bs_bs_banner

R E V I E W

Drug and Alcohol Review (March 2014), 33, 115–128 DOI: 10.1111/dar.12095

COMPREHENSIVE REVIEW

A systematic review and meta-analysis of naltrexone implants for the treatment of opioid dependence SARAH LARNEY1,2, LINDA GOWING3, RICHARD P. MATTICK1, MICHAEL FARRELL1, WAYNE HALL4 & LOUISA DEGENHARDT1,5 1

National Drug and Alcohol Research Centre, University of New South Wales, Sydney, Australia, 2Alpert Medical School, Brown University, Sydney, Australia, 3School of Medical Sciences, University of Adelaide, Adelaide, Australia, 4Centre for Clinical Research, University of Queensland, Brisbane, Australia, and 5School of Population and Global Health, University of Melbourne, Sydney, Australia

Abstract Introduction and Aims. Naltrexone implants are used to treat opioid dependence, but their safety and efficacy remain poorly understood.We systematically reviewed the literature to assess the safety and efficacy of naltrexone implants for treating opioid dependence. Design and Methods. Studies were eligible if they compared naltrexone implants with another intervention or placebo. Examined outcomes were induction to treatment, retention in treatment, opioid and non-opioid use, adverse events, non-fatal overdose and mortality. Quality of the evidence was assessed using the Grading of Recommendations Assessment, Development, and Evaluation approach. Data from randomised studies were combined using meta-analysis. Data from non-randomised studies were presented narratively. Results. Five randomised trials (n = 576) and four non-randomised studies (n = 8358) were eligible for review.The quality of the evidence ranged from moderate to very low. Naltrexone implants were superior to placebo implants [risk ratio (RR): 0.57; 95% confidence interval (CI) 0.48, 0.68; k = 2] and oral naltrexone (RR: 0.57; 95% CI 0.47, 0.70; k = 2) in suppressing opioid use. No difference in opioid use was observed between naltrexone implants and methadone maintenance (standardised mean difference: −0.33; 95% CI −0.93, 0.26; k = 1); however, this finding was based on low-quality evidence from one study. Discussion. The evidence on safety and efficacy of naltrexone implants is limited in quantity and quality, and the evidence has little clinical utility in settings where effective treatments for opioid dependence are used. Conclusion. Better designed research is needed to establish the safety and efficacy of naltrexone implants. Until such time, their use should be limited to clinical trials. [Larney S, Gowing L, Mattick RP, Farrell M, Hall W, Degenhardt L. A systematic review and meta-analysis of naltrexone implants for the treatment of opioid dependence. Drug Alcohol Rev 2014;33:115–128] Key words: opioid dependence, heroin, naltrexone, investigational therapies, systematic review.

Introduction Naltrexone is an opioid antagonist that blocks the euphoric and analgesic effects of opioids. It has been investigated for use in treating opioid dependence under the rationale that motivation to use opioids will be reduced if a person is unable to experience their pleasurable effects.

Naltrexone was originally approved as an oral medication for management of opioid dependence, but its effectiveness was limited by poor adherence and low retention in treatment [1]. This situation led to the development of several sustained-release formulations of naltrexone that could be injected or implanted. Naltrexone depot formulations have been approved for use in the USA and Russia. Only Russia has approved

Sarah Larney PhD, Post-doctoral Research Fellow, Linda Gowing PhD, Associate Professor, Richard P. Mattick PhD, Professor, Michael Farrell MB FRCP FRCPsych, Director, Wayne Hall PhD, Professor, Louisa Degenhardt PhD, Professor. Correspondence to Dr Sarah Larney, The Miriam Hospital, RISE Building, Room 115, 164 Summit Avenue, Providence, RI 02904, USA. Tel: +1401 533 8663; Fax: +1401 793 4779; Email: [email protected] Received 17 September 2013; accepted for publication 6 November 2013. © 2013 Australasian Professional Society on Alcohol and other Drugs

116

S. Larney et al.

a naltrexone implant for the treatment of opioid dependence [2]. In 2008, a systematic review of depot and implantable formulations of sustained-release naltrexone concluded that there was insufficient evidence to evaluate their effectiveness in treating opioid dependence [3]. In 2010, a review of naltrexone implants by the Australian National Health and Medical Research Council (NHMRC) concurred with the 2008 review that there was insufficient evidence to determine the efficacy of this approach, either alone or in comparison with conventional, first-line treatments such as opioid substitution therapy (OST) [4]. More recently, a review of sustained-release naltrexone formulations concluded that there was evidence of the safety, efficacy and patient acceptability of naltrexone implants [5]; however, this review failed to include all published studies of naltrexone implants, did not assess the quality of the evidence and did not employ weighting in meta-analysis. It also did not acknowledge the major limitation that none of the reviewed studies compared naltrexone implants with OST. In light of recently published studies, the limitations of recent reviews and ongoing clinical use of unregistered implants [6], it is appropriate to provide an updated review of the evidence regarding naltrexone implants and issues requiring research attention. We aimed to assess the efficacy and adverse events of naltrexone implants when used to treat opioid dependence.

except if this was in comparison with naltrexone implants. There were no restrictions on the type of participants or length of follow-up. Publications in all languages were eligible. Search strategy and study selection

This review is reported in line with the Preferred Reporting Items for Systematic Reviews and MetaAnalyses Statement [7]. The review protocol is provided in the Supplementary Materials.

The search strategy built on that used by the NHMRC’s 2010 systematic review of naltrexone implants [4]. All studies included in that review were assessed for inclusion in the present review. Studies published since the NHMRC review were located by searching EBM Reviews, the Cochrane Central Register of Controlled Trials, MEDLINE, Embase, CINAHL and PsycINFO. We also searched for current trials using naltrexone implants on ClinicalTrials.gov, the EU Clinical Trials Register (clinicaltrialsregister.eu) and the Australian and New Zealand Clinical Trials Register (anzctr.org.au). The strings used in these searches were developed by a specialist drug and alcohol librarian, and searches were limited to 2009 onwards. Finally, reference lists of prior systematic reviews on this topic were searched for any additional publications. The search was originally completed in December 2012 and updated in August 2013. Search strings for each database are provided in the Supplementary Materials. Retrieved citations were screened by the lead author (S. L.) and clearly irrelevant citations (e.g. animal studies; editorials; non-implant formulations of naltrexone) were excluded. The complete text of the remaining studies was retrieved and read in full to determine eligibility for inclusion. The list of studies selected by S. L. for inclusion was circulated to all authors for discussion, and the final list of included studies was approved by all authors.

