Hearing loss and dementia: is treating it one of the biggest levers against cognitive decline?

The scale of the claim: the biggest modifiable lever

Every few years, a group of dementia researchers convenes under the banner of the Lancet Commission on dementia prevention, intervention, and care — a standing expert panel that pools the best available evidence to estimate how much dementia might, in principle, be prevented by tackling risk factors we can actually change. Their 2020 report, led by Gill Livingston, is one of the most cited documents in the field. It identified twelve potentially modifiable risk factors that together account for around 40% of dementia cases worldwide — and at the top of that list, as the single largest individual contributor, sat hearing loss.²

The specific number the Commission attached to it is a population attributable fraction of roughly 8% — meaning that, in their modelling, about 8% of dementia cases could theoretically be avoided if midlife hearing loss were eliminated from the population. That figure edged out more familiar culprits like less education, smoking, and physical inactivity. It is worth being precise about what a population attributable fraction is: it is a modelled estimate of how much of a disease burden is statistically attributable to a factor across a whole population, assuming the association is causal. It is not a measured cure rate, and it rests on that big “assuming causal” caveat — a caveat this entire article circles back to.

Still, the ranking is not an accident of one report. It reflects a genuinely large and consistent body of observational evidence. The landmark early study came from Frank Lin and colleagues in 2011, published in Archives of Neurology: they followed 639 dementia-free adults for a median of nearly 12 years and found that the risk of incident dementia rose with the severity of baseline hearing loss, in a dose-dependent way, even after adjusting for age, sex, and vascular risk factors.³ Compared with normal hearing, mild, moderate, and severe hearing loss carried progressively higher dementia risk (hazard ratios of roughly 1.9, 3.0, and 4.9 respectively). A dose-response gradient like that — more of the exposure, more of the outcome — is one of the classic fingerprints that makes epidemiologists take an association seriously.

The 2018 meta-analysis by David Loughrey and colleagues in JAMA Otolaryngology–Head & Neck Surgery pulled the whole literature together: 36 studies and more than 20,000 participants. Age-related hearing loss was significantly associated with declines across all the main cognitive domains and with a higher risk of cognitive impairment and incident dementia.⁴ The direction and consistency of the finding, across many independent cohorts, is what earns the association its STRONG grade on the Evidence Radar above. What that grade covers is the association — that hearing loss and dementia travel together, and that on the Commission’s modelling hearing loss is the largest modifiable factor. It does not, on its own, establish that treating hearing loss reverses the risk. That is a different question, and it needs a different kind of study.

Why hearing loss might harm the brain: four hypotheses

A strong association is more believable when there is a plausible biological story for why it might be causal. For hearing loss and dementia there are several, and they are not mutually exclusive — the truth may be some combination. Timothy Griffiths and colleagues laid out the leading candidates in a 2020 Neuron review, and they map onto four broad ideas.⁵

1. Cognitive load, or “effortful listening.” When the ear delivers a degraded signal, the brain has to work harder to decode it. The idea is that the extra mental effort of straining to follow a conversation continually siphons cognitive resources — attention, working memory — away from encoding and remembering what is actually being said. Over years, the hypothesis goes, this chronic overload may erode cognitive reserve, the brain’s buffer against decline. It is an appealing, intuitive mechanism, though hard to prove is doing lasting damage rather than just making listening tiring in the moment.

2. Reduced social engagement. Hearing loss makes conversation exhausting and isolating, and people often withdraw from the social life that once kept them mentally stimulated. Social isolation and loneliness are themselves recognized dementia risk factors in the Lancet Commission’s own model.² So hearing loss may act partly through isolation: it quiets the world, the person pulls back, and the loss of rich social and cognitive input accelerates decline.

3. Accelerated brain atrophy. Neuroimaging studies have observed that people with hearing loss show faster shrinkage of brain tissue over time, including in regions of the temporal lobe involved in both hearing and memory. One interpretation is “use it or lose it”: an under-stimulated auditory system may drive structural changes that spill into neighbouring memory circuitry. Whether the atrophy is caused by the deprivation, or is a shared consequence of some upstream process, is not settled.

