ERA in hearing screening in neonates and infants

CLINICAL PROBLEM

There is general agreement about the desirability of early detection and management of hearing loss. Even a comparatively mild hearing impairment can cause considerable speech abnormality and may delay the development of communication skills.

Various hearing screening programmes have been suggested, raising the questions of the selection of population to be screened, which tests to use, and what level of hearing loss one should aim to identify.

Population

Accurate hearing levels are difficult to measure in early infancy, and the epidemiological picture is inadequate. A World Health Organization (WHO) report (1967) suggests that the incidence of severe hearing loss in neonates is about 1:1000, and that of some degree of hearing impairment may be as high as 5:1000.

Certain children have an increased risk of hearing loss and are in a special ‘at risk’ group. It has been shown that in some groups, for example, very low birth-weight children (less than 1500 g) in intensive care, the incidence of bilateral severe hearing loss is 4% and the incidence of any degree of hearing loss, including mild and unilateral, reaches 9% for those aged 3–6 years (Abramovich et al 1979). It should be noted that low birth-weight alone is not a ‘risk’ factor for hearing loss. However, factors such as perinatal illness, particularly those conditions, including apnoea, known or likely to have caused hypoxia in the neonatal period, are significantly associated with sensorineural hearing loss, and jaundice appears to have an additive effect (Abramovich et al 1979). A register of the main risk factors is a useful way of identifying children who are most likely to suffer impaired hearing and who, therefore, need reliable testing. A practical version of a risk register, similar to that proposed by the Joint Committee on Infant Hearing (1982), has proved useful; it includes the following factors:

1. Family history of childhood hearing impairment

2. Congenital perinatal infections

3. Anatomical malformations involving head and neck

4. Birth-weight less than 1500 g

5. Hyperbilirubinaemia

6. Asphyxia with acidosis

The Joint Committee on Infant Hearing (1982) has estimated the incidence of moderate to profound hearing loss at 2.5–5% in the ‘at risk’ group. The incidence of risk factors is under 5%, and, as most babies with risk factors 1 and 3 (above) are to be found in the normal wards, the high-risk group is not necessarily concentrated among those infants treated in special-care baby units. In these groups, the incidence of severe hearing loss increases by at least tenfold, and that of some degree of hearing loss increases by much more. Some risk factors, for example, recessive familial hearing loss, viral infections and asphyxia depend on the investigation pattern for detection, and are especially prone to variability. About 95% of babies do not fall into a risk category; however, one should bear in mind that only half of the children with a hearing loss manifest a known risk factor (Simmons 1980, Feinmesser et al 1982, Stein et al 1983). This emphasizes the need for general awareness of the possibility of hearing loss.

It is relatively easy to schedule children from the special baby-care units; however, it is unsatisfactory to restrict testing to only this group of children. Systematic assessment of risk factors in the general nursery would be a major undertaking. Therefore, it is imperative that all medical practitioners involved in paediatrics become familiar with the risk categories and refer such children for early testing.

CHOICE OF SCREENING TESTS

Both behavioural and non-invasive electrophysiologic tests have been suggested and tried in early hearing loss detection. Various screening programmes have been suggested to assess the hearing of babies just before they leave hospital, or at a later date within the first 6 months of life.

The value of a test depends on its sensitivity and specificity. The sensitivity is the probability of a positive screen, which signifies the presence of a hearing loss. The specificity is the probability of a negative screen. A negative screen means that the test shows hearing within normal limits. It is important to avoid false-negative error rates, because this means missing identification of a hearing loss. False-positive error rates are undesirable, because they cause a false alarm which may lead to parental anxiety and further unnecessary investigations. Of two tests with the same sensitivity, the test which has a higher specificity is more valuable. However, one should bear in mind that the clinician’s skill and experience influence the error rate of any technique.

Observational audiometry and automated behavioural screening tests, such as the auditory-response cradle, identify neonates with severe and profound hearing loss fairly reliably (Feinmesser & Tell 1976, Bennett 1979, 1980, Weiss 1983, Bhattacharya et al 1984). Several reports have indicated high error rates with the standard behavioural tests and with the automated auditory-response cradle (Feinmesser & Tell 1976, Northern & Downs 1978, Davies 1984, McCormick et al 1984).

Oto-acoustic emissions from the cochlea recorded by means of a very sensitive microphone in the ear canal, are being evaluated as a screening test in neonates, and the initial data appear to reflect the results obtained in adults, namely that for hearing losses of greater than 30 dB HTL, the evoked emission is absent. It is a particularly suitable technique for neonatal screening, as neonates have a much larger evoked emission than do adults.

Auditory-evoked potentials have been used with enthusiasm by many investigators. ECochG is undesirable because it is an invasive procedure requiring general anaesthesia. The ABR, easily recorded in sleeping neonates and infants, seems to be superior to other tests in the screening of mild to profound hearing loss using broad-band clicks (Fig. 12.1, A, B, C). Frequency specificity can be improved by masking parts of the cochlea with band-stop or high-pass masking noise.

Fig. 12.1 (A) The Algotek screener in use on a neonate. Acoustic stimuli are delivered by plastic tubes to a circumaural, sealed cavity. This avoids the closure of the ear canal, which can occur with earphones. (B) ABR responses being recorded from a neonate. A miniature insert transducer is used to deliver the stimulus. (C) ABR screening of a 4-month-old child. The mother holds the earphone a few centimetres from the ear.

The postauricular muscle response (PAM) has been used for screening neonates and infants (Douek et al 1973, Fraser et al 1978). PAM depends on the tonus of the scalp muscles and is elicited only at relatively high intensities. It gives a high false-positive error rate, which limits its usefulness.

The middle-latency responses (MLR) and 40 Hz MLR have limited application in neonatal hearing screening. It has been reported that the MLR may be difficult to recognize in a sleeping neonate, and is less consistent and more difficult to record than ABR (Ozdamar & Kraus 1983, Rotteveel et al 1986). Frye-Osier & Hirsch (1980) tested normal neonates and recorded MLRs in response to 500; 2000; and 4000 Hz tone bursts at levels of 10, 30, and 50 dBnHL. The MLR was recordable at the 10 dB level for the 500 Hz stimulus and at the 30 dB level for the 4000 Hz stimulus, but only at the 50 dB level for the 2000 Hz stimulus. There is clear evidence of maturational changes in the MLR as it is symmetrically distributed over the scalp in adults, but show marked asymmetry in neonates (Wolf & Goldstein 1978). Both the MLR and the 40 Hz MLR thresholds are reported to increase in the sleeping infant. However, both the MLR and the 40 Hz MLR are more frequency-specific than ABR when using tone pips, and can be of value to assess hearing in the low frequencies. The middle-latency responses, including MLR, 40 Hz MLR, and SN10 can be considered in some difficult cases to supplement hearing investigation in a test battery.

The SVR is frequency specific, but in a sleeping infant the waveforms are unpredictable and variable.

In conclusion, ABR is the test of choice for screening neonates and infants with broad-band frequency clicks. If a more detailed audiometric contour is required, then some frequency specificity can be achieved by masking with the techniques of band-stop or high-pass masking noise. Other more frequency-specific MLR potentials can be recorded to assess hearing loss at low audiometric frequencies. Available ERA tests for screening along with the recommended parameters are shown in Table 12.1.

Table 12.1

ERA tests for screening