Prediction of Hearing Outcomes by Multiple Regression Analysis in Patients With Idiopathic Sudden Sensorineural Hearing Loss

Hideaki Suzuki,Takahisa Tabata,Hiroki Koizumi,Nobusuke Hohchi,Shoko Takeuchi,Takuro Kitamura,Yoshihisa Fujino,Toyoaki Ohbuchi
DOI: https://doi.org/10.1177/0003489414538606
2014-06-18
Abstract:Objective: This study aimed to create a multiple regression model for predicting hearing outcomes of idiopathic sudden sensorineural hearing loss (ISSNHL). Methods: The participants were 205 consecutive patients (205 ears) with ISSNHL (hearing level ≥ 40 dB, interval between onset and treatment ≤ 30 days). They received systemic steroid administration combined with intratympanic steroid injection. Data were examined by simple and multiple regression analyses. Three hearing indices (percentage hearing improvement, hearing gain, and posttreatment hearing level [HL post ]) and 7 prognostic factors (age, days from onset to treatment, initial hearing level, initial hearing level at low frequencies, initial hearing level at high frequencies, presence of vertigo, and contralateral hearing level) were included in the multiple regression analysis as dependent and explanatory variables, respectively. Results: In the simple regression analysis, the percentage hearing improvement, hearing gain, and HL post showed significant correlation with 2, 5, and 6 of the 7 prognostic factors, respectively. The multiple correlation coefficients were 0.396, 0.503, and 0.714 for the percentage hearing improvement, hearing gain, and HL post , respectively. Predicted values of HL post calculated by the multiple regression equation were reliable with 70% probability with a 40-dB-width prediction interval. Conclusion: Prediction of HL post by the multiple regression model may be useful to estimate the hearing prognosis of ISSNHL.
What problem does this paper attempt to address?