Because methadone maintenance treatment has been shown to reduce mortality due to natural causes and overdose in opiate addict populations (Grönbladh et al., 1990; Langendam et al., 2001), it would be unethical to use placebo in the control condition of the trials. Therefore, and since methadone maintenance treatment constitutes the most applied standard reference treatment for opiate addiction in the Netherlands, methadone was used as medication in the control condition. In the experimental condition of the trials, heroin was used as an add-on to methadone (add-on study). The rationale for prescribing methadone in combination with heroin in the experimental condition is that methadone, with its longer pharmacological action, prevents the occurrence of withdrawal symptoms during periods that the prescribed heroin is not available to the patient, e.g. when a patient is not able to visit the treatment site.

Given the nature of the medications under study - co-prescribed heroin and methadone in the experimental condition and methadone alone in the control condition - serious constraints are laid upon the use of the normally favored double-blind randomized, placebo-controlled study design (see also: Bammer et al., 1999). Application of the double-blind procedure - by the addition of a placebo treatment to methadone in the control condition of the trials - is problematic, because the psychopharmacological effects of heroin and methadone - versus placebo - are immediately recognized by opiate users. The blindness of the study would be compromised, because the subjects recognize immediately that they are receiving placebo, because they do not experience a "buzz" or "hit". For these reasons, the application of the double-blind placebo-controlled study design was rejected in this population of subjects, and with this pharmacological compound. The comparison between the control condition and the experimental condition was, therefore, made in a randomized, parallel groups design without blinding (open-label trial).
In order to reduce the risk of information bias, outcome assessments were conducted by independent assessors, who used standardized instruments and evaluation procedures. In addition, the validity of the self-report data was checked through the application of urinalysis, with regard to the concurrent use of illicit drugs, and collection of registered data from the police and justice system, with regard to committed offenses and periods of detention. These latter types of data are insensitive to information bias.