Predicting chronic benzodiazepine use in adults with depressive disorder: Retrospective cohort study using...

To identify predictive variables of incident chronic benzodiazepine (BZD) use that could be assessed by prescribing physicians.

Retrospective cohort study using public health and drug insurance administrative data.

Quebec.

New adult BZD users from January 1, 1999, to March 31, 2006, with a diagnosis of depressive disorder in the previous year were included. Chronic BZD use was defined as BZD availability at least 50% of the days between day 181 and day 365 following initiation.

Potential associations between chronic BZD use and age; sex; drug insurance status; recent hospitalization; comorbidity; presence of chronic pain; use of psychotropic medication; mental health diagnoses; number, type, and duration of BZDs prescribed; and the prescribing physician's specialty.

Selection led to an exhaustive cohort of 13 688 patients aged 18 to 64 years, and 3683 aged 65 and older. For the 18 to 64 age group, the combination of disability insurance and more than 1 BZD increased the proportion of chronic users from 14.4% to 53.4%. For patients 65 and older, the main correlates of chronic BZD use included claiming more than 1 BZD (adjusted odds ratio 2.24, 99% CI 1.65 to 3.06) and recent hospitalization (adjusted odds ratio 1.70, 99% CI 1.38 to 2.10). Recently hospitalized older patients with a prescription duration of less than 8 days were the highest-risk group identified (57.8%).

Physicians should be aware that patients are more likely to become chronic BZD users if they receive disability insurance or following a hospitalization. Combination of BZDs is a potentially problematic practice that could be increasing the risk of chronic use.