Heroin maintenance treatment for chronic heroin-dependent individuals: A Cochrane systematic review of effectiveness

1. Introduction 

The life span of heroin users is dramatically shortened, as shown by mortality studies (Davoli et al., 1997). Causes of death include AIDS, overdose, and other violent causes. Heroin users are exposed not only to the biologic effects of the substance but also to the social consequences of its use. Criminal laws affect heroin availability and purity, and users are often prosecuted for activities related to heroin use at the expense of their social integration (loss of work, family disruption, and difficulties in social relationships). Social and health service offerings vary across countries and cultures. As a consequence, the effects of heroin dependence are heterogeneous over time and between countries (Bargagli et al., 2002).

The similarity of heroin dependence with other health conditions such as diabetes and hypertension (in terms of genetic heritability, individual behaviors, pathophysiology, and treatment response) suggests that these patients would benefit from patterns of treatment similar to those provided to chronic patients, with continuing care and monitoring over time (McLellan et al., 2000, O'Brien, 1997). Treatments aimed at the stabilization of patients over time (maintenance) achieve better results than those aimed at rapid detoxification (Sees et al., 2000). Medications adopted for maintenance include methadone (Faggiano et al., 2003, Mattick et al., 2003), LAAM (Clark et al., 2003), buprenorphine (Mattick, Kimber, Breen, & Davoli, 2003), and heroin (Ferri, Davoli, & Perucci, 2003).

The first experiences with the prescription of heroin to heroin-dependent individuals were in the United Kingdom (Metrebian, Shanahan, & Stimson, 1996). In 1926, the Rolleston Committee (Strang, Ruben, Farrell, & Gossop, 1994) supported the role of physicians in the prescription of opiates for the management of chronic opioid dependence. When, in the 1960s, the population of heroin users rapidly increased, the black market for pharmaceutical heroin became common and some restrictions were introduced (Hartnoll et al., 1980).

Although some clinical trials were conducted in Switzerland, United Kingdom, and the Netherlands, a debate arose (Farrell & Hall, 1998, Venning, 1998, Wodak, 1998) about the opportunity to widen the access to heroin treatment for heroin-dependent individuals. Authorization for similar trials has been obtained in Spain (Bosch, 2001) and Germany. Australia, Canada (Fischer & Rehm, 1997), Denmark, and Luxembourg are considering the implementation of such trials (Cooper-Mahkorn, 1998).

Some researchers have argued that more resources should be devoted to the consolidation of treatment for which reliable evidence is already available, such as methadone (Farrell & Hall, 1998). The general debate focused on some points: do not offer heroin prescription instead of or at the expense of methadone maintenance, do not prescribe heroin to young users or people with a short history of heroin dependence, do not prescribe heroin as a first choice treatment, and minimize the possibility that prescribed heroin will leak into illicit markets.

This review aims at assessing the efficacy of heroin maintenance versus methadone or other substitution treatments for opioid dependence in keeping people in treatment, in reducing the use of illicit substances, and in improving health and social functioning.

2. Materials and methods 

3. Results 

The search resulted in 19 eligible studies (Battersby et al., 1992, Fischer et al., 1999, Ghodse et al., 1990, Haemmig & Tschacher, 2001, Hartnoll et al., 1980, Hendriks et al., 2001, Jasinski & Preston, 1986, Krausz et al., 1999, McCusker & Davies, 1996, Mello et al., 1981, Metrebian et al., 1998, Moldovanyi et al., 1996, Oppenheimer & Stimson, 1982, Perneger et al., 1998, Rehm et al., 2001, Strang et al., 2000, Uchtenhagen et al., 1999, van den Brink et al., 2003). Four met our inclusion criteria and enrolled a total of 696 patients (Hartnoll et al., 1980, Perneger et al., 1998, van den Brink et al., 2003, van den Brink et al., 2002); only 577 patients were enclosed in the present review because one arm of one of the enclosed studies was not considered relevant (in fact, the trial enclosed 375 patients; the third-arm patients were given methadone alone for 6 months followed by 6 months of methadone + heroin). Reasons for exclusion were study design (Battersby et al., 1992, Krausz et al., 1999, McCusker & Davies, 1996, Metrebian et al., 1998, Oppenheimer & Stimson, 1982, Rehm et al., 2001, Uchtenhagen et al., 1999), study outcomes (Ghodse et al., 1990, Haemmig & Tschacher, 2001, Hendriks et al., 2001, Jasinski & Preston, 1986, Mello et al., 1981, Moldovanyi et al., 1996), interventions (Strang et al., 2000), and type of participants (Fischer et al., 1999) not in the inclusion criteria of this review.

