Is Heroin-Assisted Treatment Effective for Patients with No Previous Maintenance Treatment? Results from a German Randomised Controlled Trial

Background/Aims: Until now, the medical prescription of diamorphine (heroin) has been suggested as suitable for patients who have failed previous maintenance treatments. The aim of this paper is to assess the effects of diamorphine on opioid-dependent patients with no previous maintenance treatment experience (NPME). Methods: The German heroin trial compared diamorphine versus methadone maintenance treatment and included 107 patients with NPME. This paper is a sub-analysis of these patients. Results: When comparing this subsample with the rest of the participants in the study, large baseline differences were found, showing a more severe drug use profile in patients with NPME. However, no differences were found in terms of treatment outcome and treatment retention. In the subsample with NPME, outcome measures on the reduction of illicit drug use were significantly better under diamorphine compared to methadone treatment, while there was no difference in health outcomes. Conclusion: Controlled studies are now necessary to examine whether diamorphine treatment could be considered as one of several options in treating severely opioid-dependent patients, regardless of previous maintenance treatment experience.

criterion for the Dutch randomised controlled trials, in which HAT was designed to be an add-on treatment to MMT [11] . Inclusion into the Spanish controlled trial required 2 previous MMT attempts [12] . Inclusion into the Canadian study required at least 1 previous episode of maintenance treatment with an adequate dose for at least 30 days [16] . Inclusion into the ongoing British Randomised Injecting-Opioid Treatment Trial requires continuous methadone treatment for at least 6 months while continuing the injection of heroin on a regular basis [17] . The recently initiated Belgian HAT trial also requires patients to have had 2 failed attempts at MMT or to presently be a non-responder to MMT [18] . Previous abstinence-based treatment was found to be a predictor of effectiveness in the Dutch study [19] , but no other influence of previous treatment experience on outcome measures has been reported in other trials.
In contrast to these studies, the German heroin trial required patients to be nonresponders to MMT or to have undergone 2 unsuccessful previous addiction treatment attempts (mainly maintenance, but also outpatient or inpatient abstinence-based treatment [20] ). Therefore, patients could be included with 2 previous unsuccessful treatment attempts in abstinence-based programmes with no previous maintenance treatment experience (NPME). The aim of this study is to assess the efficacy of diamorphine versus methadone treatment for patients with NPME, in order to clarify whether previous maintenance treatment should remain an inclusion criterion for HAT.

The German Heroin Trial
In a randomised controlled trial, HAT and MMT were compared in a multicentre study among 1,015 patients in 7 cities in Germany. This intent-to-treat sample was the result of the randomisation of 1,032 heroin-addicted patients fulfilling the inclusion criteria and attending examination from a previous screening of 2,038 patients. Recruitment was stratified into 2 target groups: (1) methadone non-responders and (2) patients not in treatment for the last 6 months but with 2 previous treatment attempts (either abstinence-based or maintenance; for details, see Haasen et al. [13] ). Patients were randomised into 4 subgroups, depending on the type of medical treatment (HAT or MMT) and psychosocial care (psychoeducation plus individual counselling or case management plus motivational interviewing) received. Heroin or methadone was dispensed over a 12-month period. HAT patients received an individually adjusted maximum of 3 doses of intravenous diamorphine per day with an additional 60 mg (maximum) of oral methadone when needed, while MMT patients received 1 individually adjusted single dose of oral methadone daily. Long-term effects have been analysed for HAT patients who continued treatment for another 12 months [21] .

Measures
Information used in this study included: (1) Self-reported information on drug use and criminal activity, according to the German version [22] of the EuropASI [23] , based on the fifth edition of the Addiction Severity Index [24] . (2) Psychopathology, measured with the health scale and Global Severity Index of the Symptom Checklist-90-Revised [25] . (3) Health status measured with the Opiate Treatment Index Health Symptoms Scale [26] . (4) Urine samples for determining heroin and cocaine use. (5) Mental and physical health improvement as well as a reduction of illicit drug use (difference between baseline and 12 months). This information was used to determine dichotomous outcome measures (see Haasen et al. [13] for details).