4. Shared or common-cause pathology. This is the deflating possibility. Perhaps hearing loss doesn’t cause dementia at all — perhaps a single underlying process, such as microvascular disease, neuroinflammation, or a common neurodegenerative pathway, damages both the auditory system and the memory centres in parallel. On this view, hearing loss is an early marker of a brain already in trouble, not a driver of the trouble. Griffiths and colleagues take this alternative seriously, including a role for the medial temporal lobe in auditory cognition that could tie the two together mechanistically.⁵

These mechanisms are plausible and partially supported, which is why the Evidence Radar grades them MODERATE rather than strong. Plausibility is necessary but not sufficient: the cognitive-load and isolation stories are the ones a hearing intervention could, in principle, reverse; the shared-pathology story is the one under which hearing aids would help your hearing but do nothing for your brain. Only a randomized trial can start to separate them — which brings us to the study everyone was waiting for.

If hearing loss causes cognitive decline, treating it should slow that decline. If hearing loss merely marks a brain already declining, treating it will not. A trial is how you tell the difference.

The ACHIEVE trial: the result the headlines skipped

Observational studies can show that hearing loss and dementia travel together, but they can never fully prove that fixing one fixes the other. For that you need a randomized controlled trial (RCT) — a study that randomly assigns people to receive an intervention or not, so that the two groups differ, on average, only in the treatment. Randomization is the tool that breaks the confounding and reverse-causation problems that haunt observational data. The pivotal RCT here is ACHIEVE — the Aging and Cognitive Health Evaluation in Elders trial — led again by Frank Lin and reported in The Lancet in 2023.¹

The design was serious. ACHIEVE enrolled 977 adults aged 70 to 84 with untreated hearing loss and no dementia, and randomly assigned them either to a comprehensive hearing intervention (audiological assessment, fitted hearing aids, and ongoing support) or to a control condition of successful-aging health education. Everyone was followed for three years, with a battery of memory and thinking tests. Crucially, the participants came from two different populations: a group of older adults already enrolled in a long-running cardiovascular study — the Atherosclerosis Risk in Communities (ARIC) study — who were on average older and at higher risk of cognitive decline, and a group of healthy community volunteers recruited specifically for ACHIEVE, who were younger and healthier.

The primary result — the one ACHIEVE was designed and powered to answer — was null. Across the full cohort of 977 people, three years of hearing intervention did not slow cognitive decline compared with health education.¹ The two groups declined at statistically indistinguishable rates. For a field that had spent a decade building the “treat hearing, protect the brain” narrative, this was a sobering headline — and it is the honest headline. A well-designed RCT built to test the causal claim in the general population of older adults with hearing loss did not find the effect. That is why the Evidence Radar grades “hearing intervention slows decline in all older adults” as WEAK: the best trial we have came back null on exactly that question.

But the story does not end there — and this is where careful reading matters. ACHIEVE had a pre-specified plan to examine its two sub-populations separately, precisely because the researchers anticipated they might respond differently. And they did. In the higher-risk ARIC sub-cohort — the older participants with more vascular risk and faster baseline decline — the hearing intervention slowed three-year cognitive decline by roughly 48% relative to control.¹ In the healthier community volunteers, there was essentially no effect. Averaged together, the strong benefit in one group and the absence of benefit in the other produced the null primary result. The signal was real; it just lived entirely in the higher-risk half.

Reading the subgroup carefully

A 48% slowing of cognitive decline is a large, exciting effect — and exactly the kind of finding that demands caution rather than a victory lap. Subgroup results, even pre-specified ones, are among the most over-interpreted numbers in medicine, and it is worth naming why so you can weigh the ACHIEVE finding honestly.