The included studies (Table 1) were conducted in different years and countries, which may explain the different criteria for the choice of interventions, outcomes (Table 2), and data reporting. Table 1 also reports on the issues related to methodological quality. All the included studies were described as randomized and none was blinded; lack of blinding was not considered a weakness because it does not seem feasible to blind patients and/or treatment providers to heroin or methadone. In three studies, allocation was concealed (Perneger et al., 1998, van den Brink et al., 2002, van den Brink et al., 2003). Information about people who left the study are reported in all the studies.

Table 1. Systematic review on heroin maintenance treatment for opioid dependence: Characteristics of enclosed studies

	Study	Methods	Participants	Interventions and doses	Outcomes
	CCBH (A) 2002	Multicenter trial (six sites), The Netherlands, July 1998–December 2001; duration of trial: 12 months; allocation: randomized controlled trial; randomization performed centrally; blindness: not performed, all the patients were informed about the treatment to which they had been allocated; quality assessment = A	Diagnosis: heroin-addicted individuals (inhaling use) registered in local methadone maintenance programs who had been prescribed with effective-dose methadone for at least 4 consecutive weeks in the past 5 years and had had regular contact with the methadone maintenance program in the preceding 6 months. N = 375; sex = 79.7% male; mean age = 39.6 years (SD = 5.7 years)	Group A (n = 139): 12 months of methadone; Group B (n = 117): 12 months of heroin (inhalable) + methadone; Group C (n = 119): 6 months of methadone + 6 months of heroin (inhalable) + methadone. Doses: Group A/control (treatment completers), mean methadone dose = 67 mg (SD = 23.2 mg); median = 330 days. Group B/experimental (treatment completers), 57 mg (SD = 18.1 mg); mean heroin dose = 263 mg (SD = 59.0 mg) per visit and 547 mg (SD = 173.5 mg) per day. Group C (treatment completers), second 6-month mean methadone dose = 66 mg (SD = 27.4 mg); mean heroin dose = 252 mg (SD = 68.0 mg) per visit and 539 mg (SD = 151.2 mg) per day.	Treatment completion, treatment response after 12 months (considered as an improvement of at least 40% in the 12-month outcome assessment), compared with their situation at baseline in at least one of the areas in which they functioned poorly: physical health, psychiatric status, social functioning, and substance use, and with no substantial deterioration (>40%) in any of the domains compared with baseline and no increase (>20%) in illegal drug use. The multidomain outcome consisted of MAP-HSS, SCL-90, days of illegal activities, days without personal contact with non-drug users, days of cocaine use, and days of amphetamine use
	UNLISTED: CCBH (B) 2002	Multicenter trial (six sites), The Netherlands, July 1998–December 2001; duration of trial: 12 months; allocation: randomized controlled trial; randomization performed centrally; blindness: not performed, all the patients were informed about the treatment to which they had been allocated; quality assessment = A	Diagnosis: heroin-addicted individuals (intravenous use) registered in the local methadone maintenance programs who had failed several methadone programs. N = 174; sex = 82.2% male; mean age = 38.5 years (SD 5.7 years)	Group A (n = 98): 12 months of methadone; Group B (n = 76): 12 months of methadone + heroin injectable. Doses: Group A (treatment completers), mean methadone dose = 71 mg (SD = 23.6 mg) per day during 336 days. Group B, mean methadone dose = 60 mg (SD = 17.0 mg); mean heroin dose (in the treatment completers) = 254 mg (SD = 62.5 mg) per visit and 549 mg (SD = 193 mg) per day.	Treatment completion, treatment response after 12 months (considered as an improvement of at least 40% in the 12-month outcome assessment), compared with their situation at baseline in at least one of the areas in which they functioned poorly: physical health, psychiatric status, social functioning, and substance use, and with no substantial deterioration (>40%) in any of the domains compared with baseline and no increase (>20%) in illegal drug use. The multidomain outcome consisted of MAP-HSS, SCL-90, days of illegal activities, days without personal contact with non-drug users, days of cocaine use, and days of amphetamine use
	Hartnoll et al., 1980	London, 1972–1975; duration of trial: 12 months; allocation: computer-generated list of random numbers; blindness: patients were aware of the treatment provided but not that they were part of a trial; blindness over interventions by patients and treatment providers and outcomes measurers not mentioned; quality assessment = A	Diagnosis: regular opiate use including daily heroin injection in the last 3 months sufficient to convince clinical staff to enroll patients in a substitution program. N = 96; age range = 18–35 years, mean age = 23.9 years; sex = 75% male/25% female; history: mean duration of opiate use = 5.9 years; age at first use = 18.0 years; criminal activity: 87% had criminal convictions, the mean number of convictions being 4.3; 52% had been convicted at least once in the 12 months before intake; mental state: at intake, 60% of the patients were recorded as mildly anxious, 27% as anxious, 35% as mildly depressed, and 3% as euphoric	Heroin maintenance (n = 44): dose = 30–120 mg/day; oral methadone (n = 52): dose = 10–120 mg/day; interviews by a clinic psychiatrist at 3, 9, and 12 months	Health; use of substances = total opiate consumption (prescribed + illicit); frequency of injection during 12 months; proportion of days spent with other users; crime activity: crime as source of outcome during 12 months; arrests during 12 months; employment
	Perneger et al., 1998	Geneva, Switzerland, September 1995–March 1996; duration of trial: 6 months; allocation: randomized controlled trial; blindness: not described; quality assessment = A	Diagnosis: heroin-addicted individuals (N = 51): age > 20 years; use > 2 years (average = 12 years); more previous attempts of drug treatments (average = 8, range = 2–21) and had experienced four drug overdoses (range = 0–30); high prevalence of mental disorders and health status scores 1–2 standard deviations below population norms	Heroin injected by the patients themselves + oral methadone if a patient travels or wants to reduce attendance in the clinic; mean daily dose of intravenous heroin = 509 mg/day in one to three injections; in addition to heroin, all the patients occasionally received oral opiates and 16 patients received clorazepate substitution therapy (median dose = 60 mg/day); psychological support, n = 27; any other conventional drug treatment, n = 24	Consumption of street heroin; frequency of overdoses; risk behavior for HIV; number of days ill in the past months; use of health services, health status, work status, living arrangements, quality of social relationships, monthly living and drug-related expenditures, sources of income, and criminal behavior

