To assess the association between use of benzodiazepines (including the Z-drugs zopiclone and zolpidem) and cardiovascular mortality in women aged over 50 years.
We used data from the E3N cohort. Data self-reported in biennial questionnaires was matched with drug reimbursement and vital status/causes of death data. In Cox models, exposure to benzodiazepines was fitted using time-varying variables, the reference category being women with no benzodiazepine delivery since January 2004.
Among 85,353 women born 1925-1950 and followed between 2004 and 2011, 506 cardiovascular deaths occurred. Exposure to benzodiazepines was associated with increased cardiovascular mortality when hazard ratios (HRs) were adjusted only for age (HRever use 1.65; 95% CI 1.39, 1.97), but not when further adjusted for antidepressant use (HRever use 1.15; 95% CI 0.94, 1.40), nor in the multivariable model (HRever use 0.93; 95% CI 0.75, 1.16). Exposure to hypnotic benzodiazepines remained associated with increased cardiovascular mortality (HRever use 1.23; 95% CI 1.01, 1.51), but with no consistent trend across duration/dose or time since last use, while exposure to anxiolytic benzodiazepines was not (HRever use 0.83; 95% CI 0.67, 1.02).
In our study, adjustment for antidepressant use strongly attenuated any benzodiazepine-cardiovascular mortality association. Whether the modest association observed with hypnotic benzodiazepines is due to residual confounding deserves further investigation.