Accuracy of tympanometric middle ear pressure determination in secretory otitis media: dose-dependent overestimation related to the viscosity and amount of middle ear fluid.

HYPOTHESIS: Tympanometric measurements of middle ear pressure in children with secretory otitis media are overestimated in a dose-response manner because of increased hysteresis explained by the viscosity and amount of middle ear fluid. BACKGROUND: Tympanometric middle ear pressure is important in evaluating children with secretory otitis media. These measurements are influenced by hysteresis appearing as a peak pressure difference in bidirectional tympanometry. This represents an inaccuracy of 0.5 x peak pressure difference, which is only 5 to 25 daPa in normal ears. However, previous experiments found increased hysteresis, suggesting an inaccuracy of 225 daPa in secretory otitis media ears. MATERIALS AND METHODS: In 56 patients with secretory otitis media, bidirectional tympanometry was performed; Type B curves were excluded. The middle ear fluid was semiquantified subsequently at surgery according to viscosity (serous, seromucoid, or mucoid) and amount (small, medium, or large). A control group included 28 normal children. Peak pressure difference was calculated by the difference between middle ear pressure determined by a positive and negative pressure sweep. RESULTS: Mean peak pressure difference was 10 and 69 daPa in the normal and secretory otitis media groups, respectively (p <0.001). However, peak pressure difference ranged to 205 daPa in the secretory otitis media group and showed a significant positive correlation to viscosity and amount of the fluid (both p <0.0001). CONCLUSION: Peak pressure difference is significantly increased in secretory otitis media because of additional damping explained by the viscosity and amount of the fluid. The mean error was 5 daPa in normal ears and 35 daPa in secretory otitis media ears, but ranged to greater than 100 daPa. These results were only a low estimate of the inaccuracy, because patients with Type B tympanograms could not be included, and errors of more than 100 daPa can be anticipated.