Eligibility criteria

Data extraction

Studies were eligible for inclusion if they investigated naltrexone implants versus any other comparator in treating opioid dependence. We included randomised trials presenting data on induction to treatment, retention in treatment, opioid use, non-opioid drug use, adverse surgical site-related events, and other adverse events and effects possibly related to treatment. Additionally, we included randomised controlled trials or non-randomised observational studies that reported on opioid overdose, non-opioid drug overdose or mortality. We included non-randomised studies in the assessment of these latter outcomes because they are infrequent and therefore difficult to examine in individual randomised trials [8]. Studies were excluded if they reported on non-implant formulations of naltrexone (e.g. oral or sustained-release injectable naltrexone),

Data extraction was undertaken independently by S. L. and a research assistant. Data were entered into RevMan 5.2 [9]. When necessary, the corresponding authors of studies were contacted to clarify published data and request additional data. The two completed RevMan files were compared, and discrepancies were resolved through discussion and referral to the senior author (L. D.).

Methods

© 2013 Australasian Professional Society on Alcohol and other Drugs

Risk of bias assessment Risk of bias in each study was assessed using the domains and criteria set out by the Cochrane Collaboration [8] and additional domains and criteria developed by the Cochrane Drugs and Alcohol Review Group [10]. Domains assessed for risk of bias were

Systematic review of naltrexone implants for opioid dependence

random sequence generation, allocation concealment, blinding of outcome assessment, incomplete outcome data, selective reporting, comparability of cohorts, representativeness of the exposed cohort, selection of the non-exposed cohort and ascertainment of exposure. Blinding of outcome assessment was assessed separately for subjective and objective outcomes. Subjective outcomes were self-reported substance use, selfreported overdose, site-related adverse events and other adverse effects or events possibly related to treatment. Objective outcomes were induction as randomised, retention in treatment, substance use determined by urinalysis, overdose as determined by hospital records and mortality. Risk of bias due to incomplete outcome data was assessed separately for each outcome. A summary risk of bias was determined for each outcome within a study, across the domains listed above, following Cochrane Collaboration guidelines [8]. The summary risk of bias for each outcome was used to inform judgements about the quality of evidence for each outcome. Publication bias was assessed by checking that results from all registered, completed clinical trials had been published. With low numbers of studies for each outcome, funnel plots were considered unlikely to enhance assessment of publication bias. Quality of evidence Quality of evidence was determined using the Grading of Recommendations Assessment, Development, and Evaluation approach [11]. ‘Quality’ refers to confidence in the calculated effect estimate. Evidence is rated as high, moderate, low or very low quality. A quality rating of ‘high’ means that it is likely that the true effects lies close to the estimated effect, whereas ‘very low’ means that the true effect is likely to be substantially different from the estimated effect. Randomised trials begin with a quality rating of high, and observational studies begin with a quality rating of low. These ratings can be downgraded on the basis of limitations in any of five criteria: risk of bias in included studies; inconsistency in results; indirectness of evidence; imprecision of effect estimates; and risk of publication bias. Ratings can be upgraded if the effect size is large, there is evidence of a dose-response effect or if all plausible confounding is controlled for [8,11]. Data synthesis Statistical analyses were undertaken in stata 12 (StataCorp, College Station, Texas, USA) using the metan command. Only results from randomised studies were used in meta-analyses. Non-randomised studies were excluded from meta-analysis because of the high risk of bias within these studies and the lack of data

117

appropriate for meta-analysis. In all meta-analyses, results are presented by comparator. Given the broad range of comparators used, an overall effect for each outcome was not calculated. An exception to this was the outcome of surgical site-related adverse events, for which all studies that compared an active naltrexone implant to a placebo implant were combined, even if participants receiving a placebo implant also received oral naltrexone.This was considered appropriate as this analysis included only adverse events related to the implant site. For dichotomous data (e.g. retention in treatment), we calculated risk ratios (RR), and for continuous data (e.g. number of days of opioid use), we calculated standardised mean differences. Studies with zero events of interest in both arms were excluded from the metaanalyses [8]. Studies were combined using a fixed effect (Mantel–Haenszel) model. Heterogeneity of study outcomes was assessed using the P-value for the χ2 statistic and I2. Data from non-randomised studies are presented as narrative summaries. Results We considered 196 unique citations for inclusion in the review (Figure 1). Of these, 164 were excluded as irrelevant. The complete text of the remaining 32 studies was retrieved and read in full. Of these, 23 were excluded, for reasons summarised in Figure 1 (further details available in the Supplementary Materials). Nine studies were eligible for the review: five randomised controlled trials and four non-randomised studies. Characteristics of included studies Characteristics of included studies are described in Table 1. Randomised trials compared naltrexone implants with the following comparators: placebo implant (with oral placebo [12] and without oral placebo [13]); oral naltrexone with placebo implant [12,14]; treatment as usual (which included outpatient counselling, applying for methadone maintenance treatment, readmission to inpatient care or vocational counselling and social services) [15]; and methadone maintenance treatment [16]. One randomised trial included two comparators: placebo implant with oral placebo and oral naltrexone with placebo implant [12]. In the non-randomised studies, naltrexone implants were compared with oral naltrexone [6,17], methadone maintenance treatment [18] and buprenorphine maintenance treatment [19]. In two of the five randomised trials, all participants were currently opioid dependent (by the Diagnostic and Statistical Manual of the American Psychiatric Association; DSM-IV) adults who were living in the community © 2013 Australasian Professional Society on Alcohol and other Drugs

118

S. Larney et al.

Figure 1. Study flow diagram. NHMRC, National Health and Medical Research Council; RCT, randomised controlled trial.

