Aims To examine risk factors associated with non-fatal heroin overdose, particularly frequency and route of heroin administration.
Design Data from cross-sectional surveys were analysed as a case–control and as a case cross-over design.
Setting and participants 2556 subjects treated for heroin dependence in 164 outpatient facilities in Spain.
Measurements Prevalence of overdose involving emergency care in the 12 months before treatment admission.
Case control design Odds ratio (OR) adjusted by logistic regression.