A novel audiogram model for predicting the prognosis of sudden sensorineural hearing loss.

Abstract

Background: Sudden sensorineural hearing loss (SSNHL) is an otological emergency involving hearing loss greater than 30 dB across three consecutive frequencies within 72 hours. Prognosis is influenced by the initial pure tone audiogram (PTA), with ascending and U-shaped patterns showing better outcomes and descending patterns linked to poorer outcomes. Traditional audiograms are classified into 4 to 7 types based on hearing thresholds at 4 to 6 frequencies, such as flat, ascending, or descending types. However, the definitions of these patterns are often ambiguous and not easy to apply in clinical practice. This study aimed to compare the prognostic value of the traditional seven-type audiogram model with a novel three-type audiogram model.

Methods: This retrospective cohort study included SSNHL patients from 2012 to 2023. Hearing outcomes, measured by PTA, were classified using both the traditional seven-type audiogram model and a new three-type model based on the 4 kHz threshold (the best, intermediate, the worst). The traditional seven-type model classifies audiograms into flat, descending, ascending, tent-shaped, U-shaped, jagged, and profound types based on the threshold patterns at four frequencies (0.5, 1, 2, and 4 kHz). The novel three-type model classified audiograms according to the relative position of the 4 kHz threshold: (1) 4 kHz the best (lowest threshold), (2) 4 kHz intermediate (threshold in between), or (3) 4 kHz the worst (highest threshold). Complete recovery (CR) was defined as a final hearing threshold ≤25 dB HL. Multiple and stratified linear regression analyses assessed its association with the CR rate, defined as the proportion of patients achieving CR.

Results: In total, 965 patients with SSNHL were recruited (51.5% men; mean age 50.8 ± 15.1 years). Left ear involvement was more common (53.3%). CR was achieved in 21.1% of patients. The three-type audiogram model classification were significantly associated with complete recovery (overall p < 0.001, multivariable logistic regression). 4 kHz was the worst, the CR rate was 20%, intermediate, 36%, the best, 49%. In contrast, the traditional seven-type audiogram model did not demonstrate such predictive characteristics.

Conclusion: The novel three-type audiogram model provides better prognostic predictions for SSNHL, with 4 kHz frequency being a key factor influencing recovery. Traditional audiogram classification is relatively complex, whereas the novel three-type audiogram model, which uses the 4 kHz threshold for classification, is simple, clear, and provides better prognostic prediction for SSNHL, making it more suitable for clinical application.