[12,14]. One randomised study recruited opioiddependent (criteria not specified) adults prior to discharge from abstinence-based inpatient treatment [15]. Another study recruited adult prisoners who were opioid dependent (DSM-IV) prior to incarceration and nearing release from prison [16]. The final randomised study recruited concurrently amphetamine and opioid dependent (DSM-IV) adults living in the community [13]. All four of the non-randomised studies utilised administrative data from community-based adults seeking treatment for opioid use. One of these studies stated that all participants met DSM-IV criteria for opioid dependence [18], but it is unclear how this was determined. Outcomes assessed in each study are shown in Figure 2. All randomised trials assessed induction to treatment as randomised; opioid use during treatment; and non-site related adverse events and effects possibly related to treatment. Site-related adverse events were assessed by all randomised trials, but data were only reviewed for those trials with a placebo implant comparison. Retention in treatment, non-opioid drug use, opioid overdose and mortality were assessed in some © 2013 Australasian Professional Society on Alcohol and other Drugs

but not all randomised trials. Three of the four nonrandomised studies assessed mortality and one assessed opioid and non-opioid drug overdose. Quality of evidence The quality of the reviewed evidence was affected by bias as a result of limitations in study design and implementation, choice of comparator and the generalisability of the results from the study sample to opioid dependent persons. Table 2 presents the risk of bias assessment for various domains within each study and a summary rating of risk of bias for each outcome within a study. Reasons for risk of bias judgements are provided in the Supplementary Materials. The studylevel risk of bias for each outcome is discussed further below with the results for each outcome. We found no evidence of publication bias. Only one randomised controlled trial directly compared naltrexone implants with the conventional firstline treatment, OST (in the form of methadone maintenance) [16]. This trial studied a small sample (n = 44) and suffered from high participant dropout: only 69% (16/23) of participants randomised to

Systematic review of naltrexone implants for opioid dependence

119

Table 1. Characteristics of included studies Randomised studies Author, year

Methods

Participants

Intervention

Hulse et al., 2009 [14]

Randomised, double-blind, placebo controlled trial of naltrexone implants compared with oral naltrexone, with 6-month follow-up

70 DSM-IV opioid-dependent persons aged 18 years or older

Krupitsky et al., 2012 [12]

Randomised double-blind double-placebo controlled trial of naltrexone implants compared with oral naltrexone, with 6-month follow-up Open-label randomised controlled trial of naltrexone implants compared with treatment as usual, with 6-month follow-up

306 DSM-IV opioid-dependent adults ‘with physiological features for at least 1 year’, aged 18–40 years

Naltrexone implant: 2.3 mg ‘O’Neil implant’ manufactured by Go Medical Inc. + daily oral placebo (lactose). Comparison: daily 50 mg oral naltrexone tablets (Orphan Australia) + placebo implant Bi-weekly drug counselling plus one of the following: 1 g naltrexone implant (Prodetoxon) + oral placebo; placebo implant + 0.5 g oral naltrexone daily; placebo implant + oral placebo 2.2 g naltrexone implants (Go Medical Industries) compared with treatment as usual, which included options to engage in outpatient counselling, apply for methadone maintenance treatment, readmission to inpatient care or vocational counselling and social services 2.2 g naltrexone implants (Go Medical Industries) compared with methadone maintenance treatment, 30 mg per day building up to 80–130 mg per day within 3 weeks. Treatment was commenced 1 month prior to release from prison 1 g naltrexone implant (Prodetoxon, manufacturer not specified) compared with placebo implant ‘that was identical in appearance’ to the active implant (no further description provided)

Kunøe et al., 2009 [15]

56 opiate-dependent adults (criteria not defined, but DSM-IV used in follow-up assessments) receiving abstinence-oriented in-patient treatment

Lobmaier et al., 2010 [20]

Open-label randomised controlled trial of naltrexone implants compared with methadone maintenance treatment

44 prisoners (41 male, 3 female) with pre-incarceration heroin dependence (DMS-IV)

Tiihonen, 2012 [13]

Double-blind randomised controlled trial comparing naltrexone implants with placebo implants, with 10-week follow-up

100 concurrently amphetamine- and opioid-dependent (DSM-IV) adults. n = 50 in each groups; NI group mean age 28 years, PI group mean 29 years

Non-randomised studies Author, year Kelty & Hulse, 2012 [6]

Ngo et al., 2008 [21]

Methods Retrospective cohort study (record linkage) comparing mortality outcomes for patients receiving naltrexone implant or entering oral naltrexone Retrospective cohort study (record linkage) comparing overdoses requiring hospitalisation following naltrexone implant or entry with methadone maintenance treatment, with follow-up at 6 months and 3.5 years

Participants

Intervention

3856 adults seeking treatment for ‘problematic opioid use’

Implant naltrexone (manufacturer: Go Medical) and oral naltrexone provided as part of routine clinical practice

836 heroin-dependent (DSM-IV) adults

Naltrexone implants (Go Medical) provided under clinic-specific treatment protocols; methadone maintenance treatment provided under statewide (Western Australia) treatment protocol. Both treatments provided as part of routine clinical practice Wedgewood and Go Medical naltrexone implants (dose not specified); buprenorphine maintenance treatment (dose not specified). Treatment provided as part of routine clinical practice 2.2 g naltrexone implants (manufacturer: Go Medical) provided as part of routine clinical practice. Methadone maintenance treatment provided under statewide (Western Australia) treatment protocol as part of routine clinical practice

Reece, 2010 [19]

Retrospective clinical record review of naltrexone implant and buprenorphine maintenance patients

2634 adults seeking treatment for opioid use

Tait et al., 2008 [7]

Retrospective cohort study (record linkage) comparing mortality outcomes for patients receiving naltrexone implants or entering methadone maintenance treatment

1032 adults seeking treatment naltrexone implants or methadone maintenance treatment

© 2013 Australasian Professional Society on Alcohol and other Drugs

120

S. Larney et al.

Figure 2. Outcomes assessed in each included study. Note: Although Kunøe et al. [12] and Lobmaier et al. [20] both include reports of adverse site-related events, these data were not included in the review as there was no placebo implant comparison in these studies. - Non-randomised studies were not reviewed for data on these outcomes.

