunisation at two months of age. While there were no exclusion criteria for the 1305 young infants tested who were in this category, all other children were screened only upon receipt of a referral from clinic nurses. Results indicated that 182 children (7.9%) did not complete the screening for both ears within the time constraint (usually 15 min) of a child health clinic visit. Three hundred and sixty-two children (15.7%) failed the ®rst screening. Of the 226 children who returned for a second screening test, separated from the ®rst by at least two weeks, 121 children failed. With parents' consent, 107 children (4.6% of all participants) were referred for diagnostic or medical assessment. Subsequently, 77 out of 94 children who received audiological or medical assessment were found to have some degree of hearing impairment. The majority of positive screening results were associated with middle ear disorder. The results suggest that TEOAE screening has potential as a technique in the community health setting but improvements in instrumentation are required to reduce `could not test' cases and to separate probable conductive hearing loss from cases likely to have other disorders. Keywords: children; hearing; screening; transient otoacoustic emissions

Introduction There is a growing public health policy emphasis in many countries on diagnosing hearing impairment as early as possible in childhood. This is to allow medical and nonmedical intervention to commence before the development of possible adverse speech and language sequelae, and subsequent educational, psychological and social disadvantage. To achieve this goal screening all newborns for hearing loss prior to discharge from hospital appears an attractive option. In recent years a two-tier strategy using transient evoked otoacoustic emission (TEOAE) techniques followed by auditory brainstem response audiometry (ABR) has been advocated for this population.1 TEOAEs are low intensity sounds emitted by the normal cochlea in response to brief acoustic stimuli.2 Children with more than a mild hearing loss due to middle ear or cochlea disorder do not show a response for the affected frequency range.3 However it has been reported that TEOAE screening in this context has a high false positive rate.4 Noisy hospital environments,5 debris and ¯uid in neonatal ears6 and low TEOAE levels relative to noise during the ®rst 24± 48 h post partum7 contribute to this ®nding. In addition, not all neonates are readily available for screening: as well as nonhospital births, some neonates are discharged within 24 h of birth while others are withheld from screening for medical management reasons or because parental consent is not available.8 An alternative to a universal neonatal hearing screening program may be to later test infants when they visit community child health clinics or family doctors for routine immunisation and developmental assessments. If screening is Correspondence: Dr B McPherson, Department of Speech and Hearing Sciences, University of Hong Kong, 5F Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong. Accepted 22 December 1997

carried out in the community during early infancy, before many mothers return to their workplace, favourable attendance and compliance rates may be achieved. To date, there have been no reports investigating the feasibility of obtaining TEOAEs from infants in a community setting or that document the outcomes of a community based program. The aims of this study were to: (1) investigate the feasibility of obtaining TEOAEs in children at community child health clinics; and (2) examine the factors that may affect the outcomes of a community hearing screening program. Method Subjects Infants and young children attending eight community health clinics in the northern Brisbane region were enrolled in the study. The clinics were state government managed and offered a wide variety of health services for children and their mothers, including immunisation. The primary target group in this study were infants attending for their initial immunisation for diphtheria-tetanus-pertussis, poliomyelitis, and Haemophilus in¯uenzae type b at approximately two months of age. Participation was solicited through waiting room advertising and direct approaches to parents. However, younger or older attending children who were suspected of having hearing problems were referred by clinic nurses and also screened by the research group. Informed consent to conduct hearing screening measures was obtained from the parent or guardian of each child involved in the study. Procedure Testing was carried out by an audiologist experienced in the use of TEOAE procedures or a community health nurse