Table 2. Systematic review on heroin maintenance in opioid dependence: Comparison of study treatments and outcome measures

	Outcomes	Treatment
		Injected heroin vs. oral methadone maintenance	Injected heroin vs. waiting lists for other treatments chosen by the patients	Inhaled heroin + oral methadone vs. oral methadone	Injected heroin + oral methadone vs. oral methadone
	Primary outcomes
	Retention in treatment	1	1	1	1
	Relapse to street heroin use (self-reported)	1	1	Part of a multidomain outcome	Part of a multidomain outcome
	Use of other substances	Only a narrative comment	1	Part of a multidomain outcome	Part of a multidomain outcome
	Death	1	1	1	1
	Medical adverse events	No datum	No datum	1	1
	Secondary outcomes
	Criminal offense	1	1	Part of a multidomain outcome	Part of a multidomain outcome
	Incarceration/imprisonment	1	No datum	Part of a multidomain outcome	Part of a multidomain outcome
	Social functioning (integration at work, family relationship)	1	1	Part of a multidomain outcome	Part of a multidomain outcome

Numbers represent the number of studies (among the included studies) that used the treatment and outcome indicated.

3.1. Summarizing results across studies 

Two studies compared injected heroin with oral methadone (270 patients, 1 year) but measured different outcomes; one study compared injected heroin and methadone with oral methadone (51 patients, 6 months); and another one compared inhaled heroin and methadone with oral methadone (256 patients, 1 year). No meta-analysis could be performed because of the heterogeneity of interventions and measures; therefore, the main results are described for each study (Table 3).

