Conductive hearing loss tracks with dementia, which is an argument about mechanism

Design

Cross-sectional multivariable regression analysis of the All of Us Research Program

Population

338,048 adults aged 18 or over, mean age 54 years; 393 with coded bilateral conductive hearing loss, 6,597 with dementia

Primary outcome

Odds of dementia with bilateral conductive hearing loss

Effect

Adjusted OR 2.05 (95 per cent CI 1.18 to 3.32, P = 0.006); 2.03 adding surgical treatment terms, 1.98 adding hearing devices

The reason to study conductive hearing loss and dementia is methodological. Sensorineural loss has a neural component, so an association with dementia can always be read backwards - the disease affecting the ear as well as the brain. Conductive loss is purely mechanical, so an association with dementia cannot be explained that way.

Across 338,048 adults in the All of Us Research Program, 393 had bilateral conductive hearing loss coded and 6,597 had dementia. After adjustment for age, sex, ethnicity, race, education and smoking, the odds of dementia were 2.05 times higher with conductive loss (95 per cent CI 1.18 to 3.32, P = 0.006). Adding surgical treatments - stapedectomy, tympanoplasty, mastoidectomy, myringoplasty - to the model barely moved it (OR 2.03), and adding hearing aids or implants moved it only slightly further (OR 1.98).

The association is real and the mechanistic argument is elegant. What the study cannot do is establish direction or benefit. It is cross-sectional, so nothing tells you which came first; 393 exposed people is a small base for an odds ratio; and the observation that treatment terms barely change the estimate is not evidence that treating conductive loss fails to protect cognition - that would need people randomised or followed, not covariates added. Read it as a reason to take conductive hearing loss seriously as a potentially modifiable exposure, and as a reason to want the trial.

• Do not tell patients that treating conductive loss prevents dementia; this study cannot support that.
• Do take a long-standing untreated conductive loss seriously rather than as a cosmetic complaint.
• Note the base: 393 people with coded bilateral conductive loss out of 338,048.
• Remember the design is cross-sectional, so dementia and hearing loss were measured at the same moment.
• The mechanistic point - that conductive loss cannot be caused by dementia - is what makes this worth reading.

The statistics, in plain English

An odds ratio of 2.05 with an interval from 1.18 to 3.32 is statistically clear but imprecise, because only 393 participants carried the exposure. Cross-sectional design means there is no temporal ordering: dementia may lead to conductive problems going undiagnosed and untreated as easily as the reverse. Adding treatment variables to a regression is not a test of whether treatment helps - it is a test of whether treated status explains away the association, which is a different and much weaker question.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.