Agreement and reliability of tinnitus loudness matching and pitch likeness rating

Abstract

The ability to reproducibly match tinnitus loudness and pitch is important to research and clinical management. Here we examine agreement and reliability of tinnitus loudness matching and pitch likeness ratings when using a computer-based method to measure the tinnitus spectrum and estimate a dominant tinnitus pitch, using tonal or narrowband sounds. Group level data indicated a significant effect of time between test session 1 and 2 for loudness matching, likely procedural or perceptual learning, which needs to be accounted in study design. Pitch likeness rating across multiple frequencies appeared inherently more variable and with no systematic effect of time. Dominant pitch estimates reached a level of clinical acceptability when sessions were spaced two weeks apart. However when dominant tinnitus pitch assessments were separated by three months, acceptable agreement was achieved only for group mean data, not for individual estimates. This has implications for prescription of some sound-based interventions that rely on accurate measures of individual dominant tinnitus pitch.

Figure 1. Example data from a Study 1 participant with tonal tinnitus.

In Step 3 of the ‘Tinnitus Tester’ participant always selected ‘Tonal’. Data are presented as mean ± SD for each test frequency. Loudness match is reported as the PA5 attenuator value, 0 attenuation (maximum output of the system) was 96 dB SPL.

Figure 2. Example data from a Study 1 participant with ‘atonal’ tinnitus.

In Step 3 of the ‘Tinnitus Tester’ participant selected different bandwidths across sessions (selection in brackets). Data are presented as mean ± SD for each test frequency. Loudness match is reported as the PA5 attenuator value, 0 attenuation (maximum output of the system) was 96 dB SPL.

Figure 3. Tinnitus loudness matching and pitch likeness rating.

Intraclass Correlation Coefficient (ICC) values for (A) loudness matching, and (B) pitch likeness, in Study 1 (5 repeated measures) and Study 2 (3 repeated measures), ± 95% CI. ICC values for Study 1 grouped according to three consecutive sessions for (C) loudness matching, and (D) pitch likeness. ICC> 0.7 (above red line) is considered good for group data and ICC> 0.9 is good for individual patient data.