naltrexone implants and 52% (11/21) of those randomised to methadone maintenance commenced treatment as randomised. Of the four non-randomised studies, two compared naltrexone implants with OST (one methadone maintenance, the other buprenorphine maintenance).These suffered from limitations in methodology and statistical analysis, discussed below. A second randomised trial compared naltrexone implants with treatment as usual, which included a range of interventions and the possibility of no further therapeutic contact [15].This study did not provide information on treatments received by participants in the treatment as usual group. The three remaining randomised trials and two non-randomised studies compared naltrexone implants with either placebo or oral naltrexone [12– 14]. Oral naltrexone has itself not yet been adequately evaluated in the treatment of opioid dependence [1]. Two randomised trials employed samples that may not accurately reflect the opioid-dependent, treatmentseeking population: prisoners [16] and people with concurrent amphetamine and opioid dependence [13]. It is unknown how generalisable results for these studies are to the treatment-seeking opioid-dependent population. The four non-randomised studies included in the analysis of overdose and mortality suffered from significant methodological and analytic limitations. Three studies did not take account of treatment retention in their analyses. Thus, it is not possible to determine whether observed overdoses or deaths occurred while the participant was in treatment or had ceased treatment. The remaining non-randomised study estimated in-treatment mortality by assuming that all participants implanted with naltrexone retained their implant and that implants were protective against mortality for the © 2013 Australasian Professional Society on Alcohol and other Drugs

reported average active life of the implants. It is unclear how reliable these assumptions were. There were also concerns about how cohorts were defined in two non-randomised studies. In these studies, the cohort included only people who: (i) had received a naltrexone implant or methadone maintenance treatment during a specified time period; and (ii) appeared in a hospital database for any reason [17,18]. Excluding person-years of observation for people who did not require treatment in hospital during the observation period would likely lead to overestimation of overdose and mortality rates. Furthermore, these studies excluded from the cohort people who received both naltrexone implants and methadone maintenance treatment. Evidence for specific outcomes A summary of the results for each outcome is presented below. Additional information for each outcome, including details of how studies defined the outcome and dealt with missing data, is provided in the Supplementary Materials. Induction to treatment as randomised Induction of participants to treatment as randomised was reported in all five randomised trials. Risk of bias for this outcome was low in four studies and unclear in the fifth (Table 2). There was no significant difference in induction to treatment between naltrexone implants and placebo implants [RR 1.00; 95% confidence interval (CI) 0.98, 1.02; number of trials (k) = 2], oral naltrexone (RR 1.00; 95% CI 0.98, 1.02; k = 2), methadone maintenance treatment (RR 1.33 95% CI 0.81, 2.17; k = 1) or treatment as usual (RR 0.91; 95% CI 0.80, 1.04; k = 1) (see Figure S1 for forest plot). Retention in treatment Two randomised trials with three comparisons with naltrexone implants assessed retention in treatment. Risk of bias for this outcome was high in one study [12] and unclear in the other [13] (Table 2). Definitions of retention varied between studies (see Supplementary Materials for details), and there was significant heterogeneity in results for this outcome. Naltrexone implants showed significantly better treatment retention than placebo implants (RR 3.20; 95% CI 2.17, 4.72; k = 2) and oral naltrexone (RR 3.38; 95% CI 2.08, 5.49; k = 1) (see Figure S2 for forest plot). Opioid use All five randomised trials presented data on opioid use during treatment. Risk of bias was high in three trials,

NR, not reported; N/A, Not applicable.

Inadequate random sequence generation Inadequate allocation concealment Inadequate blinding (subjective outcomes) Inadequate blinding (objective outcomes) Selective reporting Incomplete outcome data Induction as randomised Retention in treatment Opioid use Non-opioid drug use Adverse site-related events Other adverse events/effects Opioid overdose Non-opioid drug overdose Mortality Non-comparable cohorts Non-representativeness of exposed cohort Selection of the non-exposed cohort Inadequate ascertainment of exposure Risk of bias for each outcome, summarised across risk of bias domains Induction as randomised retention in treatment Opioid use Non-opioid drug use Adverse site-related events Other adverse events/effects Opioid overdose Non-opioid drug overdose Mortality

Bias due to Low risk Low risk Low risk Low risk High risk Low risk Low risk Low risk NR High risk High risk NR NR Unclear risk Low risk Unclear risk Low risk Low risk

Low risk High risk Low risk NR High risk High risk NR NR Unclear risk

Low risk NR Low risk Low risk High risk High risk Unclear risk NR NR Low risk Unclear risk Low risk Low risk

Low risk NR High risk Low risk High risk High risk Unclear risk NR NR

Krupitsky et al., 2012[12]

Low risk Low risk Low risk Low risk High risk

Hulse et al., 2009[14]

Low risk NR High risk High risk NR High risk High risk NR Low risk

Low risk NR Low risk Low risk NR Unclear risk Low risk NR Low risk Low risk Low risk Low risk Low risk

Low risk Low risk High risk Low risk High risk

Kunøe et al., 2009[15]

Low risk NR High risk High risk NR High risk NR NR Low risk

Low risk NR Low risk Low risk NR High risk NR NR Low risk Low risk High risk Low risk Low risk

Low risk Low risk High risk Low risk Low risk

Lobmaier et al., 2010[20]

Randomised controlled trials

Table 2. Risk of bias assessment

Unclear risk Unclear risk Unclear risk Unclear risk High risk High risk NR NR Unclear risk

Low risk Low risk Low risk Low risk High risk High risk NR NR Low risk Low risk High risk Low risk Low risk

Low risk Unclear risk N/A Low risk Low risk

Tiihonen, 2012[13]

N/A N/A N/A N/A N/A N/A NR NR High risk

N/A N/A N/A N/A N/A N/A NR NR Low risk Low risk Low risk Low risk Unclear risk

High risk High risk N/A Low risk Unclear risk

Kelty & Hulse, 2012 [6]

N/A N/A N/A N/A N/A N/A High risk High risk NR

N/A N/A N/A N/A N/A N/A High risk High risk NR Low risk Unclear risk Low risk Unclear risk

High risk High risk N/A Low risk High risk

Ngo et al., 2008[21]

N/A N/A N/A N/A N/A N/A NR NR High risk

N/A N/A N/A N/A N/A N/A NR NR Low risk Low risk Low risk Low risk Unclear risk

High risk High risk N/A Low risk Unclear risk

Reece, 2010[19]

Non-randomised studies

N/A N/A N/A N/A N/A N/A NR NR High risk

N/A N/A N/A N/A N/A N/A NR NR Low risk High risk Low risk Low risk Unclear risk

High risk High risk N/A Low risk Unclear risk

Tait et al., 2008 [7]