Community-based hearing screening B McPherson et al

148

who had undertaken speci®c training in TEOAE procedures. Most infants were tested before immunisation unless they were judged to be unsettled or crying at the time of arrival and could be tested more readily after postimmunisation feeding. Testing was performed in the quietest room available at each clinic. Only one testing location was sound-treated, however, and ambient noise levels ranged from 38± 48 dB A. TEOAE responses were recorded with an IL088 Otodynamics analyser using the QuickScreen v.3.94H program. The QuickScreen analysis was chosen because it is more robust in non-soundtreated environments,9 although it limits data collection to frequencies above 1 kHz.2 A previous pilot study using the default IL088 program had indicated that TEOAE recordings could not be obtained consistently in infants in the community clinic setting. For most subjects, a brief otoscopic examination was followed by the insertion of a TEOAE probe and disposable probe tip to the ear most readily accessible to the tester. If otoscopic examination could possibly disturb a sleeping child it was deferred until after TEOAE testing. Wide band clicks were applied to the test ear with an average stimulus level of 78 dB peak SPL (s.d. ˆ 4 dB). The probe ®t was checked before TEOAE recording commenced. The noise rejection level was set as low as possible to allow data to rapidly accumulate, with the level during screening dependent on clinical judgment. A pass was recorded in this study when a TEOAE spectrum was at least 3 dB above the noise ¯oor and at least halfway across the test frequency bands of 2±3 kHz and 3 ±4 kHz. A fail was indicated otherwise. Pass and fail were recorded for each ear tested. The mid to high frequency bands were selected to: (a) avoid detection of exclusively low frequency hearing loss associated with temporary otitis media with effusion (OME) or eustachian tube blockage10; and (b) minimise the in¯uence of low frequency background noise on the TEOAE results. The screening protocol required collection of at least 50 low noise TEOAE signals from each subject's ear. At that stage, if the TEOAE result indicated fail, sampling was continued until 125 low noise samples were collected, then until 260 if a pass result was still not achieved. When the ®nal TEOAE result indicated a fail the probe was re®tted and testing repeated, if time allowed. Those children who failed the ®rst test in either ear were requested to attend a second TEOAE test Table 1 Age range (in months)

0.0 to < 0.5 0.5 to < 1.5 1.5 to < 2.5 2.5 to < 3.5 3.5 to < 4.5 4.5 to < 5.5 5.5 to < 6.5 6.5 to < 18.0 18.0 to < 30.0 30.0 to < 42.0 42.0 to < 54.0 54.0 to < 66.0 66.0 to 143

approximately two weeks later. Having failed two TEOAE tests (or rarely, three tests if the infant was very active or had an obvious cold at the second appointment), children were referred for diagnostic auditory brainstem response audiometry (ABR) or for behavioural audiometry, followed by further audiological assessment and medical referral if required. ABR assessment was usually conducted within two weeks of the second TEOAE failure, typically at a university clinic associated with the project. Alternately, behavioural audiometry was conducted within approximately one month of the second TEOAE failure at an Australian Hearing Services or hospital clinic. Children for whom it was not possible to obtain TEOAE recordings in one or both ears were classi®ed as `could not test' cases and excluded from further study. TEOAE equipment was checked for calibration using the manufacturer's speci®cations before the study and at three month intervals during the course of the study. No drift in calibration occurred during the course of the study. Results A total of 1305 infants in the target age range (1.5 to < 2.5 mon) were tested between May 1994 and November 1996. Another 193 infants aged less than 1.5 mon and 807 infants and children aged 2.5 mon and over were also assessed in the same period. A total of 2305 children participated in the study (Table 1). For 92.1% of the children it was possible to obtain complete TEOAE screening results for both ears in the community health setting. However, for 7.9% of children it was not possible to obtain a valid TEOAE recording in one or both ears, within the maximum 15 min period allocated for screening. The percentage of children who could not be screened varied with age. In the immunisation attendance group (1.5 to < 2.5 mon) 6.7% of infants could not be screened whereas for children 30 mon and over only 1.8% were in this category. The most dif®cult to screen age range was from 2.5 to < 30 mon, with average could not test ®ndings of 14.3% being recorded, range 9.0 ±25.0%. The activity status of young children during TEOAE screening was rated using ®ve behaviour categories. A progressive increase in the `restless' classi®cation can be observed with increasing age (Figure 1).

Screening outcomes for community child health clinic TEOAE screening

Mean age in months (s.d.)

0.45 1.18 2.00 2.89 4.03 4.98 6.08 10.40 24.48 36.46 47.50 59.78 88.45

(0.07) (0.25) (0.21) (0.26) (0.26) (0.28) (0.29) (3.22) (3.51) (3.52) (3.39) (2.84) (21.90)

Total children

2 191 1305 221 111 46 30 100 79 89 70 28 33 Total 2305

Children screened

2 185 1217 197 101 41 25 75 64 86 70 27 33 2123

Children not tested N

%

0 6 88 24 10 5 5 25 15 3 0 1 0 182

0.0 3.1 6.7 10.9 9.0 10.9 16.7 25.0 19.0 3.4 0.0 3.6 0.0

Community-based hearing screening B McPherson et al

Figure 1

Pattern of activity of young infants (0 ± 4.5 mon) at ®rst TEOAE screening.

TEOAE screening test time for those children with complete results also varied with age. TEOAE data collection time was recorded by the QuickScreen program and shows a general reduction in the time required for data collection after 18 mon of age (Figure 2). The mean overall test time for the initial screening was 117 s (standard deviation ˆ 125 s), with a median time of 75 s. The screening outcomes for the overall group and those for the immunisation infant group are summarised in Figures 3 and 4 respectively. For all children who could be tested using the TEOAE technique, 17.1% failed the

Figure 3

Outcomes of screening for total program group.

screening. For infants in the immunisation group the failure rate was 10.8%. At the second TEOAE screen 53.5% of all returning children continued to show fail results as did 57.3% of the immunisation group. Overall, 107 children were referred for further assessment, that is 5% of all children who had been successfully screened. A total of 82% of referrals who attended further assessment were found to have hearing loss in one or both ears. For the immunisation group there were 36 referrals, that is 3% of those successfully screened. Thereferraloutcomesintermsofestimatedaveragehearing loss in the better ear are presented for the overall subject group in Table 2. The study identi®ed one case of severe bilateral hearing loss and three cases of profound bilateral loss. The immunisation group contributed the case of severe bilateral loss and one of the cases of profound bilateral loss. Participation in each stage of the project was voluntary and an average drop-out rate of 12% was noted at the acceptance of referral and attendance at referral agency stages. The majority of hearing loss cases (Figure 5) were conductive and associated with otitis media with effusion (OME). Only 7.4% of cases were mixed or sensorineural. For 18.1% of all cases, and 29% of immunisation group cases, normal ®ndings in both ears were noted at the referral assessment. Discussion

Figure 2 TEOAE data collection time for different age groups.

The criteria for failure in this study were absence of TEOAEs in the mid to high frequency range. The great

149

Community-based hearing screening B McPherson et al

150

Figure 5 Type of hearing loss reported after diagnostic referral.

the appropriate decision in this situation. If a program is designed to detect any hearing loss affecting the child at the time of test then conductive pathology must be included. However, implicit in many programs is the detection of children with permanent hearing loss that is likely to lead to long-term disadvantage if prompt intervention is not initiated. It has been suggested that children detected with OME-related loss should be retested after a delay of 4± 12 weeks to con®rm the persistence of such a hearing loss.12 If the loss is persistent then referral is advised. However, in neonates and young infants this is not practical since (a) there is no method at present that will accurately separate OME-related from non-OME related cases, and (b) to delay referral of potentially permanent hearing loss cases by up to 12 weeks defeats, to some extent, the purpose of a screening program in early life. The present program attempted to minimise the numbers of such OME-related children but still found at least 70% of failures were related to conductive pathology. An accurate method of con®rming OME-related loss in neonates and young infants is urgently required. With such information screening programs could implement a two-track referral procedure for fail cases, arranging immediate referral for likely sensorineural hearing loss infants and review screening for likely temporary conductive hearing loss. This would reduce the overall referral burden on diagnostic audiological and medical facilities associated with such a screening program.

