Prognostic validity of auditory brainstem evoked response screening in newborn infants.

The effectiveness of the auditory brainstem evoked response (ABR) test as a screening test for permanent hearing loss in neonates is examined in 32 published studies including 4,945 infants. Follow-up studies validating initial test failure were available for only 658 infants. Significant differences in methodology exist among these studies, and these differences may have affected their outcomes. Despite this problem, data from these studies were pooled, and they revealed that 16.5 per cent of neonates failed the initial ABR test. Of these, 5.1 per cent were later confirmed to be hearing-impaired or otherwise neurologically impaired. This implies that many of the false-positive results occurred in neonates who had transient disorders, such as conductive hearing loss or neurologic abnormalities, at the time of the screening test, or that the pass/fail criterion was too stringent. Lack of data on false-negative results or the rate of correct identification of normal infants makes it impossible to assess the sensitivity of the test accurately. Firm estimates of the prognostic validity of the neonatal ABR screening in identifying the impaired population can be achieved only when the actual impairment rate can be estimated, and when the distributions of ABR latencies and thresholds for both normal and impaired populations are known. These data could be obtained by following groups of high-risk infants longitudinally and by pooling raw ABR data from the normal and impaired populations that are collected using standardized procedures and similar equipment.