Comparison of pure-tone audiometry analysis in sudden hearing loss studies: lack of agreement for different outcome measures

Abstract

Hypothesis: Different methods of pure-tone audiometry analysis in sudden hearing loss studies show a lack of agreement with respect to outcome, and the choice of outcome measure influences sample size calculation.

Background: There is an increasing number of clinical reports on the treatment of sudden hearing loss. For comparison of efficacy between different therapies and for meta-analysis, it is important to consider the outcome measures in different studies. Statistical analysis for comparison of different outcome measures is used in this study.

Materials and methods: Controlled clinical studies on the therapy for idiopathic sudden sensorineural hearing loss (ISSHL) were reviewed with respect to their choice of measurement method for the primary endpoint based on pure-tone audiometry. A selection of outcome measures, including different frequencies evaluated, absolute and relative hearing improvement, and different criteria for success, were compared by applying them on the same patient population treated for ISSHL. Agreement between different clinical outcome measures was assessed by using a method described by Bland and Altman. Based on the change in hearing level in the patient population, sample sizes were calculated for various outcome measures and for different criteria of hearing improvement.

Results: In 52 controlled studies on the systemic treatment of ISSHL, more than 40 different outcome measures for the primary endpoint were identified. Comparison of measurement methods showed no acceptable agreement between most absolute and relative measures of hearing improvement and between endpoints on the basis of different criteria of success. The differences in outcome in the treatment of ISSHL were more affected by the definition of success than by frequencies included in the pure-tone average. Estimated sample sizes for controlled clinical trials varied substantially with the choice of outcome measure.

Conclusion: The lack of an acceptable agreement between most measurement methods assessed in this report prevents a meaningful comparison of published clinical studies on ISSHL. For comparison of efficacy between different treatments in ISSHL and for the planning of controlled clinical trials, it is necessary to agree on standardized outcome measures, including the frequencies averaged and the criteria of success.