Figure 4 Outcomes of screening for immunisation group.

majority of sensorineural and permanent conductive hearing loss cases requiring hearing aid provision have de®cits in this range.11 Therefore children with greater than mild hearing loss in this critical range were targeted with this strategy. However, it is apparent that the great majority of children identi®ed with hearing loss were still those with conductiveÐand most likely temporaryÐhearing loss. Should such cases be targeted in infant hearing screening programs or not? There is controversy in the literature on

Table 2 Degree of hearing loss found in referred children Degree of hearing loss diagnosed in better ear Age range (in months)

Normal Mild loss Moderate Moderately severe Severe Profound Did not attend Total hearing (26 ± 40 dB HL) (41±55dB HL) (56 ± 70dB HL) (71±90dB HL) (  91dB HL) referral clinic referred

0.0 0.5 1.5 2.5 3.5 4.5 5.5 6.5 18.0 30.0 42.0 54.0 66.0

0 0 9 1 1 1 3 1 0 1 0 0 0 Total 17

to < 0.5 to < 1.5 to < 2.5 to < 3.5 to < 4.5 to < 5.5 to < 6.5 to < 18.0 to < 30.0 to < 42.0 to < 54.0 to < 66.0 to 143

0 0 8 2 3 2 1 2 1 1 0 1 1 22

0 0 10 2 1 0 0 8 9 7 7 2 13 49

0 0 2 0 0 0 0 0 0 0 0 0 0 2

0 0 1 0 0 0 0 0 0 0 0 0 0 1

0 0 1 0 0 0 1 1 0 0 0 0 0 3

0 1 5 1 0 0 0 2 0 0 1 1 2 13

0 1 36 6 5 3 5 14 10 9 8 4 6 107

Community-based hearing screening B McPherson et al

This study found that age had a marked effect on the activity status of infants during TEOAE screening. Findings from an earlier study of a subgroup of this cohort13 indicated that activity status has an impact on both test time and failure rate of infants. Similarly, for the overall study group, it was found that with increasing age and increasingly active behaviour the feasibility of rapid screening was reduced. Neonates and infants under 3 mon and infants over 30 mon of age showed the greatest proportion of successfully completed test results. Routine mass audiometry with school-age children produces many false positives,14 resulting in the considerable, but accepted, expense of further diagnostic assessments.15 False positives are usually associated with noisy test conditions, misunderstood test instructions or poor test technique. Objective TEOAE screening in infants eliminates test instruction problems and test quality can be reviewed by checking computer records, but noise is still a potential cause of high false positive rates. Future development of test instrumentation may result in faster data acquisition and greater tolerance of physiological noise. Fail criteria can also have a major effect on TEOAE screening outcomes16,17 and may need further re®nement. In addition, infant activity may result in considerable numbers of `could not test' cases in TEOAE screening. An effective screening technique must be able to rapidly test virtually all members of a target population. In the present study 7.9% of children overall, and 6.7% of the immunisation group, could not be screened on the ®rst attempt. This is a high ®gure and ways to reduce numbers of could not test cases need to be found prior to standard introduction of such a screening procedure. Two possible means are by age restriction in the screening protocol to ensure only infants within an optimal activity time window are tested and by `preparing' children for screening clinics by advising parents on ways to ensure children are more likely to be asleep or quiet at the time of testing (such as pre-screening feeding). In our earlier study13 it was found that the mean total consultation time for successfully tested infants in a quiet activity state was 11.7 min compared to 16.8 for children in noisy activity states. This is another reason for attempting to ensure infants are likely to be in an appropriate state at the test of screening. The additional 30% duration associated with noisy activity states would impact signi®cantly on the costs associated with a mass screening program. The TEOAE screening test is reported to be highly sensitive18 but further longitudinal monitoring of the participants is required to derive indications of test sensitivity and speci®city from the present study, as children who passed TEOAE screening were not given a diagnostic test. A limitation of the present study was that diagnostic testing was delayed for most children. The percentage of false positives who actually had transient OME at the time of screening is therefore unknown. Ideally, diagnostic testing should be carried out immediately after TEOAE screening but in practice this is rarely possible. Several infants with unsuspected severe or profound bilateral hearing loss were detected during the study. In Australia, the median age of diagnosis for permanent hearing loss is 18 mon and in over 50% of cases there is no parental suspicion of loss.15 A number of cases of substantial unilateral hearing loss were also detected. It is unlikely that these would have been otherwise diagnosed until early school years.

