1. Clinical Studies

Five randomized controlled trials (RCTs)^{27,32,34–36} and five non-randomized studies^37–41 met the inclusion criteria. Secondary publications were available for two randomized controlled trials: Otiashvili et al., 2013,³⁵ (one additional publication)⁴² and Saxon et al., 2013,²⁷ (four additional publications).^31,33,43,44 All were secondary analyses that contained findings on additional outcomes; one publication also included long-term follow-up data (mean of 60 months).⁴⁴ Detailed tables on the study characteristics are available in Appendix 2, Tables A1 and A2.

The RCTs were open-labeled trials except for one trial that used a double-blind, double dummy study design.³² One trial was conducted in Georgia³⁵ and the other four trials were conducted in the US. The number of recruited patients in the included trials ranged from 54 patients to 1,269 patients. The diagnosis of opioid use disorder was confirmed based on the Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria in three RCTs.^27,32,36 One trial was conducted in incarcerated men with sentences of 10 days to 90 days and with a dependence to heroin and other non-prescribed opioids.³⁴ Patients in the other trials had addictions to heroin or prescription opioids with a duration of drug use of 6 years^35,36 and nine to 12 years.³² In Magura et al.³⁴ and in Saxon et al.²⁷ (and its secondary analyses), the mean duration of the addiction was not reported. All five RCTs’ goal was to evaluate maintenance treatment and two trials offered medication tapering over 8 weeks⁴³ or 12 weeks at the end of the trials.³⁵ The incarcerated men in Magura et al. received treatment for 23 days and 32 days with buprenorphine/naloxone and methadone respectively, while in jail.³⁴ Duration of treatment post-release was not stated although outcomes were assessed at 3 months. In the other trials, the patients were treated for 3 months to 6 months, and one trial²⁷ had long-term data with a follow-up duration of 2 years to 8 years post-randomization (mean of approximately 60 months).⁴⁴ One trial reported using methadone tablets.³⁶ Three trials reported mean daily doses of 8.5 mg, 14.9 mg, and 22.1 mg of buprenorphine (as part of the buprenorphine/naloxone mix), and 20.9 mg,39 mg and 93.2 mg of methadone.^27,35,36 The other two trials had daily doses of up to 16 mg and 32 mg of buprenorphine (buprenorphine/naloxone mix) and 70 mg to 90 mg of methadone; mean daily doses were not reported in these two trials.^32,34 The outcomes measured in the trials included: treatment retention (measured as the number of weeks patients remained in treatment or measured as number of patients remaining in treatment), opioid or heroin use (measured from urine samples or from self-reports), and harms. One trial used instruments (Timeline Followback, Addiction Severity Index, and Risk Assessment Battery) to measure drug use and behaviours related to needle-sharing and sexual activities.³⁵ Magura et al. also measured the frequency of re-arrest and re-incarceration.³⁴

The non-randomized studies were observational and originated from Finland, the United Kingdom, Italy, and the US (2 studies). Four studies enrolled opioid dependent patients undergoing opioid maintenance treatment.^37,39–41 Diagnosis of opioid use disorder was confirmed with DSM criteria in one study.³⁷ One of these four studies enrolled men only.⁴¹ One retrospective cohort study was conducted in mothers and their newborn babies.³⁸ The non-randomized studies selected data obtained over 6 months to a year in study periods ranging from one year to two years. The number of patients enrolled in these non-randomized studies ranged from 62 patients to 3,812 patients. The buprenorphine mean daily doses (as part of the buprenorphine/naloxone mix) ranged from 9.75 mg to 14.8 mg. Four studies reported methadone mean daily doses of 64.6 mg to 79.3 mg; one study did not report the methadone dose.⁴¹ The outcomes measured included treatment retention, opioid use (measured from urine samples or from self-reports), improvements in social life and education level, and presence of neonatal abstinence syndrome.

