Effectiveness of co-prescribed injectable heroin versus methadone alone treatment

7A.1 Treatment response after 12 months

7A.1.1 Treatment response after 12 months in the intention-to-treat population

As described in paragraph 2.9.3, patients were considered as responders if they showed at least 40% improvement at the month 12 outcome assessment, compared to their situation at baseline, in at least one of the areas in which they functioned poorly at the start of the study (i.e. on the basis of which they were included at baseline), while these improvements should not have gone at the expense of similar (40% or more) deterioration in functioning in any of the other outcome domains. The primary study question, into the effectiveness of co-prescribed heroin versus methadone alone treatment after 12 months, was investigated by testing the difference in percentage responders between the experimental and control condition in the intention-to-treat population, using a logistic regression model, with "treatment site" as the only covariate, and applying the last observation carried forward method to deal with missing endpoint-assessments.

Figure 10. Effectiveness of co-prescribed injectable heroin versus methadone alone treatment (n =174)

In the heroin group (1), the percentage responders amounted to 56.6%, compared to 31.6% in the methadone group (see Figure 10: left-hand bars). Controlling for differences in response between the treatment sites, the difference of 25.0% corresponded with an adjusted Odds-Ratio of 2.99 (95%-CI: 1.58-5.65; p=0.0008). Neither "treatment site", nor the treatment-by-site interaction were significantly related to treatment response in the logistic regression model. As indicated by the Hosmer and Lemeshow goodness-of-fit test, the data fitted the regression model (X²=2.84; df=7; p=0.90).

To illustrate the clinical meaning of the observed response rates, Table 9 shows the changes in mean scores from baseline to month 12 on the scales and items used as components of the multi-domain outcome index of response in the trial.

Table 9. Changes in patient status from baseline to month 12 (mean scores)

Among the responders (right-hand column), considerable changes from baseline to month 12 were observed for most scales and items. For example, the mean MAP-HSS score decreased in the methadone alone group (A) from 13.0 at baseline to 7.6 at the month 12 assessment, and in the co-prescribed heroin group (B) from 12.5 at baseline to 5.9 at month 12. The mean score of 5.9 on the MAP is quite comparable to the mean score of 4.8 found for the general population in the Netherlands (see paragraph 2.9.1). Similarly, the SCL-90 total scores among responders decreased to mean values (40.3 and 38.8), comparable to those found in the general Dutch population (males: 27.2; females: 38.9). In the social domain, a decrease from baseline to month 12 was observed for both the number of days of illegal activities, and - to a lesser extent - for the number of days without personal contact with non-drug-users. A similar decrease occurred for number of days of cocaine use among the responders in the heroin group (B). Amphetamine use at baseline was rare (10.9%), and among the responders limited to an average of less than one day a month.

Among the non-responders (middle column), on the other hand, changes from baseline to month 12 were absent on nearly all scales and items. The only exception occurred with regard to the number of days of illegal activities in the previous month. This variable showed a decrease not only among the responders in both treatment groups, but also among the non-responders in the co-prescribed heroin group.

Robustness of the findings
At the month 12 assessment-point, data were obtained from 97.4% of the patients in the heroin group, and from 89.8% in the control group. To investigate the robustness of the observed 25% difference in percentage responders between the two groups, the (2.6%) missing endpoint-assessments in the heroin group were considered as non-responders, and those in the control group (10.2%) as responders. In this most conservative scenario, treatment response decreased from 56.6% to 55.3% in the heroin group, and increased from 31.6% to 39.8% in the methadone group (see Figure 10: middle bars). Hence, the difference in response decreased from 25.0% to 15.5%. This difference was still significant in the logistic regression model, with an adjusted Odds-Ratio for treatment group of 1.93 (95%-CI: 1.04-3.58; p=0.037), and no significant treatment-by-site interaction. Also in this analysis, the data fitted the model (X²=1.27; df=7; p=0.99).

Anticipation effects
In the follow-up phase of the study, the treatment with heroin could only be reinstated in patients who had responded to the heroin treatment at the end of the experimental phase. To investigate the influence of the anticipation of such a reinstatement (see paragraph 3.2.6) on the results of the study, the month 12 analysis described above was similarly conducted for the month 10 assessment-point. After 10 months, the percentage responders in the heroin group, utilizing LOCF, amounted to 48.7%, and in the methadone group to 28.6% (see Figure 10: right-hand bars). As in the main analysis of the month 12 effects, the difference in percentage responders between the treatment groups of 20.1%, when controlled for "treatment site", and given non-significant relationships between treatment site and response, and between the treatment-by-site interaction and response, was significant at the month 10 assessment-point (adjusted OR=2.46; 95%-CI: 1.30-4.65; p=0.0056). Again, the goodness-of-fit test indicated that the data fitted the regression model (X²=0.50; df=7; p=0.99).

(1) For reasons of readability, the term "heroin group" is used in this chapter to refer to the treatment group receiving co-prescribed heroin and methadone, whereas the term "methadone group" refers to the methadone alone treatment group.