Extended‐release pharmacotherapies for substance use disorders in incarcerated populations: A systematic review

Abstract

Background and aims

Substance use (SU) is prevalent among individuals in the criminal justice system (CJS). However, there is often poor access to treatment. We aimed to assess the effectiveness of two medications, extended‐release naltrexone (XR‐NTX) and extended‐release buprenorphine (XR‐BUP) for the prison population.

Methods

We searched Scopus, OVID/Embase, PubMed/Medline, ProQuest, EBSCO, Cochrane Library and Australian Criminology Database for original articles published from 1 January 2002 to 31 December 2022. Inclusion criteria: 18+, substance use disorder; XR treatment; recent incarceration. We extracted study, participants, treatment characteristics and outcome variables. We conducted risk of bias assessments using the RoB‐2, ROBINS‐I, JBI tools and Evers et al.

Results

We identified 25 papers (16 studies) examining 3403 participants. Sixteen papers (9 studies) focused on XR‐NTX, eight (7 studies) on XR‐BUP and one on both. Eighteen papers (11 studies) were from the US, with the remainder from Norway, Australia, UK, Canada and Germany. There were eight RCTs (10 papers), four secondary observational analyses, four cohort studies, four economic analyses, two case series and one qualitative paper. Most studies had small–moderate samples, with varying retention and follow‐up periods. Among RCTs, two XR‐NTX studies for opioid use found no difference in retention vs treatment as usual and placebo, while one reported improved retention for XR‐NTX implant vs methadone. One RCT showed mixed retention results for XR‐NTX vs placebo in alcohol use. One XR‐BUP study showed improved or equivalent treatment retention (depending on measures) vs sublingual buprenorphine. There was no difference in overdoses. SU for XR‐NTX was challenging to assess due to differing definitions, measures and comparators. XR‐BUP yielded mixed SU results, with one indicating a greater effect and another no difference from comparators.

Conclusions

There is no clear evidence for the effectiveness of extended release naltrexone and buprenorphine among individuals in the criminal justice system compared with shorter acting formulations. But there is growing evidence for the effectiveness of extended release buprenorphine in reducing opioid use and improving treatment retention in that population, with potential cost offsets from initial medication expenses.

INTRODUCTION

People in the criminal justice system (CJS) face significant health challenges, including mental health conditions, chronic diseases and substance use disorders (SUDs) [1, 2, 3, 4]. SUDs exacerbate health issues, contributing to cycles of reincarceration [5, 6, 7, 8], and limiting reintegration into employment, relationships and society. One study found that among people with an opioid use disorder (OUD), 37% (43% of men and 24% of women) had experienced at least one episode of incarceration [9]. OUDs also increase the risk of blood borne virus transmission [10, 11] and overdose [12, 13, 14, 15]. Approximately 24% of the prison population [16] are affected by alcohol use disorders (AUDs), which can lead to liver damage, injuries [1] and family violence [17].

The CJS is a critical point of contact with healthcare systems for individuals who may not otherwise seek medical care. Relapse prevention pharmacotherapy can interrupt cycles of substance use (SU) and reincarceration [3, 18, 19], and mitigate physical, social and economic harms [5, 14, 20, 21]. Opioid agonist therapy (OAT), including buprenorphine and methadone, can reduce opioid use [5, 19], overdose [14, 15, 22] and the transmission of blood borne viruses [23, 24]. OATs activate the same neuroreceptors as illicit opioids, but offer a longer and more stable duration of action than drugs like heroin, therefore, reducing cravings and use [19]. Naltrexone, used for AUDs and OUDs [19, 25], is an opioid receptor blocker. It is used instead of OAT in some jurisdictions [26]. Although there are some concerns about naltrexone's effectiveness in harm reduction for opioid use [25, 27], it has reasonable evidence for the reduction of alcohol use [28].

For participants recently released from the CJS, who often face housing and financial instability, daily dosing of medications can be challenging [29, 30]. Additionally, the opioid overdose risk is highest in the first few weeks of release from CJS [12, 14, 15, 31]. Extended‐release (XR) formulations of buprenorphine (XR‐BUP) (administered weekly or monthly) [32] and naltrexone (XR‐NTX) (administered monthly or as a 6‐month implant) [33, 34, 35, 36, 37] can increase the duration of pharmacotherapy action over this high‐risk period, as well as offering improved accessibility, adherence and safety.

The evidence for use of XR formulations in the CJS context, however, remains unclear. Our aim is to address this gap by describing the evidence for effectiveness, safety and feasibility of XR medications for people with SUDs in prison or within 3 months of release.

METHODS

This Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) [38] compliant systematic review was registered with the international Prospective Register of Systematic Reviews (PROSPERO; CRD42021276447) on 1 October 2021.

