Background Recent studies have reported conflicting associations between drug prescriptions and incident dementia. Any association between drug and dementia could be due to the drugs directly causing or preventing dementia; the drugs being associated with a risk factor for dementia; or the drugs being prescribed as a consequence of prodromal dementia. Based on methodology developed for genome-wide association studies, we systematically analyzed the effect of 733 drugs on incident dementia in a population-wide linkage study and clinically reviewed the associations.

Methods Using linked, routinely-collected electronic health records from hospital admissions, mortality records and primary care consultations, we followed-up 574,237 Welsh residents from their 60th birthday onwards to classify exposure (drug prescriptions) and dementia incidence. During follow-up, 13,786 (2.4%) of the study population developed dementia. We used time-dependent Cox proportional hazard models to study the effect of exposure on dementia incidence, controlling for the effects of age, sex, year, deprivation and smoking status. To account for multiple testing, we first analyzed a 50% household-area stratified random sample of the study population (discovery cohort), selected results with a Bonferroni-corrected p-value, re-run the analysis of ‘significantly’ associated drugs in the remaining 50% (validation cohort) and once again selected results with a Bonferroni-corrected p-value. We displayed the results (hazard ratio and p-value) from the complete cohort in several stratified volcano-plots and clinically reviewed the findings to identify potential pathways of effect.

Results 177/733 (24%) of the analysed drugs were significantly associated with dementia incidence. Of those, 7 were for neurodegenerative conditions that can cause dementia, 14 were for vascular diseases, 13 for diabetes, 16 for depression and 39 for symptoms or complications of dementia. Only four, all travel-related vaccines, were associated with a lower dementia incidence. Some drugs associated with an increased hazard of dementia clustered around several unexpected indications, including: gastro-oesophageal reflux disease, altered bowel habit, lower urinary tract symptoms and infections, anxiety, sleep disturbance, pain and nausea/vertigo.

Discussion By grouping drugs by indication, we identified several drugs with a potential of having a direct association with increased risk of dementia. We also identified drugs which are related to (known) risk factors for dementia, including those prescribed for cardiovascular disease and diabetes. The effect of travel-related vaccines is puzzling and might be more related to a preventative association of travelling with dementia incidence. Most interestingly, we identified several drugs which might have been prescribed as a consequence of a preclinical, non-cognitive syndrome in dementia.