Detecting hearing-impairment in neonates--the statistical decision criterion for the Auditory Response Cradle

The Auditory Response Cradle (ARC) has great potential for detecting severe hearing impairments in neonates. This article reviews the ARC's statistical decision algorithm and presents data that show its present efficiency is below the proposed criterion (Bennett and Wade, 1980) in a high proportion of cases. Two alternative decision rules are proposed for use in screening neonates with the ARC. The first uses a test of fixed length and the second uses a statistically more controlled method based on the sequential probability ratio test of Wald.