Authors' reply

Drs Al-Adwani & Nahata raise an important issue when evaluating the outcome of maintenance treatment, namely how to evaluate the retention rate in an unmasked trial. The special incentive for patients randomised to methadone treatment was the option to switch to the heroin group after completing 1 year of treatment. Since retention is considered one of the main outcome measures for maintenance treatment, our trial shows that heroin-assisted treatment has two advantages: it reaches a higher number of potential patients (percentage initiating treatment) and the retention rate of those initiating treatment is significantly higher (68.3 v. 56.3%, log rank χ²=14.1, P<0.001). Therefore, it is incorrect to say that ‘retention rates would possibly differ insignificantly’: the difference is certainly less, but still significant.