STATISTICAL EVALUATION OF HEARING SCREENING BY DISTORTION-PRODUCT OTOACOUSTIC EMISSIONS

Citation
P. Bonfils et al., STATISTICAL EVALUATION OF HEARING SCREENING BY DISTORTION-PRODUCT OTOACOUSTIC EMISSIONS, The Annals of otology, rhinology & laryngology, 106(12), 1997, pp. 1052-1062
Citations number
35
ISSN journal
00034894
Volume
106
Issue
12
Year of publication
1997
Pages
1052 - 1062
Database
ISI
SICI code
0003-4894(1997)106:12<1052:SEOHSB>2.0.ZU;2-N
Abstract
The aim of this study was to provide a statistical evaluation of the s creening properties of distortion product otoacoustic emissions (DPOEs ) in individuals with clinically normal hearing and in patients with p ure sensorineural deafness of various degrees. The main informational parameters used were sensitivity, specificity, positive predictive val ue (PPV), negative predictive value (NPV), and the analysis of receive r operating characteristic (ROC) curves, For each frequency tested, ea rs were classified as a function of their audiometric threshold. Two g roups were defined relative to an arbitrary reference, the ''audiometr ic criterion.'' The PPV decreased and NPV increased with increases in the audiometric criterion. Each point of the ROC curve represents the relationship between the false alarm rate and the hit rate for each au diometric criterion ranging between 10 and 75 dB hearing level: the lo wer the audiometric criterion, the fewer the hit rate value, and the l ower the false alarm value. The audiometric criterion giving the highe st hit rate and the lowest false alarm rate was 55 to 60 dB hearing le vel for primaries at 60 and 70 dB sound pressure level, or 25 to 30 dB hearing level for primaries at 30, 40, and 50 dB sound pressure level , These two different behaviors of ROC curves are consistent with the hypothesis that DPOEs do not represent activity at a single location a long the basilar membrane.