# The anticholinergic medication index and dementia risk: evidence from the UK Biobank and All of Us research program

## TL;DR

This study shows that higher use of anticholinergic drugs is linked to increased dementia and mortality risks in two large groups of people.

## Contribution

The study externally validates the Anticholinergic Medication Index (ACMI) as a potential tool for dementia risk prediction.

## Key findings

- Baseline use of ACMI-listed drugs was associated with increased dementia risk in both UK Biobank and All of Us cohorts.
- Higher ACMI scores correlated with increased mortality in both study populations.
- The genetic influence of APOE on dementia risk was replicated across both cohorts.

## Abstract

Anticholinergic drugs are associated with adverse effects, including cognitive decline. In this study, we externally validated the Anticholinergic Medication Index (ACMI) by investigating the association between baseline anticholinergic burden and dementia risk in two large prospective cohorts: the UK Biobank (UKB) and the US All of Us (AoU) program.

We analysed data from the UKB (n = 125 260; study period, 2000–15) and AoU (92 047; 2000–22). Cox proportional hazards models, adjusted for clinical and genetic covariates with death as a competing risk, assessed the association between baseline annual ACMI-computed anticholinergic burden and dementia risk. Exploratory genetic analyses included candidate gene analysis of acetylcholine signalling pathway genes in the UKB and development of a polygenic hazard score in AoU.

Prescription of any of the 88 ACMI-listed drugs at baseline was associated with increased dementia risk (UKB HR: 1.15, 95% CI 1.09–1.21; AoU: 1.06, 1.04–1.09) and mortality (UKB: 1.23, 1.19–1.27; AoU: 1.16, 1.13–1.19). We replicated the genetic influence of APOE on dementia risk (UKB [APOE ε4 vs ε3 carriers] HR: 2.05, 1.86–2.26; AoU: 1.61, 1.44–1.80). No significant gene–drug interactions were identified.

Higher baseline ACMI scores were associated with increased risks of dementia and mortality in two large cohorts, providing external validation of the ACMI. While causal inference is not possible, these findings support the ACMI’s potential utility as a prognostic tool for risk stratification and for informing future work on safer prescribing.