Subjects
In order to carry out this study, the intent-to-treat sample (n = 1,015) was divided into two subsamples: patients with PME (n = 899) and patients with NPME (n = 107). Nine patients were not included in the analysis due to missing data on their previous maintenance treatment.

Statistical Analysis
Baseline characteristics were compared using t tests for continuous variables and 2 tests for nominal variables between patients in HAT or MMT and subsamples with PME or NPME. As noted in our previous publication [13] , this trial demonstrated the stronger effect of HAT on treatment outcomes without the influence of other factors, such as target group (cases of methadone treatment failures vs. cases not in treatment), type of psychosocial intervention (psychoeducation vs. case management) and study site. As demonstrated by Blanken et al. [19] in the Dutch study, differences between MMT and HAT were only significant in patients who had previously undergone abstinence-orientated treatment. Therefore, a three-factorial logistic regression model was used to assess the possible effect of previous maintenance treatment, controlling for type of medication and previous abstinenceorientated treatment. Additionally, 2 and odds ratio were calculated separately in the PME and NPME subsamples in order to assess the differential effect of the type of medication on outcome measures. The same analyses were carried out within groups of patients with and without previous experience with abstinenceorientated treatment. A one-factor ANCOVA was carried out in the NPME subsample between treatment groups for illegal activities in the last 30 days at end of treatment controlling for baseline data. The Cl was set at 95%. Analyses were made using SPSS version 15 for Windows. Table 1 shows the baseline characteristics of the participants. Between the 2 medication groups in the subsample with NPME, there was only 1 significant difference: HAT patients had a higher proportion of hepatitis C infections ( 2 = 7.308, p = 0.007).

Participant Characteristics
Compared to patients with PME at baseline, NPME patients included fewer females ( 2 = 7.046, p = 0.008), were younger (t = 2.292, p = 0.022), had a poorer housing situation ( 2 = 6.861, p = 0.009), had less experience with detoxification ( 2 = 9.266, p = 0.002) and drug-free treatment ( 2 = 16.304, p ! 0.0001), had fewer years of heroin (t = 5.306, p ! 0.0001) and cocaine use (t = 1.888, p = 0.059) and had more days of heroin use in the last month (t = -12.497, p ! 0.0001). They had also injected drugs (t = -6.224, p ! 0.0001) on more days in the last month, but the proportion of patients infected with hepatitis C virus was lower ( 2 = 14.586, p ! 0.0001), which corresponds to fewer years of heroin use. Finally, those in the NPME subsample had a lower rate of suicide attempts ( 2 = 12.102, p = 0.001) and a higher score on the Global Assessment of Functioning Scale (t = -2.008, p = 0.045).

Treatment Retention
Treatment retention among NPME patients (53.84%) did not differ from that found among PME patients (53.27%). In contrast with the results of the main study, no significant differences were found with regard to treatment retention between HAT (55.93%) and MMT (50.00%) in the NPME subsample ( 2 = 0.374, n.s.). No significant differences were found in terms of treatment duration between HAT (276 days) and MMT (244 days) groups in the NPME subsample (t = 1.12, n.s.

