Editor—Dehue thinks that our study was a nerve racking test for the patients because considerable sanctions were connected with participants' treatment responses. This is based on the false assumptions that patients would be expelled from the experiment if they deteriorated while receiving heroin and that patients in the control condition would lose their opportunity to enter heroin assisted treatment if they improved during the treatment with methadone alone. Measures on illicit drug use and criminal activities showed excellent agreement with urine analysis and police register data, indicating that patients in the trial were accurate and reliable in their reporting.
Reed et al assume that patients were required to be in continuous methadone maintenance treatment for only four weeks in the previous five years. Study participants were in methadone treatment for around 12 years on average and were using methadone 28 days in the month before the start of the trial. At the end of the trial the methadone dosage in the experimental groups was on average about 10 mg lower than in the control groups, but this difference was neither significant nor clinically relevant. Therefore a difference in methadone dosage between the treatment conditions is unlikely to have accounted for the observed efficacy of medically prescribed heroin.
Craighead finds it disappointing that 45-88% (actual data 45-78) of the participants did not respond to co-prescribed heroin. We believe that 22-55% response in a chronic, treatment resistant population of heroin addicts with very few assets, serious health problems, and massive social impairments is a substantial effect.
We disagree with Caplehorn's conclusion that the study data favour continuation of methadone treatment in this population. Although treatment retention is often a prerequisite for treatment effectiveness, retention can never replace effectiveness data (response) in establishing efficacy. Additional analyses show that the observed efficacy of heroin prescription is not due to some Hawthorn or Honeymoon effect (www.ccbh.nl).
Voth and Aeschbach question the validity and clinical relevance of the primary outcome variable of the study. In the trial, similar results in favour of the heroin assisted treatment were obtained with various other response definitions, including definitions that did not allow any deterioration or increase in cocaine use. Heroin prescription resulted in modest reductions in cocaine use and large reductions in criminality.
With regard to their final statement, we refer to Rehm et al, who showed that after five years only 34% of the patients were still in heroin assisted treatment and that most of the patients who left the heroin programme started another treatment, generally methadone maintenance or abstinence treatment.1 The view that heroin assisted treatment furthers the addiction and enslavement of suffering addicts is therefore not supported by the data. In contrast, the findings indicate that heroin assisted treatment can be a useful addition to the treatment armature available for these suffering, chronically impaired patients.