2. Economic Evaluations

Four economic evaluations were identified from the literature search^45–48 All four economic evaluations were cost-effectiveness analyses comparing buprenorphine/naloxone to methadone.^45–48 One economic evaluation also included a cost-utility analysis.⁴⁵ Countries of origin were Portugal, the United Kingdom, Greece, and Australia. Time horizons of 6 months in two evaluations^46,48 and of one year in the other two evaluations^45,47 were chosen. None of the evaluations reported a discount rate. Key assumptions and other characteristics are outlined in Appendix 2, Table A3.

Gouveia et al. used data reported in the literature.⁴⁵ A “social” perspective was chosen, although the perspective was really that of the public payer since indirect costs were not included. The measures of clinical effectiveness used were heroin-free days (based on negative urine test; data from the literature) and quality adjusted life years (QALYs). The cost data included medication cost (including cost of compounding the methadone), physician visits, psychotherapy, nursing labour costs, social worker visits, toxicology drug tests, and administrative/ general costs. Resource unit costs (in euros) were obtained from the Portuguese legislation.⁴⁵ The costs of crime were included in a sensitivity analysis with data obtained from an economic analysis dated 2011.⁴⁹

Maas et al. used data obtained from patients recruited from one rural and two urban community drug service clinics in the UK (n=361).⁴⁶ Effectiveness was defined as the ability to retain patients in the programme for 6 months; and facilitate cessation of illicit opiate use. All analyses were performed using intention to treat. Costs, which were estimated from the perspective of the drug treatment clinic, included the cost of medications, dispensing and supervision fees, and clinic costs such as urine tests. All costs were estimated in UK sterling at 2010–2011 financial year.⁴⁶

In Geitona et al., the data used were retrospectively retrieved from the local health authority databases.⁴⁷ The perspective was not explicitly stated but is that of the public payer as the expenses included personnel, drugs/ consumables, medical consultations/ diagnostic investigations, maintenance of equipment and buildings, and overheads. Prices (in euros) were those of the Greek National Health System in 2008. The clinical effectiveness was assessed using the completion of treatment (voluntary discharge of participants as a results of achieving abstinence from illicit opioids and completing a stabilization period of 2 years) and the number of deaths that were related to the use or overdose of illicit opioid drugs.⁴⁷

In Doran et al., a treatment provider perspective was adopted with a reference year of 1998 to 1999.⁴⁸ Resources use was identified at both the patient and facility level, which included staff time, diagnostics, medications, supplies, equipment, and ancillary services. The summation of patient and facility resource use provided an estimate of total cost of each patient’s treatment episode. The primary measure of clinical effectiveness was the change in number of heroin-free days between the month prior to treatment and the 6th month.⁴⁸

3. Guidelines

Two clinical practice guidelines met the selection criteria (Appendix 2, Table A4).^50,51 The two guidelines selected for review were developed in 2011, one by the Centre for Addiction and Mental Health (CAMH) in Canada⁵⁰ and the other by the World Federation of Societies of Biological Psychiatry (WFSBP).⁵¹ They provided recommendations for the treatment of opioid use disorder; one guideline was specific to buprenorphine/naloxone⁵⁰ and the other was more general and included all medications used in the treatment and management of opioid use disorder.⁵¹ Searches of electronic databases were performed from 1980 to 2009 in the Canadian guideline and up to January 2010 in the other guideline.

1. Clinical studies

The objectives and selection criteria were stated in all clinical studies. Patient characteristics, interventions, and outcomes were well described in eight of the clinical studies. Two non-randomized studies did not clearly describe the patients.^37,41 The interventions and the outcomes were not well described in one non-randomized study.⁴¹ The RCTs had computer generated randomizations and all were open-label with the exception of one RCT that used a double-blind, double dummy study design.³² Intention to treat analyses were performed in the RCTs by Otiashvili et al and in Saxon et al.^27,35 Saxon et al. provided justification on its choice of sample size.²⁷ One study would have results that would be generalizable to patients with opioid use disorder because it sampled 34 treatment facilities.³⁹ Three studies had findings which were generalizable to specific populations such as men,⁴¹ incarcerated men,³⁴ or newborns.³⁸