Treatment Effectiveness
The three-factorial logistic regression model analysis showed no influence of PME on primary outcome measures (POM) of health, illegal drug use or both when analysing for treatment group. Previous abstinence treatment experience (PATE) was found to be significant with regard to response in the outcome measure of drug use reduction (OR = 1.434, 95% CI = 1.091-1.884) but not for health improvement (OR = 0.829, 95% CI = 0.603-1.140).
When analyzing subsamples separately, no influence of treatment on health was found in NPME patients in contrast with PME patients, but a significantly greater response from HAT patients was found in the POM of illicit drug use ( table 2 ). The more rigorous criteria of response in both outcome measures also shows a greater response from HAT patients. Table 2 shows the relationship between treatment effectiveness and PATE. Among patients with PATE, a significant influence of treatment in reduction of drug use, health improvement and both outcome measures together was found. Nevertheless, patients without PATE profited significantly more from HAT than from MMT only in reduction of drug use. With respect to the outcome measure of reduction of illegal activity, the subsample of patients with PME showed no difference, at baseline, between the HAT and MMT groups ( table 1 ), but after 12 months illegal activity was reduced to 0.81 days (of the last 30 days) in the HAT group as compared to 5.56 days in the MMT group (ANCOVA: F = 10.120, p =0.002). Figure 1 shows the course of health indicators. Physical health (as measured with the Opiate Treatment Index Health Symptoms Scale) showed an overall high improvement in the first phase of the study. A slight deterioration can be perceived among MMT patients in the last 6 months of the study, while HAT patients continued to show a stable amelioration of their symptoms. Mental health (as measured with the Global Symptom Index of the Symptom Checklist-90R) showed a parallel course in both subgroups. While HAT patients had a better outcome in the first 3 months, at 6 months, both groups showed similar results. Toward the end of the study, symptoms among MMT patients worsened slightly, while the effects of the therapy remained stable among patients in the HAT group. Figure 2 shows the course of street use of heroin and cocaine according to self-reported data. Although both groups reduced the use of illicit heroin, a greater reduction can be seen in the HAT group. Cocaine use reduction was steady among HAT patients, while MMT patients showed a reduction of cocaine use only during the first few months. These results were confirmed by urinalysis throughout the course of the study ( fig. 3 ). The percentage of cocaine-positive urine samples found during treatment was 32.2% among HAT and 38.9% among MMT patients.

Discussion
HAT has been considered a second-or last-choice intervention. This consideration largely rests on 4 facts: First, injecting bears higher health risks than does oral treatment, so that maintenance treatment is suggested to be initiated with an oral substance. Second, injecting a substance is thought to maintain the craving-related aspects of addiction, which could be avoided with the use of an oral substance. Third, the psychoactive central-nervous effect of diamorphine also upholds craving, which is considered problematic in the long-term treatment of an addictive disorder. Finally, HAT is more expensive than MMT and requires more resources. On the other  hand, HAT might have an advantage over MMT in that it includes patients who would, otherwise, choose not to enter maintenance treatment at all. If this were the case, it would be necessary to evaluate whether HAT is more effective than MMT only for chronic opioid-dependent patients who have been in maintenance treatment before, or also for those without this type of treatment experience. The present study is the first to analyse the effect of HAT in patients with NPME. The results show that patients with NPME, who have a shorter addiction career, benefit from both HAT and MMT to almost the same extent as those with PME. The most important finding is the superior effectiveness of HAT in this subsample, considering the fact that, unlike the rest of the sample, they have had no previous (negative) experience with MMT. This was not found in all outcome measures. While no difference was found in the POM on health, a significant difference became apparent in the reduction of illicit drug use and illegal activity, which are generally considered two main goals of maintenance treatment. In the more rigorous outcome definition of having to respond to both POMs, HAT was also found to be superior to MMT in this subsample.
These findings cannot be explained by a higher dropout rate in the MMT group, as the retention rates did not differ in the 2 treatment groups of patients with NPME. Nonetheless, all 107 patients entered the study with the intention of possibly being randomised into the diamorphine group. Even those patients randomised into the methadone group had the possibility of switching into the diamorphine group after completing 1 year of methadone maintenance. In this way, the attractiveness of HAT may have still played an important role in drawing these patients into maintenance treatment, a path which they previously had not chosen to follow despite the low threshold for entry into maintenance treatment in Germany.
These results, therefore, should lead us to reconsider whether HAT should only be implemented as a secondline treatment, or whether it should be made available to all chronic severely opioid-dependent patients irrespective of PME. Other factors such as PATE seem to be of greater importance for the effectiveness of HAT. However, this study was self-selective and did not control for the factor of PME in an experimental design. In the future, it would be necessary to confirm our results in a controlled trial, as this would have important implications for the scope of this innovative treatment option.