Background: With the advent of newborn hearing screening programs, the need to verify the fit of hearing aids in young infants has increased. The recording of cortical auditory evoked potentials (CAEPs) for this purpose is quite feasible, but rapid developmental changes that affect response morphology and the presence of electrophysiological noise can make subjective response detection challenging.
Purpose: The purpose of this study was to investigate the effectiveness of an automated statistic versus experienced examiners in detecting the presence of infant CAEPs when stimuli were present and reporting the absence of CAEPs when no stimuli were present.
Research Design: A repeated-measures design was used where infant-generated CAEPs were interpreted by examiners and an automated statistic.
Study Sample: There were nine male and five female infants (mean age, 12 mo; SD, 3.4) who completed behavioral and electrophysiological testing using speech-based stimuli.
Data Collection and Analysis: In total, 87 infant CAEPs were recorded to three sensation levels, 10, 20 and 30 dB relative to the behavioral thresholds and to nonstimulus trials. Three examiners were presented with these responses: (1) “in series,” where waveforms were presented in order of decreasing stimulus presentation levels, and (2) “nonseries,” where waveforms were randomized completely and presented as independent waveforms. The examiners were given no information about the stimulus levels and were asked to determine whether responses to auditory stimulation could be observed and their degree of certainty in making their decision. Data from the CAEP responses were also converted to multiple dependent variables and analyzed using Hotelling's T2. Results from both methods of response detection were analyzed using a repeated measures ANOVA (analysis of variance) and parameters of signal detection theory known as d-prime (d′) and the area under the receiver operating characteristic (ROC) curve.
Results: Results showed that as the stimulus level increased, the sensitivity index, d′, increased for both methods of response detection, but neither reached the maximum possible d′ value with a sensation level of 30 dB. The examiners with the greatest experience and Hotelling's T2 were equally sensitive in differentiating the CAEP from noise.
Conclusions: Hotelling's T2 appears to detect CAEPs from normal hearing infants at a rate equal to that of an experienced examiner. A clinical instrument that applies Hotelling's T2 on-line, so that the likelihood of response detection can be assessed objectively, should be of particular benefit to the novice or less experienced examiner.
Multivariate Statistical Quality Control Based on Ranked Set Sampling
