This study provides an approach to estimating tympanic membrane perforation-induced hearing loss (HL) using a human middle ear model. Sixty-one cases of tympanic membrane perforation originating from fireworks were reported from the Ear-Nose-Throat Department. The otoscope, audiometry data and diagnosis records were organized, and gender, age, etiology, perforation size and diseased ear side were classified as independent variables. A multinomial regression model was used to analyze the potential effects of the variables on HL. Meanwhile, a human middle ear model was implemented to calculate the ensued HL resulting from different perforation areas and sites. In addition, linear regression models were used to establish functions between perforation size and HL. The audiometry data indicate that HL at high frequencies (f > 2kHz) is much more profound than that at the speech frequency band (f < 1kHz). Compared with mild HL (<15dB), mediate HL (15-30dB) was correlated with the perforation area (p < 0.05, 95% CI), while severe HL (>30dB) was affected by both perforation size and age (p < 0.05, 95% CI). However, other factors, including gender and diseased ear side, do not show a statistically significant effect on HL. Furthermore, the Kruskal-Wallis test result reveals that HL at frequencies of 0.25 kHz ≤ f ≤ 8kHz is strongly associated with the perforation size (p < 0.05, 95% CI). It is conclusive that HL is positively proportional to the perforation size. However, HL is not correlated with the perforation site for small perforation areas of < 10% (p > 0.05, 95% CI).
Read full abstract