Systematic review of naltrexone implants for opioid dependence 121

© 2013 Australasian Professional Society on Alcohol and other Drugs

122

S. Larney et al.

unclear in one trial and low in one trial (Table 2). Two trials assessed opioid use by urinalysis at a specific time point (10 weeks [13] and six months [12]).The remaining trials presented self-reported data on opioid use throughout the follow-up period. Naltrexone implants were significantly better than placebo (RR 0.57; 95% CI 0.48, 0.68; k = 2) or oral naltrexone (RR 0.57; 95% CI 0.47, 0.70; k = 2) at suppressing opioid use (Figure 3a). There was no significant difference between naltrexone implants and treatment as usual when opioid use was considered as a categorical outcome (RR 0.87; 95% CI 0.61, 1.26; k = 1) (Figure 3a). However, when the dependent variable was days of opioid use during follow-up, there was a significant effect in favour of naltrexone implants over treatment as usual (standardised mean difference −0.82; 95% CI −1.36, −0.27; k = 1) (Figure 3b). There was no significant difference between naltrexone implants and methadone maintenance treatment in mean days of opioid use in the past 30 days at 6-month follow-up (standardised mean difference −0.33; 95% CI −1.47, 0.04 days; k = 1) (Figure 3b). [Corrections added on 9 January 2014, after first online publication: the reported result of −82 has been corrected to −0.82, and the words ‘days’ have been removed from the parenthesis.] Non-opioid drug use Use of illicit drugs other than opioids was reported in four randomised trials. Risk of bias was high in two trials, unclear in one trial and low in one trial (Table 2). The types of drugs and level of use specified for this outcome varied across trials. In some cases, data were inconsistently described within the published study. Patients with naltrexone implants were significantly more likely than those on oral naltrexone to use nonopioid drugs (RR 1.23; 95% CI 1.01, 1.51; k = 1) (see Figure S3 for forest plot). There was no significant difference between naltrexone implants and any other comparator in the use of non-opioid drugs (placebo implants RR 0.79; 95% CI 0.60, 1.04; k = 1, methadone maintenance treatment amphetamine mean difference 2.50; 95% CI −3.88, 8.88; k = 1, benzodiazepine mean difference 2.00 days; 95% CI −4.51, 8.51; k = 1, treatment as usual mean difference 6.4 days; 95% CI 0.6, 12.3 days; k = 1). Site-related adverse events Three randomised trials reported on site-related adverse events [12–14]. All three trials were at high risk of bias, particularly related to loss to follow-up (Table 2). Recipients of naltrexone implants were significantly more likely than recipients of placebo implants to report site-related adverse events (RR 4.68; 95% CI 1.63, 13.44; k = 3) (see Figure S4 for forest plot). © 2013 Australasian Professional Society on Alcohol and other Drugs

Adverse events and effects possibly related to treatment All five randomised trials reported adverse events and effects possibly related to treatment, but only two of these trials reported data that could be included in meta-analyses. All five trials were at high risk of bias for this outcome (Table 2). There was no significant difference in the prevalence of adverse events between naltrexone implants and either placebo implant (RR 2.67; 95% CI 0.73, 9.77; k = 1) or oral naltrexone (RR 2.14; 95% CI 0.84, 5.47; k = 2); however, the upper values of these CIs suggest the possibility of substantial differences that were undetected in these trials (see Figure S5 for forest plot). Non-fatal opioid overdose Two randomised trials and one non-randomised study reported data on non-fatal opioid overdose. One randomised trial was judged to be at unclear risk of bias, and the remaining two studies were at high risk of bias (Table 2). Hulse et al. reported that there were no opioid overdoses requiring emergency department treatment or hospital admission in persons receiving either a naltrexone implant or oral naltrexone during 6 months of follow-up [14]. In Kunøe et al., there were no significant differences in number of self-reported overdoses (drug not specified): 3 of 29 naltrexone participants and four of 27 treatment as usual (TAU) participants reported an overdose (RR 0.23 95% CI 0.03, 1.95) [15]. One non-randomised study examined non-fatal opioid overdose following naltrexone implantation or methadone maintenance treatment [18]. This study, however, did not specify whether overdoses occurred in- or out-of-treatment, precluding any comparison of in-treatment overdose rates between treatments. Non-fatal non-opioid drug overdose One non-randomised study assessed non-fatal nonopioid drug overdose [18]. This study was judged to be at high risk of bias (Table 2). The study compared non-fatal non-opioid drug overdose after naltrexone implantation or entry to methadone maintenance treatment. As noted above, it did not differentiate between overdoses that occurred in- or out-of-treatment, precluding analysis of in-treatment overdose rates. Mortality Of the five randomised trials, four reported mortality during follow-up, but only three provided data suitable for inclusion in meta-analysis. Two randomised trials were at low risk of bias and two at unclear risk of bias

Figure 3. Forest plots of naltrexone implants versus comparators for opioid use during treatment, measured (a) dichotomously (yes/no opioid use during treatment) and (b) continuously (days of opioid use in the past 30 days). CI, confidence interval; OP, oral placebo; PI, placebo implant; RR, risk ratio; SMD, standardised mean difference.

Systematic review of naltrexone implants for opioid dependence 123

© 2013 Australasian Professional Society on Alcohol and other Drugs

124

S. Larney et al.

(Table 2). Krupitsky et al. reported that ‘We found no evidence of increased risk of death due to overdose after naltrexone treatment’ [12], but no mortality data (either cause-specific or all-cause) were reported.There were no deaths in either arm of studies comparing naltrexone implants with placebo [13] or methadone maintenance treatment [16]. There was no statistically significant difference in mortality rates between naltrexone implants and treatment as usual (RR 0.93; 95% CI: 0.06, 14.22); however, the width of the CI around this estimate is such that it is possible that a substantial difference may exist but was not detected in this trial (see Figure S6 for forest plot). Three non-randomised studies that reported mortality data were considered to be at high risk of bias (Table 2). Reece compared mortality rates for individuals implanted with naltrexone with that in individuals receiving buprenorphine maintenance treatment.There were no significant differences between groups in terms of overall mortality (rate ratio 0.56; 95% CI 0.17, 1.44), in-treatment mortality (rate ratio 1.44; 95% CI 0.05, 13.51) or out-of-treatment mortality (rate ratio 0.64; 95% CI 0.20, 1.67). Time in treatment with naltrexone implants was calculated on the basis of reported average active life (defined as blood naltrexone >1 ng ml−1) of the implants [22], but there is substantial variation around this average [22]. There is also disagreement regarding the minimum blood naltrexone concentration required for therapeutic effect [21], and some participants may have had their implants removed while still active.The validity of the calculated mortality rate in the presence of an active naltrexone implant is therefore unclear. Two further non-randomised studies compared mortality after naltrexone implantation with oral naltrexone and methadone maintenance treatment [6,17]. Neither of these studies distinguished deaths that occurred inor out-of-treatment. Discussion Summary of findings A summary of findings is presented in Table 3. There was evidence of moderate quality that induction to treatment did not differ between naltrexone implants and any comparator. There was low-quality evidence that naltrexone implants were superior to placebo implants and oral naltrexone in retaining participants in treatment. No randomised trials have compared retention in naltrexone implants with that in OST.There was low-quality evidence that naltrexone implants were superior to placebo implants and oral naltrexone in suppressing opioid use. Naltrexone implants were superior to treatment as usual in suppressing opioid use. [Correction added on 9 January 2014, after first © 2013 Australasian Professional Society on Alcohol and other Drugs

online publication: ‘but the observed difference was of uncertain clinical significance (a mean difference in days of opioid use of less than 1 day)’ has been deleted from the preceding statement.] There was no difference observed in opioid use between naltrexone implants and OST (low-quality evidence). Placebo implants were associated with significantly fewer surgical site-related adverse events than naltrexone implants (moderate-quality evidence). There was no significant difference in rates of other adverse events between naltrexone implants and placebo implants or oral naltrexone (moderate-quality evidence). There were insufficient data to determine if adverse event rates differ between naltrexone implants and either treatment as usual or OST. In randomised trials, there was no significant difference in the prevalence of opioid overdose between naltrexone implants and oral naltrexone or treatment as usual (moderate-quality evidence). The only nonrandomised study examining opioid overdose was of very low quality and did not provide sufficient data to determine if this outcome differed between naltrexone implants and OST. There was low-quality evidence from randomised trials and very low-quality evidence from a non-randomised study reporting no difference in mortality between naltrexone implants and any comparator during treatment. Limitations of included studies As was the case in the earlier Cochrane and NHMRC systematic reviews, the studies included in our review provide limited evidence on the efficacy, effectiveness and safety of naltrexone implants. The utility of these studies for informing clinical practice is therefore limited. Only one randomised trial compared naltrexone implants with conventional OST (in the form of methadone maintenance treatment). Two trials compared naltrexone implants with oral naltrexone [12,14], but these findings do not inform clinical decisions where an effective treatment such as OST is available. Similarly, findings that naltrexone implants are superior to placebo are not clinically useful where an effective treatment exists. Ethical concerns have been raised over the use of placebo comparisons in naltrexone studies [23]. The Declaration of Helsinki, which sets ethical standards for medical research on humans, states that the benefits and risks of new interventions ‘must be tested against those of the best current proven intervention’, except where no proved intervention exists or there are compelling methodological reasons for the use of a placebo control [24]. There are proved interventions for opioid dependence: methadone and buprenorphine have both been listed by the World Health

Unclearf Unclearg

Unknown Unknown Unknown Unknown Uncleare Unknown

Moderate Very low — Very low

No difference Unknown

Oral naltrexone superior N/A No difference

No difference Unknown

Unknown Unknown

No difference Unknown

No difference Unknown Unclear; naltrexone implant possibly superiorb No difference N/A Unclearc

Treatment as usual

Uncleare Unclear; possibly no differenceg

Unknown Uncleard

Unknown Uncleard

No difference N/A Unclearc

No difference Unknown No difference

Opioid substitution treatment

a Interpretation of quality of evidence levels: High:We are very confident that the true effect lies close to the estimated effect. Moderate:The true effect is likely to be close to the estimated effect, but there is a possibility that it is substantially different. Low: The true effect may be substantially different from the estimated effect. Very low: The true effect is likely to be substantially different from the estimated effect. bThe only study investigating this comparison found no difference between groups when data were analysed dichotomously, and naltrexone implant superior when data were analysed continuously. cThe only studies investigating these comparisons did not report numeric results or results by group. dThe only study to investigate this comparison reported overdose rates after commencing treatment but did not directly measure overdose rates during treatment (i.e. no data on treatment retention included in analysis). eNo deaths in either study arm. fThe only study investigating this comparison did not report any data for all-cause or cause-specific mortality, or any information at all regarding all-cause mortality. gIn one study making this comparison, it is unclear how reliably mortality in the presence of an active naltrexone implant was determined [19].The other study to investigate this comparison reported mortality rates after commencing treatment but did not directly measure mortality rates during treatment (i.e. no data on treatment retention included in analysis) [7]. No difference, no statistically significant difference between naltrexone implants and the comparator was identified; Unknown, this comparison has not been investigated. Unclear, this comparison has been investigated, but no clear interpretation of results is possible. N/A, comparison not applicable.

Low Very low

Unknown Unknown

No difference Placebo implant superior No difference

Low Moderate Moderate

No difference Naltrexone implant superior Naltrexone implant superior

Non-opioid use Site-related adverse events Adverse events/effects Opioid overdose Randomised studies Non-randomised studies Non-opioid overdose Randomised studies Non-randomised studies Mortality Randomised studies Non-randomised studies

No difference Naltrexone implant superior Naltrexone implant superior

Moderate Low Low

Oral naltrexone

Induction as randomised Treatment retention Opioid use

Placebo implant

Quality of evidencea

Outcome

Naltrexone implant compared with

Table 3. Summary of findings

Systematic review of naltrexone implants for opioid dependence 125

© 2013 Australasian Professional Society on Alcohol and other Drugs

126

S. Larney et al.

Organization as essential medicines in the treatment of opioid dependence [25]. There are also no good scientific reasons for using a placebo control. Unavailability of alternative pharmacotherapies in Russia, where the placebo-controlled trials were conducted [12,13], raises the question as to why these studies were not undertaken in a setting with access to standard treatments [23]. Randomised trials of treatments for drug dependence can be difficult to undertake successfully in part because trial entrants may drop out if not randomised to their preferred treatment [26,27]. The only study to compare naltrexone implants with methadone maintenance treatment suffered from high post-randomisation dropout, with treatment preference a factor in dropouts in both the naltrexone and methadone groups [16,20]. Future trials comparing naltrexone implants and convention opioid agonist treatment should build into their design approaches that manage participants’ expectations regarding treatment allocation and ensure an adequate sample size that takes into account likely dropout rates [20,27]. A major limitation in the evidence on the safety of naltrexone implants is the risk of bias arising from incomplete data. Treatment dropout during follow-up was common across trials, and several trials did not report if or how adverse effects and events were assessed in participants who dropped out. This omission is particularly important because participants may drop out because of side effects or other adverse events. One trial explicitly contacted all participants to determine vital status at the end of follow-up [13], but it did not ask about other adverse events. We identified a follow-up study of adverse events in the Hulse et al. trial that attempted to account for events in participants who dropped out of the trial. The follow-up study used record linkage to identify emergency department presentations and hospital admissions among participants during the follow-up period, including participants who dropped out [28]. These data were not included in the results here as all emergency department and hospital events were reported, rather than study-related adverse events as specified in the review protocol and in the original trial report [14]. The original trial report also did not define how adverse events in the trial were determined to be study related, so we were unable to identify study-related adverse events in the follow-up study. Ascertainment of exposure to naltrexone implants and comparators was a major limitation of the nonrandomised studies reviewed. Only one of the four nonrandomised studies attempted to incorporate data on treatment retention into analyses of overdoses and mortality, and major problems were noted with ascertainment of naltrexone implant exposure in that study. The © 2013 Australasian Professional Society on Alcohol and other Drugs

remaining three studies were unable to contribute data on overdoses or mortality while in treatment or following treatment cessation. Observational studies can be critical in identifying long-term or rare outcomes that are not apparent from clinical trials but only if they accurately specify whether deaths and adverse events occur in- or out-of-treatment. Limitations of this review Few studies were eligible for inclusion in this review, and there were differences between studies in terms of comparators, how outcomes were defined and how results were reported. These differences limited the extent to which data could be formally meta-analysed. We did not specify a particular brand or duration of implant for investigation in this review. Two studies used Prodetoxon [12,13], an implant registered for use in Russia. It contains 1 g of naltrexone and reportedly blocks the effects of opioids for ‘60 to 90 days’ [12]. One study used implants manufactured by Wedgewood Pharmacy [19].These are unregistered, contain 1.4 g of naltrexone and are reported to last for 30 days [19].The remaining studies used the Australian ‘O’Neil implant’, which is manufactured by Go Medical Industries under the Australian Code of Good Manufacturing Practice but is not registered for clinical use [14]. The O’Neil implant is available in ‘single’ and ‘double’ versions. It was originally reported by the manufacturer that the single implant contained 1.7 g of naltrexone, and the double implant contained 3.4 g of naltrexone [22]. Although the method of manufacture has not changed, the method used for measuring the naltrexone content of the implants was revised, and the manufacturer now states that the amount of naltrexone per single implant is ‘between 1.1 and 1.2 g’ [21]. The average duration of blood naltrexone greater than 2 ng ml−1 (a level commonly assumed sufficient to block opioids) is reported as 95 days (95% CI 69–121 days) for a single implant and 136 days (95% CI 114–158 days) for a double implant [21]. A duration of ‘5–6.5 months’ has also been reported for the double implant [29]. Full pharmacokinetic data are not available for all the described naltrexone implants, and review of pharmacokinetic and pharmacodynamic data is beyond scope of the current review. Therefore, we were unable to account for possible effects of varying duration of activity of the implants used in the reviewed studies. Implications This review has highlighted the limited quantity and quality of evidence for the use of naltrexone implants in the treatment of opioid dependence. Although one naltrexone implant has been registered

Systematic review of naltrexone implants for opioid dependence

for use in Russia, they remain an experimental product elsewhere. As such, they should only be used in the context of carefully monitored clinical trials [4,30] that aim to establish safety and efficacy of naltrexone implants, either alone or in comparison to current, effective treatment for opioid dependence such as OST. Any such trials must be approved by a properly constituted human research ethics committee or institutional review board; should be prospectively registered with an appropriate clinical trials register; and should be monitored by an independent safety committee. Although we have noted concerns about generalisability in several reviewed studies, it is important to note that naltrexone implants may be best suited to particular sub-groups of opioid-dependent people. Substitution therapy with opioid agonists is the mainstay of treatment for opioid dependence, but abstinence-based treatment can be clinically indicated, for example, in cases of opioid-dependent physicians and other health-care workers [31,32], or patients with relatively short histories of drug use. Abstinence may also be preferred by patients who are unwilling to be prescribed opioids for long-term maintenance. Further research is required to determine the place for naltrexone implants in the treatment armamentarium for opioid dependence.

Conclusions Naltrexone implants may be an alternative to conventional opioid agonist treatment for patients motivated towards abstinence, but more and better designed research trials are needed to establish their safety and efficacy in comparison with existing effective treatments such as OST. Non-randomised studies are likely to be useful in evaluating some important treatment outcomes. While naltrexone implants remain an experimental product, their use should be limited to approved, registered, independently monitored clinical trials.

Funding Dr Sarah Larney is supported by a National Health and Medical Research Council (NHMRC) Early Career Fellowship. Professors Richard P. Mattick and Louisa Degenhardt are supported by NHMRC Principal Research Fellowships. Professor Wayne Hall is supported by an NHMRC Australia Fellowship. The National Drug and Alcohol Research Centre at the University of NSW is supported by funding from the Australian Government under the Substance Misuse Prevention and Service Improvements Grant Fund.

127

Acknowledgements Thank you to Ms Mary Kumvaj for conducting the initial literature search; Dr Jessica Belcher for assisting with data extraction; Professor Evgeny Krupitsky, who provided additional trial data on opioid use during follow-up on request; and Dr Josiah D. Rich for reviewing the manuscript prior to submission. Conflicts of interest Louisa Degenhardt and Richard P. Mattick have received untied educational grants from Reckitt Benckiser for the conduct of post-marketing surveillance of buprenorphine-naloxone for the treatment of opioid dependence. Reckitt Benckiser had no knowledge of this study. References [1] Minozzi S, Amato L, Vecchi S, Davoli M, Kirchmayer U, Verster A. Oral naltrexone maintenance treatment for opioid dependence. Cochrane Database Syst Rev 2011; 4:CD001333. [2] Krupitsky E, Zvartau E, Woody GE. Use of naltrexone to treat opioid addiction in a country in which methadone and buprenorphine are not available. Curr Psychiatry Rep 2010;12:448–53. [3] Lobmaier P, Kornør H, Kunøe N, Bjørndal A. Sustainedrelease naltrexone for opioid dependence. Cochrane Database Syst Rev 2008;(2):CD006140. [4] NHMRC: National Health and Medical Research Council. Naltrexone implant treatment for opioid dependence. Canberra: National Health and Medical Research Council, 2010. [5] Reece AS. Epidemiologic and molecular pathophysiology of chronic opioid dependence and the place of naltrexone extended-release formulations in its clinical management. Subst Abuse 2012;6:115–33. [6] Kelty E, Hulse G. Examination of mortality rates in a retrospective cohort of patients treated with oral or implant naltrexone for problematic opiate use. Addiction 2012; 107:1817–24. [7] Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6:e1000097. [8] Higgins JPT, Green S, eds. 2011. Cochrane handbook for systematic review of interventions version 5.1.0. Available at: http://www.cochrane-handbook.org (accessed August 2013). [9] Cochrane Collaboration. Review Manager (RevMan) Version 5.2. 5.0 ed. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration; 2012. [10] Gowing L, Ali R, White JM. Opioid antagonists under heavy sedation or anaesthesia for opioid withdrawal. Cochrane Database Syst Rev 2010;1:CD002022. [11] Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol 2011;64:401–6. [12] Krupitsky E, Zvartau E, Blokhina E, et al. Randomized trial of long-acting sustained-release naltrexone implant vs oral naltrexone or placebo for preventing relapse to opioid dependence. Arch Gen Psychiatry 2012;69:973–81. © 2013 Australasian Professional Society on Alcohol and other Drugs

