Trajectories of heroin use: 10–11-year findings from the Australian Treatment Outcome Study

Aims: To identify trajectories of heroin use in Australia, predictors of trajectory group membership and subsequent outcomes among people with heroin dependence over 10–11 years. Design: Longitudinal cohort study. Setting: Sydney, Australia. Participants: A total of 615 participants were recruited between 2001 and 2002 as part of the Australian Treatment Outcome Study (66.2% male; mean age 29 years). The predominance of the cohort (87.0%) was recruited upon entry to treatment (maintenance therapies, detoxification and residential rehabilitation), and the remainder from non‐treatment settings (e.g. needle and syringe programmes). This analysis focused upon 428 participants for whom data on heroin use were available over 10–11 years following study entry. Measurements: Structured interviews assessed demographics, treatment history, heroin and other drug use, overdose, criminal involvement, physical health and psychopathology. Group‐based trajectory modelling was used to: (i) identify trajectory groups based on use of heroin in each year, (ii) examine predictors of group membership and (iii) examine associations between trajectory group membership and 10–11‐year outcomes. Findings: Six trajectory groups were identified [Bayesian Information Criterion (BIC) = –1927.44 (n = 4708); −1901.07 (n = 428)]. One in five (22.1%) were classified as having ‘no decrease‘ in heroin use, with the probability of using remaining high during the 10–11 years ( > 0.98 probability of use in each year). One in six (16.1%) were classified as demonstrating a ‘rapid decrease to maintained abstinence’. The probability of heroin use among this group declined steeply in the first 2–3 years and continued to be low ( < 0.01). The remaining trajectories represented other fluctuating patterns of use. Few baseline variables were found to predict trajectory group membership, but group membership was predictive of demographic, substance use and physical and mental health outcomes at 10–11 years. Conclusions: Long‐term trajectories of heroin use in Australia appear to show considerable heterogeneity during a decade of follow‐up, with few risk factors predicting group membership. Just more than a fifth continued to use at high levels, while fewer than a fifth become abstinent early on and remained abstinent. The remainder showed fluctuating patterns.