Clinical evaluation of 1,000-Hz tympanometry for the diagnosis of exudative otitis media in Japanese infants

Tympanometry is usually performed using a probe tone frequency of 226Hz for older children and adults. It has recently been reported that 1000-Hz tympanometry may be more useful than 226-Hz tympanometry for measurement of the eardrum condition in infants. In this study, we analyzed the tympanogram of 384 ears of 195 Japanese children by both 226-Hz and 100-Hz probe tone, and examined whether the former or latter is more useful for the diagnosis of exudative otitis media. The tympanograms were categorized as A type, B type/flat, C type, negative peaked, double peaked or unanalyzable, in accordance with the classification of Jerger, Marchant, and Kei. The waveforms of the tympanograms were clearer in sleeping children than in awake children. Of the 384 ears, 316 82.3%-showed normal otomicroscopic results, 51 ears 13.3% showed findings suggestive of exudative otitis media, and 17 4.4%-showed evidence of canal stenosis. Among the tympanograms of the 316 ears with a normal tympanic membrane, 219-69.3%-showed normal tympanograms in the 226-Hz tympanometry, while 173 ears 54.7%-showed normal tympanograms in the 1000-Hz tympanometry. Therefore, 1000-Hz tympanometry provided less accurate diagnosis of normal waveforms. Of the 51 ears with exudative otitis media, type B/flat, type C or negative peaked waveforms, suggestive of exudative otitis media, were obtained in 39 ears 76.5% in 226-Hz tympanometry, and in 40 ears in 1000-Hz tympanometry. Analysis of the data revealed sensitivity for the diagnosis of exudative otitis media of 81.3% for 226-Hz tympanometry and of 88.9% for 1000-Hz tympanometry ; thus, the detection rate of 1000-Hz tympanometry was higher. On the other hand, the specificity was 86.6% for 226-Hz tympanometry and only 67.6% for 1000-Hz tympanometry. The accuracy, diagnostic odds ratio, positive likelihood ratio, and negative likelihood ratio were 85.7%, 27.9, 6.045, and 0.217 for 226-Hz tympanometry, and 70.8%, 16.7, 2.742, and 0.164 for 1000-Hz tympanometry. Thus, analysis of the data obtained from Japanese children in our study revealed a higher overall accuracy in the diagnosis of 226-Hz tympanometry than of 1000-Hz tympanometry for the diagnosis of exudative otitis media. Tympanometry is important for judging the state of the tympanic drum in infants, however, comprehensive analysis of the tympanic membrane findings, hearing test results, and imaging findings is necessary for the diagnosis of exudative otitis media.