TEOAEs could be obtained from many children attending community child health clinics. The percentage of children who could not be tested in both ears varied, depending on age group. Children in the age range of those attending for initial immunisation had a low could not test rate and a low failure rate (3%) for those who were successfully screened. In summary, TEOAE testing appears a promising new technique for use in community based hearing screening. However, further development of instrumentation and pass=fail criteria may be required prior to the widespread introduction of the technique. Acknowledgements This study was supported by a Public Health Research and Development Committee project grant from the Australian National Health & Medical Research Council, a grant from the Sylvia and Charles Viertel Charitable Foundation and an equipment grant from the Garnett Passe and Rodney Williams Memorial Foundation. The authors would like to thank the Immunisation Services of the Brisbane City Council and the nursing staff of Queensland Health for their cooperation and assistance during the course of this study, and Janet Wong for her valued assistance in data collection. An earlier version of this paper was presented at the 10th Annual Workshop on Hearing Screening in Children, MRC Institute of Hearing Research, Nottingham, in March 1997. References 1 Early Identi®cation of Hearing Impairment in Infants and Young Children. National Institutes of Health Consensus Statement 1993, 1 ± 24. 2 Kemp DT, Ryan S. The use of transient evoked otoacoustic emissions in neonatal hearing screening programs. Sem Hear 1993; 14: 30 ± 45. 3 Glattke TJ, Kujawa SG. Otoacoustic emissions. Am J Audiol 1991; 1: 29 ± 40. 4 Salamy A, Eldredge L, Sweetow R. Transient evoked otoacoustic emissions: feasibility in the nursery. Ear Hear 1996; 17: 42 ± 48. 5 Jacobson JT, Jacobson CA. The effects of noise in transient EOAE newborn hearing screening. Int J Ped Otorhinolaryngol 1994; 29: 235 ± 248. 6 Vohr BR, White KR, Maxon AB. Effects of exam procedures on transient evoked otoacoustic emission (TEOAEs) in neonates. J Am Acad Audiol 1996; 7: 77 ± 82. 7 Welch D, Greville KA, Thorne PR, Purdy SC. In¯uence of acquisition parameters on the measurement of click evoked otoacoustic emissions in neonates in a hospital environment. Audiology 1996; 35: 143 ± 157. 8 Early discharge from the newborn nursery. A potential threat to effective newborn screening [editorial]. Screening 1994; 3: 45 ± 48. 9 Vohr BR, White KR, Maxon AB, Johnson MJ. Factors affecting the interpretation of transient evoked otoacoustic emission results in neonatal hearing screening. Sem Hear 1993; 14: 57 ± 72. 10 Lonsbury-Martin BL, McCoy MJ, Whitehead ML, Martin GK. Clinical testing of distortion-product otoacoustic emissions. Ear Hear 1993; 14: 11 ± 22. 11 Birtles GJ. A visual reinforcement orientation audiometry screening procedure. Aust J Audiol 1989; 11: 1 ± 9. 12 Haggard M, Hughes E. Screening Children's Hearing. HMSO: London, 1991. 13 Kei J et al. Transient evoked otoacoustic emission in infants: Effects of gender, ear asymmetry and activity status. Audiology 1997; 36: 61 ± 71.

151

152

Community-based hearing screening B McPherson et al

14 Braveman PA, Tarimo E. Screening in Primary Health Care. World Health Organization: Geneva, 1994. 15 Robertson C et al. Late diagnosis of congenital sensorineural hearing impairment: why are detection methods failing? Arch Dis Child 1995; 72: 11 ± 15. 16 Dirckx JJJ et al. Numerical assessment of TOAE screening results: Currently used criteria and their effect on TOAE prevalence ®gures. Acta Otolaryngol (Stockh) 1996; 116: 672 ± 679.

17 Richardson MP et al. Otoacoustic emissions as a screening test for hearing impairment in children. Arch Dis Child 1995; 72: 294 ± 297. 18 Stevens JC et al. Click evoked otoacoustic emissions compared with brain stem electric response. Arch Dis Child 1989; 64: 1105 ± 1111.