Hearing loss and dementia: what the ACHIEVE trial actually found.

The relationship between hearing loss and dementia has been one of the most discussed findings in adult hearing health over the past fifteen years. The mainstream popular framing — "treating hearing loss prevents dementia" — is a compression of a more complicated and interesting picture. The ACHIEVE trial is the cleanest test of that hypothesis yet conducted, and its result is worth understanding precisely.

How we got here: the observational evidence

The modern interest in hearing loss as a dementia risk factor traces in large part to Lin et al. (2011), a Baltimore Longitudinal Study of Aging analysis of 639 adults followed for a median of 11.9 years. In that study, the risk of incident all-cause dementia increased log-linearly with the severity of baseline hearing loss: hazard ratio 1.89 for mild loss, 3.00 for moderate, and 4.94 for severe loss, compared to normal hearing.

The authors closed with the careful framing that the right question was now whether hearing loss was a marker for early-stage dementia or an actual modifiable risk factor. A decade of additional observational work in multiple cohorts pointed in the same direction: hearing loss is consistently associated with subsequent cognitive decline and dementia diagnosis. By the time the 2024 Lancet standing Commission updated its list of modifiable risk factors for dementia (Livingston et al.), hearing loss appeared as one of the largest midlife targets in their population-attributable-fraction framework.

But association is not causation, and the only way to test the modifiable-risk claim is a randomised trial.

The ACHIEVE trial design

The Aging and Cognitive Health Evaluation in Elders (ACHIEVE) study was a multicentre, parallel-group, unmasked randomised controlled trial conducted at four U.S. sites: Jackson, MS; Forsyth County, NC; Washington County, MD; and Minneapolis, MN. Participants were enrolled between November 2017 and October 2019; primary results were published in The Lancet in 2023.

Key design features:

• Participants: Adults aged 70–84 with untreated hearing loss (no current hearing aid use) and without substantial cognitive impairment.
• Two recruitment streams: (1) older adults already enrolled in the long-running Atherosclerosis Risk in Communities (ARIC) cohort study, who are by definition higher-risk for cardiovascular and cognitive outcomes; and (2) de novo healthier community volunteers recruited specifically for ACHIEVE.
• Randomisation: 977 participants randomly assigned 1:1 to either the hearing intervention (audiological counselling and provision of hearing aids) or to a health education control (individual sessions with a health educator covering chronic disease prevention topics unrelated to hearing).
• Follow-up: Every 6 months for 3 years.
• Primary endpoint: 3-year change in a global cognition standardised factor score from a comprehensive neurocognitive battery.
• Analysis: Intention to treat.

The cohort had a mean age of 76.8 years; 54% were female; 88% were White; and the ARIC subcohort (238 participants, 24%) was older, had more risk factors for cognitive decline, and had lower baseline cognitive scores than the de novo cohort (739 participants, 76%).

What the primary analysis found

In the primary intention-to-treat analysis combining both cohorts, the hearing intervention did not significantly reduce 3-year cognitive decline. Three-year cognitive change (in standard deviation units) was:

• Hearing intervention group: −0.200 (95% CI −0.256 to −0.144)
• Health education control: −0.202 (95% CI −0.258 to −0.145)
• Difference: 0.002 (95% CI −0.077 to 0.081); p = 0.96.

The two arms declined at essentially identical rates over three years. By the strict standard of "primary endpoint hit / not hit," the trial was negative.

Several sensitivity analyses that varied analytical parameters in the total cohort did not change that result.

What the subgroup analysis found

The trial's analysis plan included a prespecified sensitivity analysis comparing the effect of the hearing intervention between the two recruitment cohorts. That analysis was significant (p = 0.010): the effect of the hearing intervention differed meaningfully between the ARIC and de novo subcohorts. In the higher-risk ARIC subcohort, 3-year cognitive decline was slower in the hearing intervention arm; in the lower-risk de novo subcohort, the two arms were indistinguishable.

Two important caveats apply to that result. First, a prespecified subgroup analysis is more credible than a post hoc one, but it is still a secondary analysis and the trial as a whole is negative on its primary endpoint. Second, the ARIC subcohort had lower baseline cognitive scores and more cardiovascular risk factors, which means there was more room to observe a slowing effect — participants with already-low scores can drop further than those at ceiling.

The authors' own conclusion is calibrated:

"The hearing intervention did not reduce 3-year cognitive decline in the primary analysis of the total cohort. However, a prespecified sensitivity analysis showed that the effect differed between the two study populations that comprised the cohort. These findings suggest that a hearing intervention might reduce cognitive change over 3 years in populations of older adults at increased risk for cognitive decline but not in populations at decreased risk for cognitive decline."

No significant adverse events were attributed to either arm.

How this fits with the 2024 Lancet Commission

The 2024 Lancet standing Commission on dementia (Livingston et al.) retained hearing loss as a major modifiable midlife risk factor for dementia. Its framework rests on the population-attributable fraction: if a risk factor is causally related to dementia, eliminating it would in principle reduce population dementia incidence by some calculable percentage. The Commission's estimate places hearing loss among the largest single targets in its list.

ACHIEVE does not contradict that framework. The Commission's claim is at the population level — over a lifetime, treating hearing loss may reduce dementia incidence — while ACHIEVE tests an intervention at a particular age over a particular window (3 years, starting in older adults already aged 70–84). What ACHIEVE clarifies is that the intervention effect is conditional: it appears stronger in higher-risk older adults than in healthier older adults, and a 3-year follow-up may be too short to detect effects in healthier populations.

How to read the result if you are an adult thinking about hearing aids

Three things follow honestly from ACHIEVE for a practical decision-maker:

1. Treat hearing loss for the everyday reasons. The strongest case for treating hearing loss is, and always has been, the one ACHIEVE was not designed to test: people with hearing loss who get appropriate amplification report better communication, less withdrawal from social settings, less listening fatigue, and better quality of life. Those benefits do not require ACHIEVE to be positive on the cognitive endpoint to be real.

2. Don't frame hearing aids as dementia prevention. The honest 2026 framing is: hearing loss is associated with cognitive decline; treating hearing loss in higher-risk older adults may slow cognitive change; the totality of evidence does not yet support marketing hearing aids as a dementia-prevention intervention. Be skeptical of any seller or clinic that markets devices on that basis.

3. Don't wait. The 2024 Lancet Commission framework emphasises midlife as the period where modifiable risk factors matter most. Untreated mild-to-moderate hearing loss is now easier to address than at any prior point — through the FDA's OTC category, Apple's AirPods Pro Hearing Aid Feature, or prescription care. The strongest, most defensible reason to act is communication; whatever cognitive-aging effect exists is, on the evidence, more likely to accrue from earlier and sustained use rather than from a late-life intervention.

The bottom line

The ACHIEVE trial is the most rigorous test to date of whether treating hearing loss slows cognitive decline in older adults. Its primary analysis was null. Its prespecified subgroup analysis suggested a meaningful benefit in higher-risk older adults. Those two findings together support a careful, calibrated message: hearing loss matters for many reasons, and may matter for cognition in some populations, but is not a guaranteed dementia-prevention intervention. The 2024 Lancet Commission still places hearing loss among the largest modifiable midlife risk factors for dementia, and that population-level framework remains intact. For an adult thinking about hearing aids in 2026, the cleanest reasoning is to treat hearing loss for the daily-life benefits the evidence is strongest on, and to view any cognitive benefit as a possible, conditional, longer-horizon bonus rather than as the primary reason to act.