Analysis of Audiometric Differences of Patients with and without Tinnitus in a Large Clinical Database

Abstract

Human hearing loss (HL) and comorbidities like tinnitus pose serious problems for people's daily life, which in most severe cases may lead to social isolation, depression, and suicide. Here, we investigate the relationship between hearing deficits and tinnitus. To this end, we conducted a retrospective study on anonymized pure tone and speech audiometric data from patients of the ENT hospital Erlangen in which we compare audiometric data between patients with and without tinnitus. Overall data from 37,661 patients with sensorineural (SHL) or conductive HL (CHL) with (T, 9.5%) or without (NT, 90.5%) a tinnitus percept in different age groups and with different tinnitus pitches were included in this study. The results of the pure tone audiometry comparisons showed significant differences in T patients compared to NT patients. In young patients, we generally found lower hearing thresholds in T compared to NT patients. In adult patients, differences were more heterogeneous: hearing thresholds in T patients were lower in low frequency ranges, while they were higher at high frequencies. Furthermore, lower thresholds were more often found in CHL patients and could rarely be detected in SHL patients. In speech audiometry, only CHL patients with high-pitched tinnitus showed lower thresholds compared to NT patients' thresholds. The results of this study may point to a biologically plausible functional benefit on hearing thresholds in HL tinnitus patients. We hypothesize that the physiological mechanism of stochastic resonance counteracts HL by adding neuronal noise to the system. This neuronal noise may induce changes in the auditory pathway and finally-as a side effect of threshold improvement-lead to the development of a tinnitus percept. We propose a general model of changed hearing thresholds in T patients, being either decreased or increased compared to NT patients.

Keywords: conductive hearing loss; human; pure tone audiometry; sensorineural hearing loss; speech audiometry; tinnitus model.

Figure 1

Hearing loss (HL in decibel) in young patients (aged 1–17 years) with and without tinnitus. (A) Results of a 2-factorial ANOVA on HL with factors frequency (upper panel) and patient group (inset), interaction plot at the bottom. Symbols depict mean values; whiskers give 95% confidence intervals. Asterisks depict significance value of the 1-factorial ANOVA part: ***p < 0.001. Gray-shaded area indicates two frequencies not significantly different between groups (Tukey post hoc tests). (B) Mean differences between the HL of NT [blue in (A)] and T patients [red in (A)] quantifying the threshold difference referred to as mean HL difference, where positive values indicate lower, negative values higher hearing thresholds in decibel. Values significantly different from 0 are colored green (single sided t-tests).

Figure 2

Hearing loss (HL in decibel) in sensorineural (A,B) and conductive HL (C,D) patients. Symbols as in Figure 1. (A,C) Red areas indicate frequencies where T patients exhibit higher hearing thresholds compared to NT patients, the green area indicates frequencies where T patients show lower thresholds than NT patients (Tukey post hoc tests). (B,D) Red bars indicate significantly higher thresholds, green bars significantly lower thresholds.

Figure 3

Age dependency of hearing loss (HL in decibel) in sensorineural (A,B) and conductive HL (C,D) patients. (A,C) show the interaction plots of the 3-factorial ANOVAs of HL with factors age, patient group, and frequency. Asterisks depict significance levels of the Tukey post hoc tests in the interaction plot of age X patient group: ns, not significant, *p < 0.05, ***p < 0.001. Red shaded areas indicate frequencies with T patients exhibiting higher hearing thresholds compared to NT patients (Tukey post hoc tests separately for each frequency). Green shaded areas indicate frequency ranges (low: <1 kHz, mid: <4 kHz, high: ≥4 kHz) with T patients exhibiting lower hearing thresholds compared to NT patients (2-factorial ANOVAs for frequency ranges and patient groups). (B,D) give the mean HL difference of NT and T patients with red bars indicating significantly higher thresholds, green bars significantly lower thresholds in T patients compared to NT patients.

Figure 4

Relation between hearing loss in (HL in decibel) and tinnitus pitch in sensorineural (A,B) and conductive HL (C,D) patients. (A,C) show the results of the 2-factorial ANOVAs of HL with factors frequency and perceived tinnitus pitch (including NT patients) in the two patient groups. Asterisks depict significance levels of the Tukey post hoc tests: ns, not significant; ***p < 0.001. (B,D) The threshold difference between T and NT patients is given, with red bars indicating significantly higher thresholds, green bars significantly lower thresholds in T patients compared to NT patients.

Figure 5

Speech reception thresholds (decibel SPL) of sensorineural (A) and conductive HL (B) patients in the Freiburger speech intelligibility test (multisyllable numbers). 2-factorial ANOVAs of the 50% threshold with the factors age and perceived tinnitus pitch (including NT patients). Asterisks depict significance levels of the Tukey post hoc tests: *p < 0.05, *p < 0.01, ***p < 0.001.

Figure 6

Relation between mean relative tinnitus loudness (dB) and age, tinnitus pitch, and hearing loss (HL). The individual tinnitus loudness was calculated relative to the threshold at that frequency. Results of the 3-factorial ANOVA of tinnitus loudness relative to hearing threshold with the three factors described above. The upper three panels represent the 1-factorial part of the ANOVA results (age, HL, tinnitus pitch), the center line panels show the 2-factorial ANOVA interaction plots (age vs. HL, age vs. tinnitus pitch, tinnitus pitch vs. age), and the bottom line panel shows the interaction plot of all three factors.

Figure 7

Model of decreasing (green) or increasing (red) effects of tinnitus on hearing thresholds. In the center panel, the general effect of tinnitus on hearing threshold is given. Tinnitus is beneficial for thresholds in the low frequency range and detrimental in the high frequency range. The horizontal broken lines symbolize the significance levels. Two main factors alter this relationship by shifting the function either down (SHL) or left (tinnitus pitch). For more explanation, refer to the text.