128

S. Larney et al.

[13] Tiihonen J, Krupitsky E, Verbitskaya E, et al. Naltrexone implant for the treatment of polydrug dependence: a randomized controlled trial. Am J Psychiatry 2012;169: 531–6. [14] Hulse G, Morris N, Arnold-Reed D, Tait RJ. Improving clinical outcomes in treating heroin dependence: randomized, controlled trial of oral or implant naltrexone. Arch Gen Psychiatry 2009;66:1108–15. [15] Kunøe N, Lobmaier P, Vederhus JK, et al. Naltrexone implants after in-patient treatment for opioid dependence: randomised controlled trial. Br J Psychiatry 2009;194: 541–6. [16] Lobmaier P, Kunøe N, Gossop M, Katevoll T, Waal H. Naltrexone implants compared to methadone: outcomes six months after prison release. Eur Addict Res 2010;16:139– 45. [17] Tait RJ, Ngo HTT, Hulse G. Mortality in heroin users 3 years after naltrexone implant or methadone maintenance treatment. J Subst Abuse Treat 2008;35:116–24. [18] Ngo HTT, Tait RJ, Hulse GK. Comparing drug-related hospital morbidity following heroin dependence treatment with methadone maintenance or naltrexone implantation. Arch Gen Psychiatry 2008;65:457–65. [19] Reece AS. Favorable mortality profile of naltrexone implants for opiate addiction. J Addict Dis 2010;29:30–50. [20] Lobmaier PP, Kunoe N, Wall H. Treatment research in prison: problems and solutions in a randomized trial. Addiction Res Theor 2010;18:1–13. [21] Ngo HTT, Arnold-Reed DE, Hansson RC, Tait RJ, Hulse G. Blood naltrexone levels over time following naltrexone implant. Prog Neuropsychopharmacol Biol Psychiatry 2008;32:23–8. [22] Hulse G, Arnold-Reed DE, O’Neil G, Chan CT, Hansson RC, O’Neil P. Blood naltrexone and 6-beta-naltrexol levels following naltrexone implant: comparing two naltrexone implants. Addict Biol 2004;9:59–65. [23] Wolfe D, Carrieri MP, Dasgupta N, Wodak A, Newman R, Bruce RD. Concerns about injectable naltrexone for opioid dependence. Lancet 2011;377:1468–70. [24] World Medical Association. 2013. Declaration of Helsinki Ethical Principles for Medical Research Involving Human Subjects. Available at: http://www.wma.net/e/policy/pdf/ 17c.pdf (accessed August 2013). [25] WHO: World Health Organization. WHO model list of essential medicines. Geneva: World Health Organization, 2013. [accessed September 2013] (Available at: http:// www.who.int/medicines/publications/essentialmedicines/ en/). [26] Bale RN, Van Stone WW, Kuldau JM, Engelsing TMJ, Elashoff RM, Zarcone VP. Therapeutic communities vs methadone maintenance. A prospective controlled study of narcotic addiction treatment: design and one year followup. Arch Gen Psychiatry 1980;37:179–93.

© 2013 Australasian Professional Society on Alcohol and other Drugs

[27] Thomson CL, Morley KC, Teesson M, Sannibale C, Haber PS. Issues with recruitment to randomised controlled trials in the drug and alcohol field: a literature review and Australian case study. Drug Alcohol Rev 2008;27:115– 22. [28] Kelty E, Ngo HT, Hulse G. Assessing the usefulness of health data linkage in obtaining adverse event data in a randomised controlled trial of oral and implant naltrexone in the treatment of heroin dependence. Clin Trials 2012; 10:170–80. [29] Waal H, Frogopsahl G, Olsen L, Christophersen AS, Mørland J. Naltrexone implants—duration, tolerability and clinical usefulness: a pilot study. Eur Addict Res 2006; 12:138–44. [30] Australasian Chapter of Addiction Medicine of the Royal Australasian College of Physicians. Position statement on the use of sustained release formulations of naltrexone in opioid dependence. Royal Australasian College of Physicians; 2013. Available at: http://www.racp.edu.au/index.cfm ?objectid=2088B865-B608-1312-3DE57C28E5CED0A7 (accessed September 2013). [31] Merlo LJ, Greene WM. Pomm R. Mandatory naltrexone treatment prevents relapse among opiate-dependent anesthesiologists returning to practice. J Addict Med 2011; 5:279–83. [32] Roth A, Hogan I, Farren C. Naltrexone plus group therapy for the treatment of opiate-abusing health-care professionals. J Subst Abuse Treat 1997;14:19–22.

Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Figure S1. Forest plot of naltrexone implants versus comparators for induction to treatment as randomised. Figure S2. Forest plot of naltrexone implants versus comparators for retention in treatment. Figure S3. Forest plot of naltrexone implants versus comparators for non-opioid use (dichotomous data). Figure S4. Forest plot of comparison of risk of adverse site-related events: Naltrexone implants versus placebo implants. Figure S5. Forest plot of naltrexone implants versus comparators for other adverse events and effects possibly related to treatment. Figure S6. Forest plot of naltrexone implants versus comparators for mortality.

A systematic review and meta-analysis of naltrexone implants for the treatment of opioid dependence.

Naltrexone implants are used to treat opioid dependence, but their safety and efficacy remain poorly understood. We systematically reviewed the litera...
688KB Sizes 0 Downloads 0 Views