When a trial slices its population into subgroups and tests each, some will show an effect by chance alone — the more slices, the more likely a false positive. That risk is lower here because ACHIEVE pre-specified the ARIC-versus-volunteer split before seeing the data, with a biological rationale (higher-risk people have more decline to prevent, so an effective intervention has more room to work). A pre-specified subgroup with a plausible mechanism is far more credible than a subgroup fished out after the fact. But it remains a subgroup, not the primary endpoint. The trial was powered to detect an effect in the whole cohort; within a subgroup the numbers are smaller, the confidence interval — the statistical range of plausible true values, given the data — is wider, and the estimate is less certain. And the result is a single trial. Nothing has yet replicated it.

This is why the Evidence Radar grades the higher-risk subgroup finding EMERGING, not strong. It is a genuine, promising, biologically coherent signal from a rigorous trial — the best evidence to date that treating hearing loss can slow cognitive decline in the people most at risk. But it is one subgroup, in one study, awaiting confirmation. The correct posture is neither dismissal nor triumph: it is interested, cautious optimism, pending replication and the longer follow-up ACHIEVE’s investigators are pursuing.

What the observational hearing-aid data show

Outside the trial, a body of observational research has asked a simpler question: do people who use hearing aids show less cognitive decline than people with untreated hearing loss? Several long cohort studies suggest they do. The most-cited is Hélène Amieva and colleagues’ 25-year study of nearly 3,700 older French adults, published in the Journal of the American Geriatrics Society in 2015. Self-reported hearing loss was associated with faster cognitive decline over the quarter-century follow-up — but among the people with hearing loss who used a hearing aid, the rate of decline was no different from those with normal hearing.⁶ In other words, treatment appeared to erase the extra decline that untreated hearing loss carried.

That is an encouraging pattern, and it aligns with the higher-risk ACHIEVE subgroup. But observational hearing-aid studies carry a serious, hard-to-remove bias that is worth being blunt about: the people who seek out, buy, and consistently wear hearing aids differ systematically from those who don’t. They tend to be more educated, wealthier, more health-engaged, more socially connected, and less cognitively impaired to begin with — all of which independently protect against dementia. So some, perhaps much, of the apparent benefit of hearing aids in these studies may reflect who chooses to use them rather than what the devices do. It is precisely this kind of confounding that the ACHIEVE randomization was designed to eliminate — which is why the trial’s null primary result is a genuinely important corrective to the rosier observational picture. (One widely publicized 2023 hearing-aid-and-dementia analysis of a large biobank was subsequently retracted for a data-handling error; we have deliberately excluded it here and rely on the durable evidence.)

Association vs causation: reverse causation and confounding

Step back, and the whole hearing-and-dementia question turns on the same distinction that governs most of preventive medicine: the gap between association and causation. The association is not in doubt — hearing loss and dementia are robustly linked. What is genuinely uncertain is the arrow of cause.

Reverse causation is the first threat. The early stages of a dementing process can begin years or even decades before diagnosis, and one of the things a subtly declining brain may do is disengage from effortful listening — struggling more with speech-in-noise, participating less, appearing to “lose hearing” that is at least partly a central processing problem rather than a purely cochlear one. In such cases the incipient dementia is degrading the person’s functional hearing, not the other way around. The hearing loss is an early symptom, and treating the ear would not touch the underlying neurodegeneration.

Confounding is the second. Education, cardiovascular and vascular risk, diabetes, socioeconomic status, and social connectedness all influence both hearing and cognition. A person with less education and more vascular disease is at higher risk of both hearing loss and dementia for reasons that have nothing to do with a causal link between the two. Good observational studies adjust for these, and the association survives adjustment, which is why it is credible.³⁴ But statistical adjustment can never fully remove unmeasured confounding, and it cannot manufacture the certainty that randomization provides.

This is the deep reason the ACHIEVE trial matters so much, and why its overall null result should temper the strongest claims. Randomization neutralizes both reverse causation and confounding by design. When a rigorous RCT fails to find, in the general population, the effect that observational studies suggested, the honest interpretation is that at least part of the observational association was not the causal, treatable relationship it appeared to be. The signal that survived — in the higher-risk subgroup — is the part most worth taking seriously going forward.

