Hearing Aids Slash Cognitive Improvement Rate Sixfold in Older Adults With MCI
A 703-person RCT finds hearing aids nearly 6x more likely to reverse mild cognitive impairment than education alone over 24 months.
Summary
Hearing loss is one of the biggest modifiable risk factors for dementia, but it is unclear whether fixing it helps people who already have mild cognitive impairment (MCI). In the CHOICE trial in Shanghai, 703 adults aged 60 or older with moderate to severe hearing loss and MCI received either free hearing aids or a hearing care education program. After two years, 3.1% of the hearing aid group and 4.7% of the education group had progressed to dementia-level impairment, a difference that was not statistically significant. However, 15.3% of hearing aid users improved to a normal global cognitive rating, compared with 2.4% of the education group. Because the study was open-label and the improvement finding was a secondary outcome, the results are encouraging but not definitive.
Detailed Summary
Hearing loss is considered the leading modifiable risk factor for dementia, with a population attributable fraction of roughly 5% to 7%. Observational studies link hearing aid use to delayed cognitive decline, but the first major randomized trial, ACHIEVE, found no overall cognitive benefit in cognitively healthy older adults. Whether hearing correction helps people who already have mild cognitive impairment (MCI), a stage at which intervention is increasingly emphasized, has remained unclear.
The CHOICE trial was a multicenter, open-label, parallel-group randomized trial run at three tertiary otology clinics and three community health stations in Shanghai. Of 21,908 adults aged 60 or older screened, 703 with moderate to severe sensorineural hearing loss (pure tone average 35 to under 80 dB HL) and a Clinical Dementia Rating (CDR) global score of 0.5 were randomized 1:1. One group (n=353) received free binaural Widex Evoke hearing aids with audiologist fitting. The other (n=350) received a structured hearing care and lifestyle education program matched for contact time.
The primary outcome, conversion to dementia-level impairment (CDR of 1 or higher) at 24 months, occurred in 3.09% of the hearing aid group and 4.74% of the education group (risk difference −2.07%, 95% CI −4.74 to 0.61; relative risk 0.59, 95% CI 0.25-1.38; P=.23). This was not significant. The trial assumed event rates of 10% versus 18%, so actual rates were far lower than planned and the study had limited power to detect the observed difference.
Prespecified secondary outcomes pointed the other way. Cognitive improvement, meaning a shift from CDR 0.5 to 0, occurred in 15.34% of the intervention group versus 2.36% of controls (risk difference 12.00%, relative risk 5.94, 95% CI 2.46-14.37). Mean CDR global scores were lower (better) with hearing aids (−0.062, 95% CI −0.094 to −0.029, P<.001), driven mainly by the memory domain (−0.066, P<.001). Other domains did not differ significantly. Results held in per-protocol, complete-case and inverse-probability-weighted analyses, and across subgroups.
Adherence was good: average hearing aid use was 5.9 hours daily, and pure tone average improved from 52.65 to 32.18 dB HL. No study-related adverse events occurred, and 20 unrelated serious adverse events were split evenly between groups. For clinicians, the findings support treating hearing loss in MCI for its own sake, with a possible cognitive upside.
Several caveats apply. Participants and audiologists could not be blinded, and cognitive improvement was a secondary, exploratory outcome based on a CDR score that can fluctuate near 0.5. Attrition was higher in the education group (60 versus 33 missing at 24 months), requiring multiple imputation of 89 data points. Baseline hearing and diabetes prevalence differed slightly between arms. The cohort was drawn from one city and was 64% male. Longer follow-up is needed to show whether the improvements translate into delayed dementia.
Key Findings
- Primary outcome not met: dementia-level conversion at 24 months was 3.09% with hearing aids vs 4.74% with education (RR 0.59, 95% CI 0.25-1.38; P=.23)
- Cognitive improvement (CDR 0.5 to 0) occurred in 15.34% vs 2.36% of controls (RD 12.00%; RR 5.94, 95% CI 2.46-14.37; P<.001), a prespecified secondary outcome
- Mean CDR global score was 0.45 vs 0.51 at 24 months (difference −0.062, 95% CI −0.094 to −0.029), with the memory domain driving the effect (−0.066, P<.001)
- Orientation, judgment, community affairs, home and hobbies, and personal care domains showed no significant between-group differences (P=.08 to .92)
- Per-protocol analysis: dementia-level impairment 1.89% vs 3.16%; cognitive improvement 15.46% vs 1.40% (RR 8.91, 95% CI 3.24-24.53)
- Hearing aid users wore devices a mean of 5.90 hours daily, and pure tone average improved from 52.65 to 32.18 dB HL
- Follow-up was completed by 86.77% of 703 participants; no study-related adverse events, and 20 unrelated serious events split 10 per group
Methodology
CHOICE was a multicenter, open-label, parallel-group RCT in Shanghai that randomized 703 adults aged 60 or older with MCI (CDR 0.5) and moderate to severe sensorineural hearing loss 1:1 to binaural hearing aids or hearing care education, with follow-up at 6, 12 and 24 months. Cognitive assessors and the statistician were blinded, and hearing aids were removed before cognitive testing. Primary analysis was by intention to treat with multiple imputation (20 imputations) for missing data, using generalized linear models and modified Poisson regression adjusted for recruitment source. Sensitivity analyses included per-protocol, complete-case and inverse probability censoring weighting datasets. The trial was registered as ChiCTR2000036139.
Study Limitations
The authors note that the open-label design meant participants and audiologists were unblinded, and that the primary analysis was likely underpowered because event rates (3.1% and 4.7%) were far below the 10% and 18% assumed. Cognitive improvement was a secondary, exploratory outcome, attrition was uneven (33 vs 60 missing at 24 months, requiring imputation), and the MCI criterion applied before randomization was not prespecified in the registration. The single-region Chinese population and 24-month horizon limit generalizability. Funding and conflict-of-interest details were not in the supplied text, although hearing aids were provided free of charge to participants.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: