Summary of Study Characteristics
Two systematic reviews,20,21 three RCTs (in four publications),22–25 six non-randomized studies,26–31 two economic evaluations,20,32 and two evidence-based guidelines13,14 were identified and included in this review. No relevant health technology assessments or meta-analyses were identified. Detailed characteristics are available in Appendix 2, , , , and .
Study Design
The two included systematic reviews20,21 had objectives and inclusion criteria that were broader than for the present report (i.e., wider in scope); only information from the subset of relevant studies is included here. Authors of one systematic review,20 published in 2018, included literature searches for published and unpublished RCTs and non-randomized comparative studies up to September 25, 2018. The second review,21 published in 2017 included RCTs and controlled clinical trials published before January, 2014. The first systematic review20 included five relevant primary studies, and the review by Minozzi et al.21 included one relevant RCT for a total of six unique primary studies (i.e., there was no primary study overlap between the included systematic reviews).
Ten primary study reports regarding the clinical effectiveness and safety of buprenorphine formulations for the treatment of OUD were identified. There were three RCTs in four publications: a randomized open-label study;22 a prospective, randomized, multi-centre, blinded then open-label, parallel-group, active-control, noninferiority study,24 a blinded, randomized, parallel-group, multi-centre, noninferiority study,25 and a secondary analysis of this later study.23 The six relevant non-randomized studies also utilized different methodologies: a retrospective cohort;26 a multi-centre, open-label, prospective cohort;28 a retrospective longitudinal study;30 a retrospective cohort study,29 and a prospective observational study.31 The sixth study27 was a multi-centre, open-label, uncontrolled, prospective cohort extension study.
The two economic evaluations20,32 employed Markov models. The clinical inputs used in these models came from various systematic reviews or individual clinical studies, as selected by the authors. Cost inputs were informed by clinical studies, various databases, or were provided directly from drug manufacturers. One economic evaluation,20 was conducted from the perspective of the United States health care sector using a five year time horizon (it also included a scenario analysis that took a modified societal perspective). The second economic evaluation32 took a US societal perspective using a 12-month time horizon.
Two evidence-based guidelines were identified regarding the treatment of OUD that contained recommendations for the use of various buprenorphine or BUP-NAL formulations.13,14 The first guideline, published in 2018 from the Canadian Research Initiative in Substance Misuse (CRISM), was based on two previous documents developed in British Columbia: “[…] the Vancouver Coastal Health/Providence Health Care Guideline for Clinical Management of Opioid Addiction released in November 2015, and the BC Centre on Substance Use/Ministry of Health Guideline for the Clinical Management of Opioid Use Disorder, released in February 2017”.13 They further updated the literature in 2016 and included meta-analyses of randomized clinical trials, clinical trials, observational reports, and expert opinion.13 The second guideline, published in 2015, from the United States’ Department of Veterans Affairs and the Department of Defense (VA/DoD) and is an update to their 2009 “Clinical Practice Guideline for the Management of Substance Use Disorders”.14 A systematic review was conducted to update the results from November 2007 onward, and included only systematic reviews or clinical studies (RCTs, prospective comparative studies).14 They interpreted the results and carried forward recommendations from the previous guidelines modifying or adding as necessary. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) tool was used to evaluate the quality of evidence and strength of recommendations in both guidelines (details in ).13,14 Recommendations were consensus-based and were developed with consideration of feedback from internal and external stakeholders and experts.13,14
Country of Origin
The included systematic reviews were by authors in the United States20 and Italy.21 Relevant primary studies included in the systematic reviews were conducted in the United States and published between 2008 and 2018.
The RCTs were conducted in the United Kingdom22 and the United States.23–25 The non-randomized studies were conducted in Australia,26 the United States,27–30 and Germany.31
The two economic evaluations were conducted in the United States. 20,32
The guidelines were developed in Canada13 and the United States.14
Patient Population
One systematic review20 included studies that enrolled patients (≥ 16 years of age) with OUD in various treatment settings. The review by Minozzi et al.21 examined adolescent patients (≤ 18 years of age) with opioid dependence. Neither systematic review excluded studies that included participants with co-morbid physical or psychological illness.
The four RCT reports focused on adults in different settings, such as: 36 participants with opioid dependency commencing buprenorphine maintenance in specialized clinical trials facilities and addictions treatment facilities;22 310 participants with opioid dependency in the past 12 months,24 and similarly in 758 participants,23,25
The six non-randomized studies also focused on different populations and settings. Three studies evaluated all participants in large databases (i.e., 4,692 participants in a jurisdictional drug monitoring database,26 4,306 participants in a private insurance claims database,29 and 495 eligible participant’s electronic medical records30) and included all patients receiving treatment for opioid dependence.26,29,30 Other studies were set in study centres (249 participants), or addiction medicine physician practice sites (384 participants).31 The sixth study27 included adult participants aged 18 to 65 years, with opioid dependence and having received buprenorphine-based opioid substitution therapy for at least 22 days.
The two economic evaluations20,32 were conducted with patient populations in the United States. One study20 assessed the cost-effectiveness of several drugs used for medication-assisted treatment among a cohort of patients who were considered for OUD treatment. The second study32 analysed adults with OUD who were classified as clinically stabilized (i.e., those who achieved prolonged clinical stability on < 8 mg of daily transmucosal buprenorphine).
The target populations for the CRISM guidelines are adolescents, young adults, and adults with uncomplicated OUD and also included specific considerations for special populations.13 The intended users are Canadian physicians, nursing and allied healthcare providers, medical educators, clinical care case managers, policymakers, healthcare administrators.13 The VA/DoD guidelines apply to service members (18 years or older) and veterans with substance use disorder, and the intended users are VA/DoD health care providers, and others involved in the care of the target population 14
Interventions and Comparators
In one systematic review20 eligible medication interventions (i.e., buprenorphine subcutaneous extended-release injection, buprenorphine implant) were compared to other active treatments for patients with OUD (e.g., BUP-NAL in sublingual and buccal formulation). The Minozzi et al.21 systematic review investigated the effectiveness of any opioid agonist treatment (including buprenorphine) alone or in conjunction with psychosocial interventions compared with no intervention, alternative opioid agonist treatments, other pharmacological interventions, any detoxification intervention, or psychosocial interventions alone.
One RCT evaluated buprenorphine oral lyophilisate wafers administered on the tongue compared to standard sublingual buprenorphine.22 The three other studies evaluated BUP-NAL (rapid dissolving)23,24 sublingual tablet25 compared to either generic buprenorphine sublingual tablets,24 or BUP-NAL sublingual film and generic buprenorphine sublingual tablets (for the induction phase).23,25
One non-randomized study evaluated buprenorphine compared to BUP-NAL.26 Two studies did not have a comparator but evaluated BUP-NAL rapid dissolving sublingual tablet,27 and BUP-NAL.31 One study evaluated the conversion (i.e., switching patients from one formulation to another and where the dosages are not necessarily equal) from BUP-NAL sublingual tablet or film to BUP-NAL buccal film formulation.28 Another,30 compared sublingual buprenorphine to sublingual BUP-NAL and other opiate substitution therapies. The sixth study,29 compared the BUP-NAL sublingual film formulation to the BUP-NAL sublingual tablet formulation.
One economic evaluation20 compared the cost-effectiveness of several opioid substitution treatments (buprenorphine subcutaneous extended-release injection, naltrexone extended-release injectable suspension, buprenorphine implant) with generic sublingual BUP-NAL. The economic study by Carter et al. evaluated the cost-effectiveness of subdermal implantable buprenorphine compared to sublingual buprenorphine for the treatment of OUD, along with monthly psychosocial counselling in both groups.
The CRISM guidelines considered the following treatments for OUD: opioid agonists and antagonists, withdrawal management strategies, psychosocial interventions, and residential treatment.13 The VA/DoD considered various OUD treatments, including pharmacological therapies, brief interventions, mutual help programs, psychotherapy, and psychosocial interventions.14
Outcomes
The outcomes considered in the SRs were illicit use of opioids,20,21 opiate withdrawal symptoms20 (e.g., COWS, SOWS), patient adherence and dropout (e.g., all-cause treatment discontinuation),20,21 adverse events,20 and mortality.20
In the RCTs, the outcomes of interest were opiate withdrawal symptoms22–25 (e.g., COWS, SOWS, OOWS), patient adherence and dropout,22,24,25 adverse events, 22–25 and mortality.22,24
The outcomes of interest in the non-randomized studies were illicit use of opioids,30 health care utilization26 (e.g. hospital or emergency department admissions), adverse events27,28,31 (including anomalous liver function laboratory values [e.g., alkaline phosphatase, glutamic-pyruvic transaminase, glutamate oxaloacetate transaminase, and gamma-glutamyl transpeptidase]27,28,31), and mortality.26–28,31
In the economic evaluations, the outcome of interest was incremental cost per quality adjusted life-years gained.20,32
The three main symptom severity scales used for opiate withdrawal were: (1) the Clinical Opiate Withdrawal Scale (COWS), (2) the Objective Opiate Withdrawal Scale (OOWS), and (3) the Subjective Opiate Withdrawal Scale (SOWS). A brief description of these three scales is provided below.
- (1)
Clinical Opiate Withdrawal Scale (COWS; used in four studies20,23–25): A validated 11-item, clinician-administered tool used to reproducibly rate common signs and symptoms of opiate withdrawal.34,35 Total scores range between 0 and 47, and withdrawal is classified as mild (5 to 12), moderate (13 to 24), moderately severe (25 to 36), or severe (>36).34,35
- (2)
Objective Opiate Withdrawal Scale (OOWS; used in one study22): A validated 13-item rating scale in which physically-observable symptoms (e.g., perspiration, tremor, vomiting, anxiety) are rated as present or absent.36 Total scores range between 0 and 13. No clinically significant ranges or thresholds were mentioned in the identified literature.
- (3)
Subjective Opiate Withdrawal Scale (SOWS; used in five studies20,22–25): A validated 16-item symptom severity scale in which patients are asked to rate various opiate withdrawal symptoms on a scale of 0 (not at all) to 4 (extreme).36 Total scores range between 0 and 64. No clinically significant ranges or thresholds were mentioned in the identified literature.
The outcomes of interest in the guidelines included OUD adherence with treatment, emergency department utilization, morbidity, mortality, overdoses, relapse, adverse events, cravings, substance consumption (e.g., alcohol, opioid), and quality of life.
Summary of Critical Appraisal
Additional details regarding the strengths and limitations of the included publications are provided in Appendix 3, , , , and .
Systematic Reviews
Strengths of both systematic reviews20,21 included: clear objectives and inclusion criteria, report of key search terms and search strategies, provision of a list of included studies and summary of their characteristics, justification for eligible study designs, and detailed descriptions of the processes used for article selection, data extraction, and quality assessment. In addition, the Minozzi et al.21 review included a list of excluded studies and reasons for exclusion. The authors of both systematic reviews20,21 published detailed protocols containing their proposed methodologies prior to conducting the reviews. These strengths of reporting increase confidence in the findings and the reproducibility of the systematic reviews. In both reviews, multiple databases were used to identify relevant literature and various strategies to identify grey literature were performed by review authors, decreasing the risk of missing relevant, non-indexed studies. One systematic review20 restricted the search to studies published in English and did not provide justification for this decision. Study selection and data extraction were performed individually by multiple authors and any disagreements were resolved by group discussions in both systematic reviews.20,21 The possibility of publication bias was discussed and investigated to the degree possible in both reviews,20,21 and no evidence of publication bias was identified. The authors of both systematic reviews20,21 disclosed their conflicts of interest and sources of funding, none of which were considered likely to have influenced the findings.
RCTs
There were several strengths common to all four RCT reports,22–25 such as: clear descriptions of objectives, interventions, main outcomes, population characteristics, and eligibility criteria; participants appeared to be representative of the population of interest; and main outcome measures used were valid and reliable (e.g., COWS scores). Power calculations, and recruitment of adequate sample size, were performed in two studies,24,25 and a description of the participant randomization process were provided. This contrasts with the two other studies,22,23 where details on the methodology used to randomize were lacking and absence of allocation predictability could not be assessed. Three studies had the strength of being multi-centric.23–25 Reported patients lost to follow-up or withdrawn were 16.7%,22 28.5%,25 30.9%,23 and 35.8%,24 but no associated patient characteristics were provided.22–25 A further limitation present in all studies revolved around conflicts of interest. Strang et al.22 reported conflicts of interest, including that key authors have patents on various novel naloxone formulations. The authors of Gunderson et al. 2016 and 2015 reported conflicts of interest, including that some authors have received funding from the manufacturer of their respective intervention.23,25 Lastly, there was no disclosure of conflict of interest in one study, therefore possible conflicts of interest could not be assessed.24
Non-Randomized Studies
There were several strengths common to all six non-randomized studies,26–31 such as: clear descriptions of objectives, interventions, main outcomes, population characteristics, eligibility criteria; participants appeared to be representative of the population of interest; and main outcome measures used were valid and reliable (e.g., COWS scores). All studies planned their data analyses at the outset,26–31 and one study performed survival analyses for the primary outcomes to adjust for the different lengths of follow-up in the intervention and comparator groups.26 None of the studies reported conducting an a priori power calculation.26–31 Actual probability values (P-values) were not reported in three studies.27,28,30 One study in particular,27 suffered a 56.1% dropout rate mostly due to patients being lost to follow-up or patient nonadherence. A further limitation of some studies was that sources of funding were either not disclosed,31 or disclosed,27–29 included received funding from the manufacturer. Similarly, some study authors disclosed conflicts of interest which may have influenced the findings of the study,27–29,31 including that some received funding from the manufacturer of their respective intervention.
Economic Evaluations
The research objectives, economic importance of the research question, time horizons, viewpoints of the analyses, and rationale for choosing interventions and comparators were clearly stated, in both economic evaluations. The five year time horizon used in one economic evaluation20 was appropriate; however, the 12 month time horizon used in the Carter et al.32 analysis may not be sufficiently long given that treatment for opioid use disorders often takes place over multiple years. The choice of form of economic evaluation was justified in both studies (four and five-state Markov models).
As for the methods of data collection, the sources of effectiveness estimates, transition probabilities, drug costs, and other relevant model inputs were included in both studies.20,32 The primary outcome measures and methods to assign value to clinical benefits were described in detail. Both studies20,32 provided details of currency and price adjustments for inflation. Although both studies stated the discount rates that were applied to their models, justification for selecting these rates was not provided. The Carter et al.32 analysis made assumptions that may limit their findings: (1) both treatment cohorts received monthly psychosocial counselling if they were retained in treatment, and (2) the model did not consider the possibility of patients transitioning to an off treatment, not relapsed state (i.e., fully recovered). These assumptions, whilst simplifying the model, may not be truly accurate in reality, as treatment with monthly psychosocial counselling is likely to have some sort of dropout rate. Also, since the study did not examine the cost-effectiveness of the buprenorphine subdermal implant alone nor sublingual buprenorphine alone, but rather in combination with psychosocial counselling, the magnitude of effectiveness and the associated costs would both be anticipated to be different than with the intervention of interest alone. Additionally, the possibility of patients recovering from their condition, therefore no longer requiring opioid agonist maintenance treatment, should be considered.
The authors of the Carter et al.32 economic evaluation disclosed conflicts of interest including former or current employment with manufacturers, and the study was funded by a manufacturer. The authors and expert reviewers of the second economic evaluation20 reported no relevant conflicts of interest related to the analysis and disclosed sources of funding. The generalizability of these two economic evaluations20,32 to the Canadian context is limited given they were conducted in the United States.
Evidence-Based Guidelines
In both guidelines13,14 the scope and purpose were described. Developed in Canada, the CRISM guideline sought the views and preferences of the target population, provided an explicit link between the recommendations and the evidence, and indicated a procedure for updating the guideline in the future.13 Overall, the methodology used to develop the guidelines were rigorous. Whereas only the VA/DoD guidelines employed systematic methods to search for evidence,14 they both described thoroughness in selecting evidence, assessing quality of the evidence, and having the guideline document externally reviewed.13,14 In both cases, the guideline development groups were comprised of experts and stakeholders who were required to declare conflicts of interest. 13,14 The recommendations were well presented and unambiguous and included information on the quality of the evidence and strength of the recommendations.13,14 Neither guideline included discussion of facilitators or barriers to implementation, resource implications with respect to application, or monitoring and auditing criteria.13,14 The views of the funding bodies did not appear to have influenced the content of the guidelines.13,14 The generalizability of the VA/DoD guideline to the Canadian context is limited given that it was conducted for a specific target population and it was in the United States.14
Limitations
A number of limitations were identified in the critical appraisal (Appendix 3, , , , and ), however, additional limitations exist. The main limitations of this review are related to risk of bias, limited study populations and generalizability of findings.
A primary limitation that should be considered when interpreting these results is that participants and outcome assessors were not blinded to the treatment received in a majority of the reviewed studies. Given that several of the outcomes reported in these trials were based on subjective measures (e.g., SOWS scores, opioid craving visual analogue scale scores, self-reported use of heroin), the findings of open-label studies may be at risk for bias in either direction depending on the perceptions and expectations of participants and clinicians involved.
Only one included study20 contained information specific to pediatric or adolescent populations with OUD; therefore, the comparative clinical effectiveness, safety, and cost-effectiveness of various buprenorphine formulations in children is largely unknown. The applicability of the evidence to Canadian settings is unclear as all relevant clinical studies20–32 were conducted outside of Canada. Furthermore, it may be difficult to generalise the results in women since most studies enrolled a disproportionately higher number of men.22,24,26–29,31 There is also uncertainty around the generalizability of the cost-effectiveness findings due to the potential for significant differences in drug prices and associated costs between Canada and the United States, where the economic studies took place.