Lisbon 2017 Abstract Poster

Predictors of heroin abstinence in opiate substitution therapy in heroin-only users and dual users of heroin and crack

Felicia Heidebrecht, Mary Bell MacLeod and Lynne Dawkins


Opiate substitution therapy (OST) for heroin dependence is recognized across the world as effective for retention in treatment and reducing illicit drug use. The current UK drug strategy based on the recovery model has resulted in a shift in treatment aims from harm reduction towards complete abstinence. Therefore abstinence from illicit drug use while on OST becomes an equally relevant treatment outcome.
Traditionally, studies have focused on heroin-only users. The high prevalence of crack/cocaine use among heroin users accessing OST in the UK however has been documented for over a decade, and national statistics show an increase in the percentage of dual users (relative to all heroin users). Although more recent studies have recognised that dual users have poorer treatment outcomes when compared to heroin-only users, no specific recommendations are made in the UK guidelines on clinical management.
The need for systematic studies of the patterns of concurrent cocaine use alongside heroin has been recognized, however the majority of studies have considered dual users as a homogeneous group.

Research Question

To analyse predictors of heroin abstinence in OST based on frequency of crack use and its interactions with other predictors in a clinical non-experimental setting.


Design: Retrospective study.
Setting: A community drug service in London, UK.
Participants: 325 clients starting OST between 2010 and 2014 (197 methadone and 128 buprenorphine).
Measurements: Logistic regression models (a general model and separate models for methadone and buprenorphine) assessed demographic and clinical data as predictors of heroin abstinence at one year after treatment start (or at the date of transfer to another service).


Most of the significant predictors in the general model were found significant only in the buprenorphine but not in the methadone model, suggesting that a general model has little predictive value.
For the methadone model only daily crack use predicted heroin use at follow up (OR = 2.62, 95% CI: 0.96 – 7.16).
For buprenorphine, higher amounts of baseline heroin use, lower buprenorphine dose and daily drinking predicted heroin use at follow up (OR = 0.85, 95% CI: 0.75–0.95; OR = 1.31, 95% CI: 1.06–1.60 and OR = 6.04, 95% CI: 1.26–28.92). Use of cannabis increased likelihood of heroin abstinence for clients not using crack compared to occasional (OR=6.68, 95% CI: 0.37–119.59) and daily (OR=57.49 (95% CI: 2.37–1396.46) users. Similarly, the experience of depression increased the likelihood of heroin abstinence for non-crack users compared to occasional (OR=106.31, 95% CI: 3.41–3313.30) and daily OR=170.99 (95% CI: 4.61–6339.47) crack users.


Crack use was a significant predictor of heroin abstinence at follow up in all models, however for buprenorphine only when depression or cannabis use was present. Further research is needed to assess effective treatment approaches for the growing population of dual users.