The two secondary objectives of the study, which are the subject of the present report, concerned a comparison of the effects of co-prescribed heroin given for six months and 12 months duration, and an evaluation of the effects of the discontinuation of co-prescribed heroin after six and 12 months of treatment. The statistical analyses of these secondary study questions are described in the following sections. The relevant assessment-points for the secondary study questions are outlined in Figure 5.

Figure 5. Primary and secondary assessment-points

Treatment effects after six months
The effectiveness of treatment with co-prescribed heroin after six months was analyzed in two ways. The first type of investigation pertained to the question whether the 12 months effects could already be achieved after six months. To this end, the percentage responders was compared between the treatment conditions B and C at the time of the month 12 assessment (in Figure 5: [10] vs. [16] ). Response was defined in a similar manner in condition C as in condition B (i.e. based on the degree of change in the status of the patient between the baseline and month 12 assessment). Since condition C was absent in the injectable heroin trial, this investigation was conducted in the inhalable heroin trial only. Analogous to the primary analysis described in the previous paragraph, the difference in percentage responders between conditions B and C was analyzed on an intention-to-treat basis, by means of a logistic regression model, with treatment site as the only covariate, and by applying LOCF for the missing month 12 outcome assessments. As in the analysis of the 12 months effects, the homogeneity of the treatment effect across the treatment sites was investigated by incorporating a treatment-by-site interaction term in the regression model.
An additional analysis focussed on the six months effects ( [10] vs. [16] ) among the treatment completers in groups B and C, utilizing the same definition for treatment completers as in the analysis of the primary study question.

The second type of investigation of the six months effects consisted of a comparison of the percentage responders between conditions A and B at the time of the month 6 assessment (in Figure 5: [3] vs. [9] ). In this case, the underlying study question pertained to the six months effectiveness of ongoing treatment with co-prescribed heroin, compared to that of ongoing methadone treatment. This investigation - utilizing a similar analysis model as in the first type of investigation - was conducted for both trials in the study.

Effects of discontinuation
The effects of discontinuing the experimental treatment with co-prescribed heroin after 12 months were investigated descriptively, by determining the percentage responders in condition B who had subsequently deteriorated considerably - i.e. at least 20% of the baseline value - on at least one of the outcome domains on which the patient had responded, two months after termination of the heroin treatment (assessment [11] ). This analysis was conducted only for treatment completers who were responders at the 12 months assessment. In the inhalable heroin trial, the percentage responders who had subsequently deteriorated was compared between subjects in conditions B and C ( [11] vs. [17] ), to gain insight into differential effects of discontinuing the heroin treatment after six and 12 months.