Drug prescriptions and dementia incidence: a medication-wide association study of 17,000 dementia cases among half a million participants

Tim Wilkinson, Christian Schnier, Kathryn Bush, Kristiina Rannikmae, Ronan Lyons, Stuart McTaggart, Marion Bennie, Cathie L M Sudlow

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

Background: Previous studies have suggested that some medications may influence dementia risk. We conducted a hypothesis-generating medication-wide association study to systematically investigate the association between all prescription medications and incident dementia.

Methods: We used a population-based cohort within the Secure Anonymised Information Linkage (SAIL) databank, comprising routinely-collected primary care, hospital admissions and mortality data from Wales, UK. We included all participants born after 1910 and registered with a SAIL general practice at ≤60 years old. Follow-up was from each participant’s 60th birthday to the earliest of dementia diagnosis, deregistration from a SAIL general practice, death or end of 2018. We considered participants exposed to a medication if they received ≥1 prescription for any of 744 medications before or during follow-up. We adjusted for sex, smoking and socioeconomic status. The outcome was any all-cause dementia code in primary care, hospital or mortality data during follow-up. We used Cox regression to calculate hazard ratios and Bonferroni-corrected p-values.

Results: Of 551,344 participants, 16,998 (3%) developed dementia (median follow-up was 17 years for people who developed dementia, 10 years for those without dementia). Of 744 medications, 221 (30%) were associated with dementia. Of these, 217 (98%) were associated with increased dementia incidence, many clustering around certain indications. Four medications (all vaccines) were associated with a lower dementia incidence.

Conclusions: Almost a third of medications were associated with dementia. The clustering of many drugs around certain indications may provide insights into early manifestations of dementia. We encourage further investigation of hypotheses generated by these results.
Original languageEnglish
JournalBritish Medical Journal (BMJ)
Issue number3
Publication statusPublished - 27 Oct 2021


Dive into the research topics of 'Drug prescriptions and dementia incidence: a medication-wide association study of 17,000 dementia cases among half a million participants'. Together they form a unique fingerprint.

Cite this