Our study of congenital hearing loss in a large Norwegian population of neonates shows a prevalence corresponding to international figures. The proportion of late-detected cases within our period of follow-up is also as expected in light of other studies. There was a decrease in the median age of diagnosis during the period, and the introduction of automated auditory brainstem response audiometry reduced the need for brainstem response audiometry under narcosis. The degree of coverage of the screening amounted to 98.8 % during the first two years, well inside the recommended degree of coverage of 98 % (10). In the ensuing years the degree of coverage was not recorded, but we have no reason to assume that it has declined.
The national and international recommendations stating that hearing loss should be diagnosed before the age of three months and treatment should be initiated before the age of six months were not reached, although we achieved a considerable reduction of the median age of diagnosis, from 13 to six months from the first to the second half of the period. In most cases, the delay in the diagnosis was due to parents who did not turn up for further examination in spite of repeated calls. This was especially the case during the first period, but occurred also during the second. The hearing centre has started to place a clearer emphasis on the likelihood of hearing loss when emissions are absent, as well as on the consequences of incomplete diagnosis and treatment.
Although eleven (35 %) children were diagnosed in the course of their first six months, there were 20 who were diagnosed too late for initiation of optimal treatment and habilitation programmes. Ten of these children were older than 12 months. The six children who were diagnosed before six months of age were examined by auditory brainstem response audiometry under narcosis after considerable pressure from parents who were either hearing-impaired or deaf themselves, or when there were strong indications of deafness after automated auditory brainstem response audiometry had been undertaken. In light of practices in which auditory brainstem response audiometry is undertaken under narcosis only from the age of six months onwards for anaesthesiological reasons, initiation of treatment within the recommended age cannot be achieved. In recent years we have therefore made use of auditory brainstem response audiometry during natural sleep, in practice also before the age of three months. This has enabled earlier diagnosis and initiation of treatment before the age of six months. Other studies have shown a median age for initiation of treatment of 2.5 months and eight months respectively (20, 21). Both these studies, as well as our own material, show that the proportion of children with normal emissions increases in pace with the number of otoacoustic emissions tests. One study found that 0.8 % of those screened were referred for auditory brainstem response audiometry after having undergone up to four otoacoustic emissions tests (20). In our study, 0.25 % of the children were referred to auditory brainstem response audiometry under narcosis after having gone through up to five otoacoustic emissions tests and automated auditory brainstem response audiometry.
The proportion of children who developed hearing loss after a normal neonatal screening is comparable to findings in a study that had a follow-up period of nine years and in which the median age of diagnosis was 26 months (22). The follow-up period is shorter in our study, where the youngest children only had 48 months of follow-up, and the eldest child was 86 months when diagnosed. Relocations out of the county also introduce a degree of uncertainty regarding the number of late-detected hearing loss cases.
Seven of the ten late-detected children had at least one risk factor for hearing loss. All of the nine children who had been admitted to the neonatal intensive care unit received treatment with aminoglycosides. This is the standard empirical treatment in cases of neonatal sepsis in Norwegian as well as most Western neonatal intensive care units. Because of the risk of hearing damage, the serum level is monitored closely in these children. Risk factors are assumed to be present in 10 – 30 % of all newborns, while 50 % of all children diagnosed with hearing loss have no known risk factors (Box 1) (10). Our figures correspond well with this: a total of 13 (42 %) of the 31 children who had no emissions in the neonatal screening had risk factors. Of these 13, nine had first-degree relatives with hearing loss and five tested positive for connexin 26. Heredity may be involved in 60 % of cases, and connexin 26 accounts for 50 – 80 % of the non-syndromic recessive cases (10). The presence of hearing loss in close family, especially among late-detected children, most likely implies that closer follow-up of these will result in earlier diagnosis and initiation of treatment.
Until April 2008, children with unilateral hearing loss detected by otoacoustic emissions testing were not referred to the hearing centre, because it was believed that hearing in one ear was sufficient for language development. Unilateral hearing loss is a risk factor for bilateral hearing loss, and 10.6 % will have progression to hearing loss also in the other ear (23). Currently, children with a unilateral outcome are therefore followed up with annual otoacoustic emissions testing and automated auditory brainstem response audiometry until their hearing threshold can be measured by pure-tone or speech recognition audiometry. One child in our material suffered from unilateral hearing loss as a neonate and was diagnosed with bilateral hearing loss at the age of 20 months.
The Joint Committee on Infant Hearing recommends that children in maternity wards be screened either with the aid of otoacoustic emissions or automated auditory brainstem response audiometry in a one-step or two-step model. The most common practice is to use a two-step model with otoacoustic emissions, followed by automated auditory brainstem response audiometry as required. It is recommended that children who have been admitted to neonatal intensive care be screened with the aid of automated auditory brainstem response audiometry to enable diagnosis of auditory neuropathy, since otoacoustic emissions testing will not reveal this condition (24). If automated auditory brainstem response audiometry had been used in the neonatal intensive care unit it is conceivable that the four children with late-detected hearing loss could have been diagnosed at an earlier stage. The time required is on average somewhat longer for automated auditory brainstem response audiometry than for otoacoustic emissions testing, but children and their parents can be spared the burden of repeated checks and worries (25). A false-positive screening results in increased worry on the part of 3.5 – 14 % of parents, and 8 % of mothers report that they treat their children differently (e.g. talking louder, clapping their hands) (2).
Before the introduction of hearing screening for otoacoustic emissions, the median age of diagnosis of serious hearing loss in children was 2.5 years (7). It is satisfying to see that a considerable reduction in the age of diagnosis has been achieved during the first ten years of hearing screening in Østfold county, although we have failed to reach the goal set by the Directorate of Health of initiating treatment before the age of six months in the majority of cases. Screening for otoacoustic emissions is an effective method for children in maternity wards, while children in neonatal intensive care units should be screened with the aid of automated auditory brainstem response audiometry.