The largest trial of hearing aids and cognitive decline found no overall effect

The one large randomised trial designed to test whether treating hearing loss slows cognitive decline found that it did not. In ACHIEVE, three-year change in global cognition was −0.200 SD units in the hearing intervention group and −0.202 in the health education control group — a difference of 0.002 (95% CI −0.077 to 0.081), p = 0.96 [s1]. The trial also found, in a prespecified sensitivity analysis, that the effect differed significantly between the two populations it recruited from (p for interaction = 0.010), and that secondary result is the source of nearly every headline claiming hearing aids protect the brain [s1].

Why the question was worth a trial

Observational evidence linking hearing loss to cognitive decline is extensive and consistent. A systematic review and meta-analysis in JAMA Neurology screened 3,243 studies and included 31 (25 observational studies and 6 trials) covering 137,484 participants [s2]. Pooling eight studies with 126,903 participants and follow-up from 2 to 25 years, hearing aid users had significantly lower hazards of cognitive decline than people with uncorrected hearing loss: hazard ratio 0.81 (95% CI 0.76 to 0.87), with no detectable heterogeneity (I² = 0%) [s2]. Eleven studies of 568 participants found a 3% improvement in short-term cognitive test scores after starting hearing aids (ratio of means 1.03, 95% CI 1.02 to 1.04) [s2]. The authors' own conclusion called for randomised trials [s2].

The scale of the underlying problem is not in dispute. The World Health Organization records that more than 5% of the world's population — around 430 million people — require rehabilitation for disabling hearing loss, defined as loss greater than 35 dB in the better-hearing ear, and that over 25% of those aged above 60 are affected [s3]. WHO projects nearly 2.5 billion people with some degree of hearing loss by 2050, more than 700 million of whom will need rehabilitation [s3]. It lists increased risk of developing dementia and accelerated cognitive decline among the consequences of hearing loss left unaddressed [s3].

What ACHIEVE did, and what it found

ACHIEVE was a multicentre, parallel-group, unmasked randomised controlled trial at four community sites in the United States [s1]. It enrolled adults aged 70 to 84 with untreated hearing loss and without substantial cognitive impairment, from two distinct populations at each site: participants in the long-running Atherosclerosis Risk in Communities (ARIC) observational study of cardiovascular health, and healthy de novo community volunteers [s1].

Participants were randomly assigned 1:1 to a hearing intervention — audiological counselling and provision of hearing aids — or to a health education control in which they met individually with a health educator on chronic disease prevention topics, with follow-up every six months [s1]. The primary endpoint was three-year change in a global cognition standardised factor score from a comprehensive neurocognitive battery, analysed by intention to treat [s1].

Between 9 November 2017 and 25 October 2019 the trial screened 3,004 people and randomised 977 (32.5%), of whom 238 (24%) came from ARIC and 739 (76%) were de novo volunteers [s1]. Four hundred and ninety went to the hearing intervention and 487 to the control [s1]. Mean age was 76.8 years (SD 4.0); 523 (54%) were female and 858 (88%) were White [s1]. No significant adverse events were attributed to either arm [s1].

The primary analysis, combining both cohorts, found the two groups indistinguishable [s1]. Other prespecified sensitivity analyses that varied analytical parameters in the total cohort did not change that result [s1].

The subgroup, stated precisely

The ARIC participants were older, carried more risk factors for cognitive decline, and had lower baseline cognitive scores than the de novo volunteers [s1]. A prespecified sensitivity analysis found the effect of the hearing intervention on three-year cognitive change differed significantly between the two cohorts (p for interaction = 0.010) [s1].

That is what the trial reported. The published Lancet paper does not present that interaction as a demonstration that hearing aids work in high-risk populations; its own interpretation is that the 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" [s1]. The distinction between a hypothesis-generating interaction in a trial whose primary endpoint was null and an established benefit in a subgroup is not a technicality. Subgroup findings from null trials are among the most reliably non-replicating results in clinical research, and the ARIC subgroup here contained 238 people out of 977.

What this does and does not settle

It does not settle whether hearing loss causes dementia. ACHIEVE tested a three-year treatment effect on cognitive scores, not dementia incidence, and three years is short against the timescale on which dementia develops. It does not overturn the observational association, which remains strong and consistent [s2]; it does show that the most obvious causal reading of that association did not survive its first randomised test at three years [s1].

Nor does a null result on cognition make hearing aids ineffective at what they are for. Nothing in ACHIEVE speaks against treating hearing loss to hear better, and the trial recorded no significant adverse events from doing so [s1]. What changed in 2023 is the strength of one specific justification for treatment — that it will protect thinking — which now rests on observational data and a subgroup, not on a positive trial.

This article describes published trial results and is not medical advice. Decisions about hearing assessment or devices belong with an audiologist or physician.