2. Economic Evaluations

In the economic evaluation reports, the research questions were well defined and the analysis methods were clearly stated. The key parameters on which the analyses were based were justified, except in Maas et al. which did not appropriately describe the dosages used.⁴⁶ The time horizons were clearly specified in all reports: All economic evaluations used a time horizon of at least 6 months which should be sufficient to determine whether or not a patient benefited from treatment. However, one evaluation included premature death, yet the time horizon of the analysis was limited to one year.⁴⁵ Sensitivity analyses were not performed in Mass et al.⁴⁶ In Gouveia et al., the range or distribution of values were not clearly described for conducting their sensitivity analyses.⁴⁵ One limitation of the Australian report was that the investigators based their analyses on retrospective data (cost and efficacy data) collected more than 10 years ago.⁴⁸ Discount rates were not applied in any of the evaluations most likely due to the fact that the time horizons were 6 months and one year. The generalizability of the study results to Canada is uncertain due to the fact that none of the evaluations used North American data.

3. Guidelines

The included guidelines were developed by professional association or expert committee based on a systematic review process which was well described in one guideline.⁵⁰ The objectives, clinical questions and the population for whom guidance was intended were well described. Guideline development groups were representative of their relevant professional groups and recommendations were peer reviewed in one guideline.⁵⁰ Conflict of interest was declared. The recommendations were clearly presented and explicitly linked to supporting evidence. One limitation was the lack of clarity regarding patient involvement in guideline development. The WFSBP guideline did not clearly provide the future update plan.⁵¹

What is the clinical effectiveness of buprenorphine/naloxone compared with methadone for the treatment of patients with opioid dependence?

Treatment retention

Treatment retention was an outcome of interest in four RCTs and one non-randomized study.^{27,32,35,36,39} Treatment retention was measured as number of weeks the patients remained in treatment or as number of patients who remained in treatment (Appendix 2, Table A2). As shown in Table 3, more than 80% of patients completed 12 weeks of treatment in one study (87.5% of patients with buprenorphine/naloxone compared with 82.5 % of patients with methadone, statistical significance not reported).³⁵ At 17 weeks, the result for methadone was numerically superior to buprenorphine/naloxone (28.9% of patients remaining in treatment with methadone vs. 24.1% with buprenorphine/naloxone, P value not reported.) At 6 months one study found no statistically significant difference between buprenorphine/naloxone and methadone³⁶ whereas two studies reported a statistically significant difference with more methadone patients remaining in treatment (48% with methadone vs. 30% with buprenorphine/naloxone, P = 0.001; and 74% with methadone vs. 46% with buprenorphine/naloxone, P < 0.01).^27,39 Patients who stopped treatment did so for a variety of reasons including being lost to follow-up and non-compliance with medication. The reasons for treatment discontinuation for each RCT are outlined in Appendix 5, Table A9.

Table 3

Treatment Retention.

When the number of weeks on treatment was considered, both methadone and buprenorphine/naloxone patients remained in treatment for a mean of 12 weeks in one study.³² Two other studies showed statistically significant differences of 17 weeks to 18 weeks for buprenorphine/naloxone compared with approximately 24 weeks to 26 weeks for the methadone group (P < 0.0001 and P < 0.001).^27,39

A secondary analysis of the Saxon et al. trial²⁷ evaluated the dose-retention relationship of the two medications.⁴³ Doses of methadone greater than 60 mg showed 80% or greater retention, and doses of 120 mg or greater showed 91% retention. Buprenorphine/naloxone showed a linear relationship, with increasing dose having better retention. At doses of 30 mg to 32 mg, a treatment retention of approximately 60% was obtained.⁴³

The Saxon et al. trial subsequently published a report on their long-term data.⁴⁴ At the 60-month interview, 46% of buprenorphine/naloxone patients and 37% of methadone patients were not in any treatment for opioid dependence. Also at 60 months, 48% of methadone patients were still in treatment with methadone compared with 12% of buprenorphine/naloxone patients being treated with buprenorphine/naloxone.⁴⁴ These findings are difficult to interpret because the reasons for stopping treatment were not described. Patients may no longer be in treatment because they were able to successfully stop the medication; the medication was stopped because of illicit opioid use while on treatment; or other reasons.

