RESUMO
AIMS: To estimate the annual incidence of heroin use in Spain. PARTICIPANTS AND DESIGN: Data on individuals' year of first heroin use (from 1971 to 2005), year of first heroin treatment between 1991 and 2005 and most frequent route of heroin administration when presenting to treatment were obtained from the Spanish Drug Observatory Register and used to calculate the delay between onset and treatment. By using a log-linear model approach it was possible to correct for missing observations (heroin users who presented for treatment before 1991 and those who had still not presented by the end of 2005) and to estimate heroin incidence over time. FINDINGS: The estimated incidence of problematic heroin use in the population aged 15-44 peaked at 190 per 100,000 in 1980--after rising rapidly from less than 40 per 100,000 in 1971--and fell subsequently to about 8 per 100,000 in 2005. On average, incidence was five times higher in men. Injecting heroin incidence peaked and declined rapidly from 1980; as heroin smoking did not decline as rapidly, from 1985 onwards its estimated incidence has remained above that of heroin injecting. The delay between starting heroin use and entering treatment had a median of 3 years. CONCLUSIONS: We demonstrate the utility of a method to estimate heroin incidence from analysis of observed trends in presentations at specialist drug treatment facilities. The estimates suggest that incidence of heroin use, especially injecting, has fallen since 1980 and is now lower than in the early 1970s.