Introduction: Emerging evidence has linked the gut microbiome to cognitive aging. Antibiotic use can rapidly reduce bacterial diversity and lead to dysbiosis that persists for months to years, potentially impairing cognitive function. Numerous studies have reported the associations between chronic antibiotic use and dementia-related diseases like obesity and cardiovascular diseases. However, prospective investigation into the association of antibiotics with dementia risk remains limited.
Methods: We analyzed prescription records from 13,571 community-based Australian adults ≥70 years from ASPirin in Reducing Events in the Elderly (ASPREE), a completed randomized aspirin trial with adjudicated dementia as a prespecified endpoint, and its extension study (ASPREE-XT). We assessed antibiotic use until the second annual visit and excluded participants with dementia at or before the second visit to minimize the possibility of reverse causation. We used multivariable-adjusted Cox models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for incident dementia and cognitive impairment no dementia (CIND) associated with antibiotic use. To assess changes in individual and composite standardized cognitive function scores (global cognition, memory, language and executive function, and psychomotor speed), we used linear mixed models with an interaction term between antibiotic use and the year of cognitive assessment (0, 1, 3, 5, 7, 8, 9) and participant-specific intercept and slope.
Results: Over a mean of 5.5 years of follow-up, we documented 461 incident dementia cases and 2576 incident CIND cases. Compared to non-users, antibiotic users were similar in age, education level, BMI, and cardiometabolic disease status, but were more likely to use other prescription medications concomitantly. Compared with nonuse, antibiotic use was not associated with dementia (HR: 1.01; 95% CI: 0.83-1.24), CIND (HR: 1.01; 95% CI: 0.93-1.10), or changes in cognitive scores over time (Figure 1), after adjusting for a wide range of other sociodemographic and lifestyle risk factors, including age, sex, alcohol use, family history of dementia, and baseline global cognition based on the Modified Mini-Metal State test (3MS). More frequent antibiotic use was not associated with higher risks for dementia and CIND or greater declines in cognitive scores (Figure 2). These null associations persist when assessed by individual antibiotic classes and by categories of other risk factors, including age, sex, and baseline cognitive score (Figure 2).
Conclusion: Among older adults, antibiotic use was not associated with incident dementia and CIND or accelerated declines in cognitive function. Our results do not support an association between antibiotic-associated gut microbiome disruption and dementia risk.

<b>Figure 1. Antibiotic use was not associated with incident dementia and CIND or changes in cognitive function scores. </b>A. Values are in mean (standard deviation) unless otherwise noted (%). B. Hazard ratios and 95% confidence intervals were calculated using Cox models adjusted for age, sex, race, education level, smoking status, alcohol use, BMI, family history of dementia, hypertension status, diabetes status, chronic kidney disease status, CESD-10 score, baseline 3MS score, the total number of concomitant drug use, anticholinergic drug use, psycholeptic drug use, and psychoanaleptic drug use. C. Linear mixed model with an interaction term between antibiotic use and the year of cognitive assessment, participant-specific intercept and slope, and adjustment for the same covariates as the Cox model.

Figure 1. Antibiotic use was not associated with incident dementia and CIND or changes in cognitive function scores. A. Values are in mean (standard deviation) unless otherwise noted (%). B. Hazard ratios and 95% confidence intervals were calculated using Cox models adjusted for age, sex, race, education level, smoking status, alcohol use, BMI, family history of dementia, hypertension status, diabetes status, chronic kidney disease status, CESD-10 score, baseline 3MS score, the total number of concomitant drug use, anticholinergic drug use, psycholeptic drug use, and psychoanaleptic drug use. C. Linear mixed model with an interaction term between antibiotic use and the year of cognitive assessment, participant-specific intercept and slope, and adjustment for the same covariates as the Cox model.

<b>Figure 2. The null associations between antibiotic use and incident dementia and CIND did not differ by the (A) subclasses of antibiotics or (B) other risk factors.</b> Hazard ratios and 95% confidence intervals were calculated using Cox models adjusted for age, sex, race, education level, smoking status, alcohol use, BMI, family history of dementia, hypertension status, diabetes status, chronic kidney disease status, CESD-10 score, baseline 3MS score, the total number of concomitant drug use, anticholinergic drug use, psycholeptic drug use, and psychoanaleptic drug use.

Figure 2. The null associations between antibiotic use and incident dementia and CIND did not differ by the (A) subclasses of antibiotics or (B) other risk factors. Hazard ratios and 95% confidence intervals were calculated using Cox models adjusted for age, sex, race, education level, smoking status, alcohol use, BMI, family history of dementia, hypertension status, diabetes status, chronic kidney disease status, CESD-10 score, baseline 3MS score, the total number of concomitant drug use, anticholinergic drug use, psycholeptic drug use, and psychoanaleptic drug use.

Jonathan Broder

Appearances