A comprehensive evaluation on the associations between hearing and vision impairments and risk of all-cause and cause-specific dementia: results from cohort study, meta-analysis and Mendelian randomization study

Abstract

Background: Epidemiological studies show inconsistent links between hearing/vision impairment and dementia risk. Using multisource data, we investigated how single or combined sensory impairments relate to risks of all-cause and specific types of dementia.

Methods: We employed a triangulation approach combining three methodologies. We analyzed 90,893 UK Biobank (UKB) adults to explore single and joint effects of hearing and vision impairments on all-cause and Alzheimer's disease (AD), vascular dementia (VD) and non-AD non-VD (NAVD). A meta-analysis of prospective studies involving 937,908 participants provided stronger evidence. Finally, we conducted Mendelian randomization (MR) analysis using genome-wide association studies from UKB (361,194 participants) and FinnGen (412,181 participants) to validate relationships between sensory impairments and dementia occurrence.

Results: In the UKB cohort study, compared to participants with normal hearing, those in the mild and severe hearing impairment groups had progressively and significantly higher risk of all-cause dementia (mild: HR1.52, 95%CI 1.31-1.77; severe: HR1.80, 95%CI 1.36-2.38), AD (mild: HR1.63, 95%CI 1.30-2.04; severe: HR2.18, 95%CI 1.45-3.27), VD (mild: HR1.68, 95%CI 1.19-2.37; severe: HR1.47, 95%CI 1.22-1.78), and NAVD (mild: HR1.47, 95%CI 1.22-1.78; severe: HR1.98, 95%CI 1.43-2.75). Besides, vision impairment was associated with an increased risk of all-cause dementia (HR1.55, 95%CI 1.18-2.04) and NAVD (HR1.51, 95%CI 1.07-2.13). Furthermore, dual sensory impairment was associated with stepwise increased risks of all-cause and cause-specific dementia than single hearing or vision impairment. In the meta-analysis of 31 prospective cohort studies, risks of all-cause dementia and AD were elevated in participants with single hearing impairment (all-cause dementia: HR1.30, 95%CI 1.21-1.40; AD: HR1.30, 95%CI 1.21-1.40) and dual sensory impairment (all-cause dementia: HR1.63, 95%CI1.14-2.12; AD: HR 2.55, 95%CI 1.19-3.91), while single vision impairment only associated with higher risk of all-cause dementia (HR1.43, 95%CI 1.16-1.71) but not AD. Finally, the MR analysis revealed a significant association between hearing impairment and all-cause dementia (OR1.74, 95%CI 1.01-2.99), AD (OR1.56, 95%CI 1.09-2.23), and NAVD (OR1.14, 1.02-1.26), as well as vision impairment and NAVD (OR1.62, 95%CI 1.13-2.33).

Conclusions: Our findings showed significant associations between hearing and vision impairments and increased risks of all-cause and cause-specific dementia. Standardized hearing and vision assessment and intervention should be emphasized in dementia prevention strategies.

Keywords: Alzheimer’s disease; Dementia; Hearing impairment; Vascular dementia; Vision impairment.

Fig. 1

Schematic diagrams illustrating the study designs. Panel A A cohort study with 90,893 participants from UK Biobank. Panel B A meta-analysis based on 93,7908 participants from 31 prospective cohort studies. Panel C A two-sample Mendelian Randomization analysis based on GWAS summary statistics derived from FinnGen (N = 41,218) and UK Biobank (N = 361,194); MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier; CI, confidence intervals

Fig. 2

Meta-analyses on the relationship of hearing and visual impairment with all-cause dementia and AD. Panel A shows the results of association between hearing impairment with all-cause dementia and panel B for AD. Panel C shows the results of association between vision impairment with all-cause dementia and panel D for AD. Panel E shows the results of association between dual sensory impairment with all-cause dementia and panel F for AD. AD, Alzheimer’s disease; CI, confidence interval; MD, mean difference

Fig. 3

MR analyses for the effects of hearing and vision impairment on the risk of all-cause and cause-specific dementia by using RE-IVW method based on GWAS of FinnGen and UKB. Panel A with the only FinnGen GWAS for exposure. Panel B with the UKB and FinnGen GWAS for exposure. Dots, mean odds radio; Horizontal lines, 95%CI; Arrows, the confidence interval extends beyond the displayed range; MR, Mendelian randomization; RE-IVW, random-effects inverse-variance weighted; UKB, UK Biobank