An observational pooled analysis of seven population ageing cohorts, published in Cell Reports Medicine on 12 May 2026, reported that hearing aid use among hearing-impaired adults aged 55 and over was associated with a lower rate of probable dementia (hazard ratio 0.91, 95 per cent confidence interval 0.88 to 0.94), and that the association was present only in participants who said their device delivered effective hearing improvement (HR 0.86, 0.80 to 0.93). Among users who reported poor improvement, there was no reduced risk (HR 0.98, 0.90 to 1.07). The team, led by Chen Shanquan of the University of Hong Kong's School of Public Health and Xu Lei of Shandong Provincial ENT Hospital, harmonised individual data from 61,089 people in CHARLS (China), KLOSA (Korea), SHARE (Europe), ELSA (England), TILDA (Ireland), HRS (United States) and MHAS (Mexico), recording 8,911 probable dementia events over a mean follow-up of 6.5 years.

The pattern differed by national income. In middle-income countries, hearing aid users had about a 24 per cent lower hazard of probable dementia than non-users (HR 0.76, 0.69 to 0.83), while in high-income countries a significant reduction appeared only among users reporting good effectiveness (HR 0.87, 0.81 to 0.95). In the authors' own framing, the priority in richer settings is fitting quality, follow-up and long-term effectiveness, while in middle-income settings affordable access to hearing care remains the constraint. This is an association, not a demonstrated effect: hearing aid users were not randomly assigned, and both the exposure and the rating of effectiveness came from what participants reported about themselves.

A separate analysis published in Neurology on 14 January 2026 shows why clinicians remain unsatisfied with this evidence base. Lachlan Cribb and colleagues at Monash University emulated a target trial in 2,777 Australian participants of the ASPREE study, aged 70 and over, dementia-free and with moderate hearing impairment at baseline and no prior hearing aid use. Over seven years, average scores on cognitive testing were essentially the same whether or not a hearing aid was prescribed (mean difference 0.03 standard deviations, 95 per cent confidence interval minus 0.14 to 0.21), yet the estimated seven-year risk of dementia was 5.0 per cent with a prescription against 7.5 per cent without (risk ratio 0.67, 0.37 to 0.97). The authors wrote that residual confounding could explain the divergence, and noted that self-reported hearing status could misclassify who was impaired.

Taken together, the two papers point in the same direction without closing the question. Neither was a randomised trial of hearing aids, and neither measured aided hearing objectively over time, which is the measurement both groups identify as missing. The practical distinction they raise is between owning a device and having one that demonstrably helps in daily life, a matter for an audiologist or clinician who can test and adjust it.