Wideband reflectance in normal Caucasian and Chinese school-aged children and in children with otitis media with effusion.

OBJECTIVES Wideband reflectance (WBR) is a middle ear analysis technique that quantifies frequency-specific sound conduction over a wide range of frequencies. One shortcoming of WBR is that there is limited normative data, particularly for pediatric populations and children with middle ear pathology. The goals of this study were to establish normative WBR data for early school-aged children; to determine whether WBR differs significantly between Caucasian and Chinese children, male and female children, and children and adults (experiment 1); and to compare the normative pediatric WBR data with the WBR data obtained from children with abnormal middle ear conditions (experiment 2). DESIGN WBR was measured from 78 children with normal middle ear status with an average age of 6.15 yrs and 64 children with abnormal middle ear status with an average age of 6.34 yrs. Control group subjects and subjects without previously diagnosed middle ear pathology were recruited from eight elementary schools in the Greater Vancouver Area. Subjects with known middle ear pathology were recruited through the British Columbia Children's Hospital Otolaryngology department. Middle ear effusion (MEE) was identified in one of the two ways. In the British Columbia Children's Hospital group, MEE was diagnosed by a pediatric otolaryngologist (OTL) using pneumatic otoscopy and video otomicroscopy. These cases (21 ears) were classified as OTL confirmed. Subjects who were assessed through screenings at their elementary schools and suspected to have MEE based on audiological test battery results including elevated air conduction thresholds, flat low- and high-frequency tympanograms, and absent transient-evoked otoacoustic emissions were classified as not OTL confirmed (21 ears). Data were statistically analyzed for effects of gender, ethnicity (Caucasian versus Chinese), age (child versus adult), and middle ear condition. WBR equipment used for this study was from Mimosa Acoustics (RMS-system, version 4.03). Data were averaged in one-third octave bands collected from 248 frequencies ranging from 211 to 6000 Hz. RESULTS Control group subject data (experiment 1) revealed no effects of gender or ear, and their interactions with frequency were not significant. There was a significant interaction between ethnicity (Caucasian versus Chinese) and frequency. Chinese children had lower energy reflectance (ER) values over the mid-frequency range. ER was significantly different between pediatric data and previously collected adult data. Diseased group ER was significantly different among all four middle ear conditions (normal, mild negative middle ear pressure, severe negative middle ear pressure, and MEE) (experiment 2). The overall test performance of ER was objectively evaluated using receiver operating characteristic (ROC) curve analyses; it was compared across frequencies averaged in one-third octave bands. Statistical comparison of the area under ROC (AUROC) plots revealed that ER above 800 Hz (except for ER at 6300 Hz) had better test performance in distinguishing normal middle ear status from MEE compared with ER at 630 and 800 Hz. Although not statistically different from other frequencies between 800 and 5000 Hz, ER at 1250 Hz had the largest AUROC curve (sensitivity of 96% and specificity of 95%) and was selected for further analysis. Comparison of AUROC curves between WBR at 1250 Hz and static admittance at 226-Hz probe tone frequency revealed significantly better test performance for WBR in distinguishing between healthy ears and MEE. CONCLUSIONS A preliminary set of normative ER data have been generated for a pediatric population between the ages of 5 and 7 yrs, which were significantly different from previously gathered normative adult ER data. In this study, pediatric normative data were warranted for testing children, but ethnic-specific norms were not required to detect middle ear pathology and changes in middle ear status. WBR shows promise as a clinical diagnostic tool for measuring the mechanoacoustic properties of the middle ear and the changes that result in the presence of negative middle ear pressure or MEE.