The Longitudinal Impact of Hearing Impairment on Cognition Differs According to Cognitive Domain

Identification and modification of the risk factors for cognitive decline throughout the adult life span are priority subjects in a progressively aging society; however, much remains to be learned. The aim of this study was to understand whether changes in cognitive function can be affected by hearing impairment (HI) and whether the impact of HI differs depending on the cognitive domain. A total of 1109 individuals aged 60-79 years at baseline who participated in the Longitudinal Study of Aging at the National Institute for Longevity Sciences (NILS-LSA) was followed up for a maximum of 13.3 years. Cognitive function was evaluated using four subtests of the Japanese Wechsler Adult Intelligence Scale-Revised Short Forms (JWAIS-R-SF): namely, Information, Similarities, Picture Completion, and the Digit Symbol Substitution subtests. The HI was defined as a pure-tone average of the better ear >25 dB. A longitudinal analysis of 4437 observations obtained during a follow-up period of approximately 12 years was performed. We estimated linear changes in subtest scores by HI status, using the time-varying mixed-effects regression model, which included fixed terms for the intercept, HI status at baseline, time (years elapsed since baseline) and an HI × time interaction term adjusted for age at baseline, sex, education, and other possible confounders. There were significant main effects of HI on the scores of the four subtests after adjustment. The HI × time interaction was significant for the scores of the Information (p = 0.001) and Digit Symbol Substitution subtests (p = 0.001). The scores of the HI group declined faster in the Information and Digit Symbol Substitution subtests compared to those in the no-HI group. The model-predicted 12-year slope using a mean baseline age (68.7 years) indicated no significant decline in the individuals without HI at baseline for the Information and Similarities subtests, however, this tolerance was lost in the individuals with HI. In conclusion, the present observation showed that the impact of HI on cognition was longitudinally significant and implied that the effect differs according to cognitive domain.