Use of Opioids or Heroin

Nine of the 10 studies assessed the use of opioids during treatment and this was measured by analysing urine samples or from patients self-reports (Table 4). Four studies found that patients on buprenorphine/naloxone were better at maintaining abstinence from opioids as measured in urine samples, and of these four studies, two reported a statistically significant difference.^35,41 Buprenorphine/naloxone was better numerically in two studies but statistical significance was not reported.^37,39 Conversely, methadone showed a statistically significant difference long-term, with 43% of buprenorphine/naloxone patients who had positive urine samples compared with 31% of methadone patients, P<0.01.^27,44 Two studies found no statistically significant difference;^32,43 however, one study was based a small sample size of 26 patients.³⁶

Table 4

Use of Opioids or Heroin.

When patients self-reported opioid use, one study reported no statistically significant differences between the two groups (approximately 14 days of opioid use in the last month).³⁴ Two studies had similar results (approximately 3 days to 4 days of opioid use in the last month in one study, and approximately 37 days to 38 days of heroin use in the last 3 months in the other study).^32,40 One study found a statistically significant difference; however this was based on the results for 26 patients, with 5 of 13 buprenorphine/naloxone patients reporting opioid use compared with no patients of the 13 methadone patients.³⁶ Longer-term, patients reported less days of heroin use with buprenorphine/naloxone (8.5 days in the last 3 months) compared with methadone (24 days in the last 3 months).⁴⁰

Harms

Harms of buprenorphine/naloxone and methadone were reported in five studies (Table 5).^{27,32,34–36} There were no statistically significant differences in the number of patients experiencing harms, including death; however one study reported more adverse events with buprenorphine/naloxone compared with methadone (108 events vs. 80 events, P = 0.003).³⁵ Examples of adverse events reported included insomnia, constipation, and depression.³⁵ Serious adverse events were reported in two studies and included persistent headache, non-cardiac chest pain, bradycardia, spontaneous abortion, suicidal ideation or threat, change in mental status, cholecystitis, gastric ulcer, benzodiazepine overdose, and hospitalization for abscess due to heroin injection, high blood pressure, lung mass with shoulder infection, intoxication or vomiting.^27,32 In one trial, one overdose with methadone was reported.²⁷ Withdrawals due to adverse events were reported in one study, with one patient withdrawing due to headaches, and another due to the presence of tic (reported as “nodding all the time”, page 8).³⁴ This same study reported that six patients in the buprenorphine/naloxone groups had to withdraw from treatment due to suspected diversion, compared with one methadone patient.³⁴

Table 5

Harms.

Two studies reported reductions in HIV risk behaviours (for example less needle sharing) and in sexual risk behaviours, with no statistically significant difference between groups.^27,31,35

What is the cost-effectiveness of buprenorphine/ naloxone compared with methadone for the treatment of patients with opioid dependence?

