Predicting the degree of hearing loss using click auditory brainstem response in babies referred from newborn hearing screening.

OBJECTIVES: The predictive ability of the auditory brainstem response (ABR) is an important factor governing the program sensitivity of neonatal hearing screens. The study examined the accuracy of the click-evoked auditory brainstem response (ck-ABR) when undertaken below the age of 6 months (from expected date of delivery) in predicting the pure-tone thresholds subsequently found to be present in children with a congenital permanent childhood hearing impairment.

DESIGN: Children with permanent childhood hearing impairment were ascertained from neonatal screening programs that have been subject to longitudinal evaluation. Ninety-two children who had ck-ABR recorded when below 6 months of age and repeatable ear specific pure-tone audiometry were recruited. Those with recognized temporary middle ear effusions at either test were excluded. The relationship between ABR and pure-tone thresholds was tested using the Pearson correlation coefficient with a linear regression model used to estimate pure-tone threshold (dependent variable) from ABR (independent variable). Correlation coefficients were obtained for pure-tone frequencies at octave intervals between 0.25 kHz and 4 kHz and at various frequency combinations. The difference between ABR and pure-tone threshold was analyzed. Those with a difference of greater than 20 dB were further examined. The ABR and pure-tone differences were also compared in babies born at term and prematurely.

RESULTS: Of the 92 children recruited to the study two children had a confirmed auditory neuropathy spectrum disorder (2%) and 10 (11%) had an audiometrically confirmed progressive hearing impairment. When these children were excluded, there was a high linear positive correlation (r = 0.90, SE = 14.3 dB) between the ABR and pure-tone thresholds averaged at 2 to 4 kHz. Although the correlation varied for different audiometric configurations, in all cases with a sloping hearing loss the correlation with their best frequency was weaker than the correlation at 2 to 4 kHz. For the total cohort the mean difference between ABR and pure-tone thresholds averaged at 2 to 4 kHz was 4.4 dB (SD = 19.29). The modal difference was 0 dB (58%) and 76 % had a difference of 20 dB or less. ABR underestimated the subsequently recorded pure-tone thresholds by more than 20 dB in 11 children and 10 of these children showed progression of their hearing loss measured by serial pure-tone audiometry. ABR overestimated the pure-tone thresholds by more than 20 dB in 15 children. Nine of these children (60%) had suffered a perinatal illness and the mean difference between the pure-tone and ABR thresholds was significantly greater in those born at <35 weeks of gestation (p < 0.001).

CONCLUSIONS: There is a high positive linear correlation between ck-ABR and pure-tone average thresholds at 2 to 4 kHz. However, the predictive value of ABR is reduced in certain neonatal groups. The cause for this is discussed as are the implications for undertaking a test battery at this age to improve the predictive accuracy.