A Large-Scale Investigation of the Influence of Tinnitus on Clinical Measures of Speech Perception in Quiet and Noise

Abstract

Purpose:: Many individuals with tinnitus report difficulty understanding speech in noise; however, it remains unclear whether tinnitus significantly affects speech perception abilities, particularly those measured in audiological practice. This study investigates the impact of tinnitus on two commonly used clinical assessments of speech perception: Word Recognition in Quiet (WRQ) and Quick Speech-in-Noise (QuickSIN). Research Design:: This retrospective study included adult patients (age 18 years or older) seen at the Stanford Ear Institute between 2015 and 2025 who underwent standard audiometry (250–8000 Hz), WRQ, and QuickSIN. Study Sample:: Our sample (n = 21,967 ears) included 20,433 control ears and 1,534 ears with a tinnitus diagnosis. All ears with conductive or mixed hearing loss, Ménière’s disease, or vestibular schwannoma were excluded. Data Collection and Analysis:: Participants were grouped according to the International Classification of Diseases, Ninth and Tenth revisions, Clinical Modification (ICD-9-CM and ICD-10-CM) codes for tinnitus (ICD-9-CM 388.3 and ICD-10-CM H93.1). Welch’s t tests were used to compare raw performance on WRQ and QuickSIN between the tinnitus and control groups. A generalized linear model was used to assess group differences while controlling for hearing sensitivity and age. Results:: Unadjusted WRQ and QuickSIN performance showed no significant differences between tinnitus (WRQ mean ± standard deviation [SD] = 95.08 ± 9.6 percent; QuickSIN mean ± SD = 4.67 ± 3.75 dB signal-to-noise ratio [SNR] loss) and control (WRQ mean ± SD = 95.21 ± 10.82 percent; QuickSIN mean ± SD = 4.61 ± 4.46 dB SNR loss) groups. After adjusting for age and hearing loss (high-frequency pure-tone average of 1, 2, 4, and 8 kHz), tinnitus was associated with slightly better performance in both quiet (β = 0.83, t = 3.62, p < 0.001) and noise (β = −0.59, t = −3.71, p < 0.001). However, using a multidimensional principal component analysis–based model of hearing, this effect disappeared for WRQ (β = 0.22, t = 0.98, p = 0.33) but persisted for QuickSIN (β = −0.49, t = −6.19, p < 0.001). Conclusions:: Tinnitus alone does not affect performance on WRQ and QuickSIN. Although formal assessment of perceived tinnitus severity or quality was not routinely conducted among our tinnitus cohort, these data suggest that although many patients with tinnitus report difficulty understanding speech, particularly in noise, these concerns are most likely attributable to hearing loss rather than to tinnitus. Clinical Relevance Statement:: The marginal differences found in WRQ and QuickSIN performance suggest that, with regard to speech recognition, the presence of a tinnitus diagnosis is unlikely to influence the management of individuals with tinnitus beyond appropriate management of their hearing loss. Thus, these results provide a valuable counseling tool for clinicians to help patients better understand and manage their expectations regarding the functional impact of tinnitus.