Association between benzodiazepines and dementia: A case-control study from Canadian health surveys and medico-administrative databases

BACKGROUND: Benzodiazepines (BZDs) are widely prescribed for insomnia and anxiety, but long-term use may accelerate cognitive decline. Evidence is inconsistent because prodromal dementia symptoms can prompt BZD prescriptions, creating reverse-causality bias. OBJECTIVE: Examine whether BZD exposure, duration and elimination half-life are independently associated with incident dementia, and explore confounding by the prodromal period. METHOD: We performed a case-control study in the Torsade Cohort, derived from the Canadian Community Health Survey linked to health-administrative databases. BZD exposure, duration and half-life were analyzed with multivariate conditional logistic regression. Model 1 adjusted for dementia risk factors; Model 2 additionally adjusted for potential BZD indications (insomnia, anxiety, depression). To test prodromal effects, the index date was moved 1-10 years before diagnosis. Cases were adults ≥50 years with dementia; controls were matched on sex, age, follow-up and education. RESULTS: Among 1082 cases and 4262 controls, Model 1 showed that any BZD use was associated with dementia (OR 1.65, 95 % CI 1.42-1.93). Risk was higher with long half-life molecules (OR 2.81) than medium half-life (OR 1.57). In Model 2, chronic use (>180 days) was linked to dementia only within four years before diagnosis. CONCLUSIONS: BZD exposure is associated with increased dementia risk, strongest for long half-life agents. The restriction of chronic-use associations to the four-year prodrome suggests confounding by indication or reverse causality. These findings stress cautious, time-limited BZD prescribing in older adults.