Table 3. Systematic review on heroin maintenance in opioid dependence

	Study	Retention in treatment	Relapse to street heroin use (self-reported)	Use of other substances	Death	Medical adverse events	Criminal offense	Incarceration/imprisonment	Social functioning (integration at work, family relationship)
	Hartnoll et al., 1980 (N = 96)	31/44 patients in the heroin group and 13/52 in the methadone group were in treatment at 12 months; RR = 2.82 (95% CI = 1.70–4.68), in favor of heroin	Data divided into three categories of quantity of illicit opiate use (0–4, 5–39, and ≥40 mg); all patients in both groups continued to use illicit opiates during treatment; people in the two higher categories included 27/42 patients in the heroin group and 27/46 in the methadone group; RR = 1.10 (95% CI = 0.79–1.53)	No numerical datum is reported, but the authors stated that there was no significant difference between the two groups regarding use of drugs other than opiates (barbiturates, amphetamine, cocaine); neither were changes during the year observed	2/44 patients in the heroin group and 1/52 patient in the methadone group died during the treatment; suicide was considered as the most likely cause	–	27/42 patients in the heroin group and 30/46 in the oral methadone relied on criminal activity as a source of income; RR = 0.99 (95% CI = 0.72–1.34); no statistically significant difference	22/42 patients in the heroin group and 33/46 in the methadone group were arrested one or more times in the year; RR = 0.73 (95% CI = 0.52–1.03); no statistically significant difference	18/42 patients in the heroin group and 23/46 in the methadone group were employed; RR = 0.86 (95% CI = 0.54–1.35); no statistically significant difference; comparing with work status at baseline, it seems that a decline in full-time occupation was registered in both groups; no datum was provided about family relationships
	Perneger et al., 1998 (N = 51)	25/27 patients in the heroin group and 22/24 in the control group were in treatment at 6 months; RR = 1.01 (95% CI = 0.86–1.19); no difference between groups	6/27 patients in the heroin group and 14/21 in the control group reported using street heroin during the treatment; RR = 0.33 (95% CI = 0.15–0.72)	No main difference between the groups was observed	No death occurred	–	5/27 patients in the heroin group and 12/21 in the control group were charged for any reason in the previous 6 months (before assessment); RR = 0.32 (95% CI = 0.14–0.78), in favor of heroin; comparing the illegal incomes at baseline and at follow-up shows a significant reduction in the experimental group (passing from 3,372 to 331 Swiss francs at follow-up) vs. 3,123 to 4,931 Swiss francs in the control group (p = .053)	No information available	6/27 patients in the heroin group and 3/21 in the control group were employed at follow-up; RR = 1.56 (95% CI = 0.44–5.50); no statistically significant difference; when compared with data at baseline, it seems that the situation remained stable across the study; marital status follows a similar pattern: 12/27 patients in the heroin group and 7/21 in the control group had a stable partner; RR = 1.33 (95% CI = 0.64–2.79); there was no substantial difference when compared with the baseline information
	CCBH (A), 2000 (N = 174)	55/76 patients in the injected heroin + oral methadone group and 83/98 in the oral methadone group were in treatment at 12 months; RR = 1.17 (95% CI = 0.99–1.38); no statistically significant difference	CCBH (A)/ injected heroin and CCBH (B)/inhalable heroin adopted a composite measure in which the relevant outcomes were integrated	Considering the information provided about these two studies [CCBH (A) /injected heroin and CCBH (B) /inhalable heroin], it is not possible to report the detailed information	Injected heroin: one patient died several hours after discharge from the hospital for an epileptic seizure treated with opioid antagonist naloxone; methadone: one patient died following a traffic accident during the 12-month treatment (not related to the treatment)	18 SAEs occurred: 11 to 9/76 patients (for 1 patient having 3 episodes) in the heroin coprescribed group and 7 to 7/98 in the methadone group	No disaggregated information available	No disaggregated information available	No disaggregated information available
	CCBH (B) 2000 (N = 375; 256 reviewed)	80/117 patients in the inhaled heroin + methadone group and 121/139 patients in the methadone only group were still in treatment; RR = 0.79 (95% CI = 0.68–0.90), in favor of methadone	CCBH (A) /injected heroin and CCBH (B) /inhalable heroin adopted a composite measure in which the relevant outcomes were integrated	Considering the information provided about these two studies [CCBH (A)/injected heroin and CCBH (B)/inhalable heroin], it is not possible to report the detailed information	No death occurred in this trial	14/117 patients in the heroin group and 11/139 in the methadone group experienced 14 SAEs	No disaggregated information available	No disaggregated information available	No disaggregated information available

The outcomes considered are described by each study.

