Reliability and effectiveness of screening

Objective-To establish the reliability and effectiveness of screening for hearing loss by brainstem auditory evoked potential testing in high risk neonates. Design-Seven year investigation of newborn babies admitted to a special care baby unit and monitored through a regional children's audiology unit. Setting-Special care baby unit and children's audiology department, Belfast. Subjects-405 neonates admitted to the baby unit, during 1 October 1982 to 31 March 1987. Main outcome measures-Presence of hearing impairment, type and severity ofhearing impairment, mortality. Results-85 children failed the screening test, 62 ofwhom were followed up. Five children had severe bilateral sensorineural impairment and 12 had conductive impairment requiring surgical intervention. A further 18 had severe neurological disorder detected. The sensitivity of screening was 100% and specificity was 88%. If the procedure was introduced into routine clinical practice the mean age at diagnosis for all children with severe perinatal hearing impairment would be 11 (median 1) months. The mean age at diagnosis with the health visitor screening service was 23 (19) months (difference 10 months, 95% confidence interval 6 to 16 months; p<0-0001). Conclusion-Screening for hearing loss in high risk neonates is highly reliable and cost effective. It also provides valuable neurophysiological information. Routine testing of these infants would result in over halfofall children with severe bilateral perinatal sensorineural hearing impairment being identified by 2 months of age. This would make an important contribution to the habilitation of this socially, emotionally, and educationally vulnerable group. first was to assess the validity and reliability of a screening programme for high risk neonates. The second was to compare the relative success of this programme with existing services and to assess the potential for improving the service by screening children at risk. The long term follow up in our study was made possible by the presence of a single regional children's audiology unit, to which all children with suspected hearing impairment were routinely referred (fig 1). Identification of false negative test results was also facilitated by the presence of two community screening tests for all children, one at 7 months and one at 3 years of age, and by routine pure tone audiometry at entry to school. iI Regional audiological service FIG 1-Flow diagram showing stages leading to full audiological assessmentfor different methods ofscreening and no screening Department of Mental Health, Queen's University, Belfast BT9 7BL R J McClelland, MD, professor ofmental health V Lawless, BA, clinical scientist Department of Otorhinolaryngology, Queen's University, Belfast BT9 7BL D R Watson, MSC, research officer D Adams, MB, senior lecturer -Children's Audiology Unit, Royal Belfast Hospital for Sick Children, Belfast H G Houston, MSC, senior audiological scientist Correspondence to: Professor McClelland. BMJ7 1992;304:806-9 Introduction Severe prelingual hearing impairment has important consequences for language acquisition, communication, and cognitive, social, and emotional development.'2 Indeed evidence is increasing that even moderate hearing loss in very young children can be detrimental.3 4 It is now generally accepted that appropriate remedial measures should be implemented at the earliest possible age and to this end screening for hearing loss in preschool children has been implemented in most health districts in the United Kingdom. The mainstay of current programmes is distraction testing carried out at 7-9 months of age.5 However, serious doubts have been raised regarding the efficacy and cost effectiveness of such programmes, and health visitor screening has been suggested to be inefficient and costly and to result in considerable delays in diagnosis. Attention has increasingly turned towards developing methods for the early and accurate evaluation of high risk infants.7'0 In view of the uncertainties regarding the relative merits of these two approaches we began a seven year study in 1982. The study had two related aims. The Subjects and methods The evaluation was based on 405 neonates admitted to the special care baby unit of the Royal Maternity Hospital, Belfast, during 1 October 1982 to 31 March 1987. A careful assessment was made of the following risk factors: familial deafness, rubella infection during pregnancy, birth weight <1500 g, congenital malformations of the ear, apnoea, asphyxia (Apgar score <7 at 5 minutes), respiratory difficulty (requiring respiratory support or more than respiratory oxygen), sepsis, hyperbilirubinaemia (requiring phototherapy or exchange transfusion"), and exchange transfusion. Auditory brainstem evoked potential testing was usually carried out after any acute conditions had settled; the modal gestational age was 37 (range 28-42) weeks. Neonates were investigated in their bassinet or incubator. Surface electrodes (silver/silver chloride) were attached to the vertex and both mastoid processes (impedance <5 kfl). The ear canals were inspected to ensure they were clear. Sound stimuli were generated by a Disa audiometer (Disa Elektronik, Denmark) using 80 ts pulses of alternating polarity delivered monaurally at a rate of 16 Hz through a TDH 39 BMJ VOLUME 304 28 MARCH 1992 806 earphone (Instrument Systems, New York) at two test levels, 30 dB normal hearing level (nHL) and 65 dBnHL. Electrical activity recorded from vertex and ipsilateral mastoid was amplified x 50000 and analogue, filtered with a band width of 30-3000 Hz by using a Grass Model 12 amplifier (Grass Instrument Company, Quincy, Massachusetts). Averages were formed from 2048 individual responses with a purpose built portable microcomputer." The criterion for passing the test was a definite response at 30 dBnHL in both