Background: A nationwide register-based cohort study from Finland including 48 124 incident benzodiazepines and related drug (BZDR) users aged 18-65 years who initiated use in 2006 and were not dispensed BZDRs during 2004-2005. The follow-up was 5 years or until death, whichever occurred first.

Aims: To investigate sociodemographic and clinical factors associated with high-dose use of BZDRs (i.e. Z-drugs) among new BZDR users.

Method: The temporal BZDR dose was calculated as a point estimate every 6 months after initiation as defined daily doses (DDDs) per day, based on the PRE2DUP method (an approach based on mathematical modelling of personal drug purchasing behaviours). Sociodemographic and clinical factors associated with dose categories were studied using multinomial logistic regression.

Results: During the 5-year follow-up, very high-dose BZDR use was observed in 7.4% (n = 3557) and medium high-dose use in 25.5% (n = 12 266) of the users (corresponding to >= 30 mg and 10-29 mg in diazepam equivalents, respectively). Very high-dose use was more common among men compared with women (10.9% versus 4.6%). Very high-dose use patterns were especially observed in younger age groups (18- to 25-year-olds). Compared with oxazepam, initiating BZDR use with clonazepam (adjusted odds ratio 3.86, 95% CI 3.24-4.60), diazepam (2.05, 1.78-2.36) or alprazolam (1.76, 1.52-2.03) was associated with increased odds for very high-dose use. Both medium high-dose and very high-dose BZDR use were associated with a lower level of education. In all, 58% of very high-dose use occurred in BZDR users who received their first prescription from general practitioners.

Conclusions: Clinicians should be aware of the dose escalation risk especially when prescribing diazepam, alprazolam or clonazepam for psychiatric indications. If BZDRs are needed, our findings suggest favouring oxazepam.