4. Discussion 

Only four studies (577 patients) met the inclusion criteria for this review: one was conducted in the 1970s in the United Kingdom, one in the 1990s in Switzerland, and two in 2000 in the Netherlands. One study (Hartnoll et al., 1980) showed a statistically significant higher retention in treatment among patients undergoing heroin treatment as compared with those undergoing methadone treatment, whereas one of the largest, most recent studies (van den Brink et al., 2003) found opposite results; however, the latter study, as referred by the authors, had more restricting rules in the heroin group than in the methadone one. Patients breaking those rules were dropped out from heroin treatment. This fact may explain part of the observed contradictory results.

Data on relapse to street heroin use were presented only in two studies (Hartnoll et al., 1980, Perneger et al., 1998), showing opposite findings. One of the two studies analyzing criminal activity shows that heroin treatment helps people keep away from illegal activities.

The multidomain outcome adopted in two studies (van den Brink et al., 2003) to take into consideration the complexity of the condition summarized measures related to physical health, mental status, social functioning, and substance use. The definition of success on this outcome was predefined in the protocol stage. The studies found statistically significant improvements in social and health functioning among patients treated with heroin.

However, the improvement was calculated as 40% in one of the areas in which those specific patients functioned poorly if that improvement did not go at the expense of any other outcome domain. The definition raises some caution on the generalizability of these results to other social contexts and to daily clinical practice, as patients would have difficulties in understanding the expected outcomes. The overall percentage of improvement does not allow weighting for each single dimension; this composite outcome seems to be more of interest for policymakers than for patients themselves, who can hardly disentangle what dimension can be of more relevance to them.

The maximum follow-up time of these studies was 1 year. The condition of heroin dependence commonly lasts for many years, such that different positive or negative outcomes of treatment cannot be excluded.

The studies were conducted in different years and social and geographical situations; these factors do affect particularly the evaluation of treatment of opioid dependence as compared with other clinical conditions because the social consequences of heroin use impact the overall well-being of a person. The experimental intervention was addressed to patients not responding to methadone treatment, but these patients might have completely different characteristics in those countries where appropriate methadone treatment is not easily accessible. Substitution treatments are aimed at reducing the social consequences of related problems (frequency of street heroin use; i.e., unknown composition of substance, contamination, dangerous modality of injection, illegal activities, loss of social relationship, unemployment) rather than the hazards derived from the biologic mechanism of heroin.

The studies considered for this systematic review were conducted in countries (United Kingdom, The Netherlands, Switzerland) that provide drug users with high-quality health and social services. The control treatment was methadone maintenance at its best performance. If replicated under different societal and clinical circumstances, the estimated effect of heroin substitution treatment would probably be much greater. Time and space should not be considered as confounders but instead as effect modifiers in estimating the efficacy of these treatments. If and when the effect of etiological factors for a specific disease and/or health effect is heterogeneous over time and space and between populations, the goal of systematic reviews is to explore and highlight interactions rather than to provide overall pooled estimates of efficacy. In view of the ongoing or projected trials on heroin prescription, we recommend that comparable outcomes be adopted to allow comparisons between different contexts and to highlight treatment effects.

The prescription of heroin to heroin users was proposed to overcome the failures of methadone treatment on some patients. In light of the observation that some people are not able to achieve detoxification in spite of several attempts and that they do not remain in methadone program even with flexible doses, it was assumed that prescription of heroin would have prevented the risks derived from street heroin use and illegal activities related to obtaining heroin. The main objectives of heroin prescription are, therefore, maintenance in treatment and social functioning. The results of the studies do not allow overall evaluation of efficacy of heroin prescription because of noncomparability of outcomes. Heroin use in clinical practice is still a matter of research in most countries.

Results of studies were sometimes in disagreement, especially for a crucial outcome such as the retention in treatment of patients. Contradictory results derived from most recent studies have been explained by the authors with the different rules adopted in the study groups. Patients given heroin were extruded more easily than those given methadone. This limits the comparability of interventions and, consequentially, the generalizability of results. New trials have been promoted in several countries and are hoped to provide results of effectiveness of heroin prescription in different social contexts (i.e., different patterns of use, different social environments, and different offerings of treatment strategies).

The adoption of comparable outcomes is therefore essential to allow comparisons among different studies. International public health institutions should allocate more resources for studies on the efficacy of substitution treatments. Services for heroin users must be based on best current scientific evidence on efficacy of treatments, avoiding biases based on prejudices and ideologies.