Who should act — and why it’s worth it anyway

Here is the pivot that keeps this from being a deflating story: even after the null trial, treating hearing loss remains one of the clearest good bets in the whole preventive-health landscape. The reason is that the cognitive question, however it resolves, is not the only reason — or even the main reason — to fix your hearing.

Untreated hearing loss degrades quality of life in ways that are large, immediate, and well established. It fuels social isolation, depression, communication breakdown, workplace difficulty, safety risk, and strain on relationships. Correcting it — with modern hearing aids that are smaller, smarter, and increasingly affordable, including over-the-counter options in some markets — reliably improves those outcomes. And the intervention is genuinely low-risk: no drug, no surgery, no meaningful downside beyond cost and the nuisance of adaptation. When an intervention improves daily life on its own merits, carries little risk, and might also protect cognition in higher-risk people, the risk-benefit math is easy. This combination — strong quality-of-life benefit plus low risk — is what earns treating hearing loss its STRONG grade on the Evidence Radar, independent of the dementia question.

Who should be most motivated by the cognitive angle specifically? On the current evidence, it is older adults with hearing loss who also carry higher dementia risk — those with vascular risk factors, existing cognitive concerns, or the ARIC-like profile in whom ACHIEVE found benefit.¹ For a healthy 60-year-old with early hearing loss, the honest message is: treat it for your quality of life and social connection, and know that any cognitive dividend is plausible but unproven for you specifically. For a 78-year-old with hypertension, diabetes, and untreated hearing loss, the case that treatment may also help protect thinking is more compelling. Either way, the first step is the same: a proper hearing assessment with an audiologist, not a self-diagnosis. If you want to see where this sits among the other genuinely modifiable levers of brain aging, our brain & cognitive hub and longevity hub lay out the fuller picture, and our Evidence Radar index shows how we grade claims like these.

The hype: the guaranteed-shield claim

With a true and eye-catching premise — “hearing loss is the #1 modifiable dementia risk factor” — the marketing writes itself, and it overshoots. The claim to be wary of is the one that turns a plausible, subgroup-supported possibility into a certainty: that hearing aids prevent dementia, full stop, for everyone who wears them. That is where the evidence and the sales pitch part ways.

The ACHIEVE trial — the single best test of this exact claim — found no cognitive benefit in the overall population of older adults with hearing loss.¹ The benefit was confined to a higher-risk subgroup, in one study, awaiting replication. To leap from that to “hearing aids are a proven dementia shield” is to ignore the primary result of the very trial being invoked. It is the same error the wellness industry makes with every biomarker: taking a real association, or a real subgroup signal, and inflating it into a guarantee. That guaranteed-shield framing is why the Evidence Radar grades “hearing aids definitively prevent dementia for everyone” as HYPE. Hearing aids are excellent devices with real, well-documented benefits. Universal dementia prevention is not, on current evidence, one you can promise anyone.

What this article is not saying

This is not “hearing loss doesn’t matter for the brain.” The opposite: it is the largest single modifiable dementia risk factor in the leading prevention model, the association is strong and replicated, and there is a credible — if unproven — causal case, now backed by a positive higher-risk trial subgroup.¹² Dismissing the link is as wrong as over-hyping it.

This is not “don’t bother treating hearing loss.” Treat it. The case for correcting hearing loss stands firmly on quality of life, social connection, and mental health alone, at very low risk — and there is a plausible cognitive bonus on top, strongest for those already at higher risk. The point is to treat it for the right, well-evidenced reasons, not on the strength of a guarantee the data don’t support.

And this is not a personal prognosis. The attributable fractions and hazard ratios in this piece are population statistics, not sentences handed to any individual. Plenty of people with untreated hearing loss never develop dementia, and hearing aids are not a countdown you can stop. If your hearing has changed, or you have concerns about your memory, the right move is a hearing assessment and a conversation with a clinician — not a purchase made in fear, and not reassurance bought from a marketing claim. Read the evidence for what it is: a strong reason to care for your hearing, and a careful, honest limit on what that care can promise your brain.