Treatment with buprenorphine/naloxone was more costly and more effective in the 2015 Portuguese economic evaluation.⁴⁵ The total costs for one year were 2,079.3 euros and 1,965.0 euros for buprenorphine/naloxone and methadone, respectively. The buprenorphine/naloxone group reported 284 heroin-free days whereas the methadone group reported 247 heroin-free days. The incremental cost per heroin-free days was 3.06 euros. Total QALYs were estimated to be 0.5901 and 0.5707 with buprenorphine/naloxone and methadone, respectively, for an incremental cost per QALY gained of 5,914.1 euros. When premature death attributable to treatment was no longer considered in a sensitivity analysis (in the base case analysis, a mortality rate of 0.02 per 1,000 patients with buprenorphine/naloxone and of 2.75 per 1,000 patients with methadone was assumed), the incremental cost-utility ratio increased to 16,604. In another sensitivity analysis, buprenorphine/naloxone became dominant (lower cost, higher QALY) when the costs of crime were considered.⁴⁵

The 2013 UK cost-effectiveness analysis reported greater effectiveness with methadone compared to buprenorphine/naloxone, as buprenorphine/naloxone was 19.4% less effective at retaining patients at 6 months.⁴⁶ Yet, 7% more patients stopped using illicit opiates with buprenorphine/naloxone. The differential cost was £63, with buprenorphine/naloxone having higher costs. Considering treatment retention, buprenorphine/naloxone was dominated by methadone. An incremental cost-effectiveness ratio (ICER) of £903 was reported for patients who stopped illicit opiate use.⁴⁶

In the 2012 Greek cost-effectiveness analysis, the estimated patient total costs for one year were 2,876 euros for treatment with buprenorphine/naloxone and 5,626 euros for treatment with methadone.⁴⁷ In terms of the clinical effectiveness, buprenorphine/naloxone increased the percentage of treatment completion by approximately 1.5-fold. The percentage of deaths in the buprenorphine/naloxone group was 2.5-fold less than that reported with methadone. As a result, the cost-effectiveness analysis demonstrated that buprenorphine/naloxone therapy was dominating methadone. The ICER for buprenorphine/naloxone versus methadone was -795.03 with respect to treatment completion, and was -1,410.7 with respect to percentage of avoided deaths. Variations in the different cost parameters, measured in sensitivity analyses, did not reverse the findings of the evaluation and in fact buprenorphine/naloxone became dominant for some scenarios.⁴⁷

In the 2005 Australian economic evaluation, the mean treatment costs over a 6-month period were AUD$1,593 for buprenorphine/naloxone, and AUD$1,415 for methadone.⁴⁸ The changes in the number of heroin-free days between the month prior to treatment (baseline) and the sixth month were 7.34 days for buprenorphine/naloxone and 6.84 days for methadone. Therefore, the ICER for the comparison between buprenorphine/naloxone and methadone was AUD$357 (confidence interval: -1,520 to 2,367), suggesting that buprenorphine/naloxone was more expensive but more effective than methadone. Variations in the different cost parameters, measured in sensitivity analyses, did not reverse the findings of the evaluation.⁴⁸

Table 6 summarizes the findings of the economic evaluations.

Table 6

Summary of Economic Evaluations.

What are the evidence-based guidelines associated with the use of buprenorphine/ naloxone for the treatment of patients with opioid dependence?

The CAMH guideline recommends that clinicians consider prescribing buprenorphine/naloxone or methadone for patients diagnosed with opioid dependence.⁵⁰ Care can be given in a primary care setting or in a specialized addiction treatment setting. The choice of buprenorphine/naloxone or methadone should be guided by the individual clinical circumstances and patient preference. Maintenance treatment with buprenorphine/naloxone should be initiated at a maintenance dose titrated as rapidly as possible to achieve stabilization while avoiding over-sedation or precipitated withdrawal. Once maintenance is achieved, non-daily dosing of buprenorphine/naloxone is as effective as daily dosing. All of these recommendations received the highest rating. Recommendations are given with respect to when buprenorphine/naloxone is preferred over methadone, for example in case of presence of prolonged QT interval, history of adverse events with methadone, significant respiratory illness, history of opioid use disorder is less than one year, and being elderly, adolescent or young adult.⁵⁰

The WFSBP guideline has only one recommendation with respect to buprenorphine/naloxone. It stated that buprenorphine/naloxone is a standard treatment for the treatment of opioid use disorder (recommendation not graded).⁵¹