Hearing loss is on the Lancet Commission's list of modifiable dementia risk factors, and of the fourteen it is the one people find most surprising. It is also the one where the gap between "associated with" and "causes" matters most — and where a large randomised trial has actually tested the question rather than leaving it to observation.
This is what the association is, what the trial found, and what a reasonable person should do about it.
The association
The 2024 Lancet Commission places hearing loss among the midlife risk factors and reports, from its meta-analysis, a hazard ratio of 1.37 (95% CI 1.00–1.87) for dementia in people with hearing loss compared with those without.
Note where that confidence interval starts. The lower bound sits at 1.00 — the value meaning no effect at all. The association is real and consistently observed across studies, and the precision around it is not as tight as the headlines suggest. That is worth holding in mind before anyone tells you hearing loss causes dementia as a settled fact.
The Commission's own summary of the fourteen factors, and what the 45% figure does and doesn't mean, is covered separately here.
Three explanations, and only one of them is good news
An association this consistent has at least three candidate explanations, and they have completely different implications.
One — sensory deprivation and cognitive load. If hearing is degraded, the brain spends more of its available resources reconstructing speech from an incomplete signal. That effort has to come from somewhere, and the proposal is that it is taken from the processes that would otherwise encode and remember what was said. Over decades, reduced input to auditory pathways may also contribute to structural change.
Two — social withdrawal. Hearing loss makes conversation exhausting, particularly in groups and in noise. People withdraw: fewer dinners, fewer phone calls, less of the sustained back-and-forth that is cognitively demanding in a useful way. Social isolation is itself on the Commission's list, so this route would mean hearing loss acts partly through another risk factor rather than independently.
Three — common cause. This is the uncomfortable one. The same underlying processes — small-vessel vascular disease, or early neurodegeneration — could damage both auditory pathways and cognition. Hearing loss would then be an early sign of the disease rather than a contributor to it, and treating it would change nothing.
Observational studies cannot separate these three. That is exactly why a trial was needed.
ACHIEVE: the trial that tested it
ACHIEVE is the substantial randomised test of whether treating hearing loss slows cognitive decline. Its design:
- 977 participants — 490 to hearing intervention, 487 to control
- Aged 70 to 84, all with untreated hearing loss and no substantial cognitive impairment at baseline
- Intervention: hearing aids, plus audiological counselling and a toolkit for self-management
- Control: health education sessions on chronic disease prevention — an active control, not nothing
- Three years, with the primary outcome being three-year change in global cognition on a standardised neurocognitive battery
The design detail that turned out to matter most is that participants came from two different sources. 238 were drawn from ARIC, an ongoing cardiovascular health study — an older group, with more risk factors for cognitive decline and lower baseline cognitive scores. The other 739 were newly recruited community volunteers, healthier on every relevant measure.
What it found
Across the whole trial, there was no significant difference between the groups (p=0.96). That is about as flatly null as a primary outcome gets, and any honest account of this evidence has to lead with it.
In the ARIC cohort, the hearing intervention was associated with a 48% reduction in three-year cognitive change compared with control (p=0.027).
In the newly recruited cohort, no significant difference (p=0.18).
How to read that honestly
A null primary result with a positive finding in one subgroup is the shape of result that gets over-claimed in both directions, so it is worth being careful.
The case for taking the subgroup seriously. The healthy volunteer group declined very little over three years in either arm. If there is almost no decline to slow, an intervention has nothing to act on, and a null result in that group tells you less than it appears to. The higher-risk cohort was the group in whom cognitive change was actually occurring and therefore the only group where a slowing effect could be detected in three years. The cohort split was part of the trial's structure rather than something found after the fact.
The case for caution. A subgroup of 238 people is small. Subgroup findings in trials with null primary outcomes have a poor track record of replicating, and three years is a short window for a condition that develops over decades. One trial does not settle this.
Where that leaves it. The reasonable reading is that hearing intervention may slow cognitive decline in older adults who already carry elevated risk, and that this has not been demonstrated in lower-risk populations. Anyone telling you hearing aids prevent dementia is ahead of the evidence. Anyone telling you the trial showed hearing aids do nothing has read only the first line of it.
The word that does the work: untreated
Everywhere in this literature the risk factor is specified as untreated hearing loss — and the gap between having hearing loss and doing something about it is enormous. People wait years, often a decade or more, between noticing and acting, and a large share of those who would benefit from hearing aids never get them.
The reasons are well documented and mostly not about the hearing: cost, the perception that hearing aids signal age, the belief that it isn't bad enough yet, and the fact that hearing loss arrives so gradually that there is no moment at which it becomes obvious. Partners usually notice before the person does.
Which produces a practical asymmetry worth naming. The dementia-prevention case for treating hearing loss is suggestive rather than proven. The case for treating hearing loss because you cannot hear is not in dispute at all — and it delivers the communication, the relationships and the reduced daily effort immediately, whatever the long-run cognitive question turns out to be. The cognitive argument is a bonus reason to do something already worth doing, not the reason.
What to actually do
Get tested if any of these apply: you ask people to repeat themselves · you struggle in restaurants or groups more than you used to · you have turned the television up and others comment · someone close to you has raised it · you are over 50 and have never had a hearing test.
A hearing test is quick, painless and widely available, and over-the-counter hearing aids have brought the cost of the milder end of this substantially down in several markets.
And treat it as one item on a longer list. Hearing is one of fourteen factors, and it is not the biggest. Blood pressure, lipids, blood sugar, smoking, alcohol, movement and social contact are all on the same list, and several of them are more actionable. The vascular cluster in particular does double duty, since it drives stroke risk as well.
Where we sit
Nothing Proco sells treats hearing loss or prevents dementia, and there is no supplement anywhere on the Commission's list of fourteen. The useful action here is an audiologist, not a bottle.
If your actual worry is your memory rather than your hearing, the more relevant article is this one — most memory worries turn out to have causes other than dementia, and several of them are reversible. A few of those are simple blood tests, which we've listed separately.
This article is not medical advice. Hazard ratio from the 2024 Lancet Commission on dementia prevention, intervention and care; trial figures from the ACHIEVE randomised controlled trial as reported at AAIC 2023 and published in The Lancet.
Sources: ACHIEVE trial, The Lancet · AAIC 2023 presentation of ACHIEVE results · Lancet Commission on dementia (2024)