Losing Both Senses: Hearing, Vision and Dementia Risk

Abstract

Losing hearing raises the risk of dementia. So does losing vision. Losing both raises it by considerably more than either alone — and by more than the two added together would suggest.

A 2025 meta-analysis pooled 13 studies and 383,326 participants, and found a hazard ratio of 1.53 for dual sensory impairment against no impairment.

What makes this worth reading is not the headline number. It is that every sensitivity analysis the authors ran made the finding stronger and the heterogeneity smaller, which is the opposite of what usually happens.

1.The Analysis

Nina Zumbrunn, Kate Beckett, Fiona Newell, David McGovern and colleagues, working between Dublin City University, Trinity College Dublin and Stanford, searched PubMed, Web of Science and PsycINFO for studies comparing dementia risk in people with both visual and hearing impairment against those with one or neither.

The review was pre-registered on PROSPERO and followed PRISMA reporting standards. Thirteen studies reported hazard ratios and entered the meta-analysis, covering 383,326 participants.

Heterogeneity was low to moderate, at an I-squared of 32.75 per cent, and neither funnel plot inspection nor Egger's test found evidence of publication bias. Those two facts do more for the credibility of what follows than the effect size does.

2.The Result, and Why It Holds Up

Dual sensory impairment against no sensory impairment: hazard ratio 1.53, with a confidence interval from 1.41 to 1.64.

The sensitivity analyses are where this becomes interesting. When the authors restricted the pool to studies using objective measurement of sensory function rather than self-report, the hazard ratio barely moved — 1.52, interval 1.41 to 1.64 — while heterogeneity fell to zero.

When they excluded studies without a clinically verified dementia diagnosis, keeping only DSM or ICD criteria, the hazard rose to 1.63, interval 1.49 to 1.77, and heterogeneity again fell to zero.

Restricting to follow-up of five years or more raised it slightly, to 1.56.

This pattern is unusual and it is the reason we regard the finding as solid. Tightening measurement and diagnosis usually shrinks an effect and reveals that the pooled estimate was carried by the weakest studies. Here it did the reverse: the more rigorous the subset, the larger and more consistent the association.

3.Why Both Is Worse Than Either

Hearing impairment alone, against no impairment, carried a hazard ratio of 1.18 with an interval from 1.12 to 1.25. Vision impairment alone carried 1.25, interval 1.14 to 1.35 — though with substantial heterogeneity at 72 per cent, which the authors flag and we repeat.

Formal comparison found the dual-impairment hazard significantly greater than either single impairment, at p below .001 in both cases.

That is more than additive in a loose sense, and it points at something specific. A person losing one sense can lean on the other: someone hard of hearing reads faces and lips, someone losing vision relies on sound to locate and identify. Losing both removes the compensation as well as the input.

It also removes most routes to conversation, which is where sensory loss meets social isolation — itself an established risk factor for dementia.

4.Association, and the Attempt at Causation

Everything above is observational, and the standard three possibilities apply. Sensory loss might cause cognitive decline; early dementia might cause apparent sensory loss through impaired central processing; or both might follow from shared vascular, inflammatory or ageing pathways.

Reverse causation is genuinely plausible here rather than a formality. Hearing and vision tests require attention and comprehension, and someone in the prodromal phase of dementia may perform worse on them without their ears or eyes having changed.

A 2024 study in BMC Medicine by Fan Jiang and colleagues attacked this directly, combining a cohort analysis and meta-analysis with Mendelian randomisation — using genetic variants as instruments, drawn from UK Biobank with 361,194 participants and FinnGen with 412,181.

Genetic variants are fixed at conception and cannot be caused by later dementia, so if the association survives, reverse causation is largely excluded. For hearing impairment it did survive, with an odds ratio for all-cause dementia of 1.74 — but the interval runs from 1.01 to 2.99.

A lower bound of 1.01 is the narrowest possible escape from the null. That result is consistent with a causal contribution and it does not establish one, and anyone citing the 1.74 without the interval is misrepresenting it.

5.What This Does and Does Not Support

It supports treating combined hearing and vision loss as a marker of substantially elevated dementia risk, and it supports assessing both rather than one.

It does not establish that correcting sensory impairment reduces dementia risk. That is a separate claim requiring a trial, and the trial exists — we cover it separately, and its primary outcome was null.

A 2024 analysis in JAMA Network Open by Erin Ferguson and colleagues adds a further wrinkle worth knowing: visual impairment and specific eye conditions relate differently to subsequent diagnoses of neurodegeneration, which suggests that lumping all vision loss together may obscure more than it reveals.

Editorial Comment

MindHeaven® sells supplements. Nothing in our range affects hearing, vision or dementia risk, and this article recommends none of our products.

We publish it because our own exercise series concluded that the honest ordering of the evidence puts hearing above every compound in our catalogue, and it would be dishonest to write that sentence and then never cover the topic.

The practical implication costs nothing and is not something we sell. If an older relative is struggling to follow conversation, or has quietly stopped going to things they used to enjoy, a hearing test and an eye test are worth more than any supplement in this library — including ours — and both are ordinary, available and often free.

Sensory loss that has developed gradually is also easy to miss in oneself, which is part of why it goes uncorrected for years. Noticeable changes in hearing or vision are a matter for a clinician, and the point of testing is not to prevent dementia but to find something treatable.