Search strategy

The search strategy was developed with a librarian specializing in drug and alcohol research. The search was limited to quantitative and qualitative peer‐reviewed studies in English from 1 January 2002 to 31 December 2022. Seven platforms were searched: (a) Scopus; (b) OVID/Embase; (c) PubMed/Medline; (d) Proquest; (e) EBSCO; (f) Cochrane Library; and (g) Australian Criminology Database (Figure 1). Search terms were grouped into four domains: (a) substance dependence; (b) relapse prevention pharmacotherapy; (c) XR‐formulations; and (d) prison‐based participants. Reference lists of included articles were hand‐searched. Title and abstract, English language and human limits were used. A manual search of reference lists of included studies was conducted.

FIGURE 1.

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) flow‐chart.

Study selection

Inclusion criteria were: (a) participants age 18+ with primary treatment focus being SUD ¹; (b) XR (formulations requiring a dosing interval of five or more days) relapse prevention medication; and (c) prison setting or a population released within the past 3 months. Only studies presenting original data or analysis were included. Separate articles reporting on the same sample and interventions were only included if the articles reported on new outcomes. Exclusion criteria were: (a) participants age <18 because of medical approval issues; and (b) primary focus on conditions other than SUD, such as mental health treatments.

Duplicates were removed using EndNote [39]. Two reviewers (A.W. and W.A.D.) screened the title/abstracts independently. Disagreements were resolved by consensus, with a third reviewer (K.C.) available for disagreements. This was repeated for full‐text reviews.

Data extraction

Aims, methods and results were summarised by A.W., following the Cochrane Collaboration Handbook for Systematic Reviews of Health Promotion and Public Health Interventions [40].

Outcome measures

Up to 17 variables were recorded relating to the primary outcomes. Any relevant measurement of SU was extracted for primary and secondary drugs of concern, as was any measurement and definition of retention in treatment (RIT) and recidivism. We extracted all measurements of adverse events (AEs), although focused on those effects that were likely to be treatment related. We extracted all economic results, regardless of the individual measures used. We synthesized any information relating to participants' treatment perceptions. Meta‐analysis was not feasible because of excessive heterogeneity in methods and outcome measures.

Critical appraisal of methods

Quantitative articles were evaluated using the Cochrane Risk‐of‐Bias tool for randomized trials (RoB‐2) [41], the risk of bias in non‐randomised studies of interventions (ROBINS‐I) [42] and the Joanna Briggs Institute's (JBI) tools [43].

A.W. reviewed all articles, and a random sample of 11 articles (44%) was re‐reviewed by a blinded coder (W.A.D.), with discrepancies resolved through discussion. Initial agreement on risk of bias was 72% (n = 8/11), with the remaining three resolved in discussion.

Qualitative articles were evaluated using the JBI Qualitative Research tool, covering 10 domains: philosophical perspective, methodology, question and objectives alignment, data collection, analysis, results, reflexivity, participant representation, ethics and conclusions.

RESULTS

Our search yielded 792 references (Figure 1). After removing 318 duplicates, we screened 474 titles and abstracts and 127 full text articles. Twenty‐five articles (n = 16 studies) were finally included.

Characteristics of included studies

Eleven articles were published from 2010 to 19 [36, 37, 44, 45, 46, 47, 48, 49, 50, 51, 52] (n = 7 studies), and 14 between 2020 and 22 [53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66] (n = 9 studies) (Table 1). There was a total of 3403 participants across the 16 studies. ² Most articles (18/25; 11 studies) were from the United States (US) [36, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 57, 59, 60, 61, 62, 63, 64], and others from Norway [37, 44], Australia [56, 66], United Kingdom [54], Canada [65] and Germany [58].

TABLE 1.

Characteristics of included studies (n = 25).

	Author (y), country		Aim/research question a	Substance focus	Methods (including substance withdrawal status, duration of intervention and FU)	Participants –Total (n)	Gender (n, % male) b Age (mean, SD) [unless otherwise stated]	Ethnicity
	Study	Article
XR‐NTX	Farabee (2020), US [53]	Farabee (2020), US [53]	What is the effectiveness of XR‐NTX compared to treatment‐as‐usual (TAU; fact sheet on overdose, referral list of OAT providers) for inmates with OUD?	Opioids	RCT, open label Presumed withdrawal completed (but not stated) 7 (monthly) injections (1 pre‐release, 6 post‐release) FU: 12 months post‐ release	n = 101 2 intervention groups NI: (XR‐NTX) = 53 c NC: (TAU) = 48	65 (72%) ‐‐‐‐‐‐‐‐‐ Not reported	White = 62/90 (69%) First Nations (American Indian) = 8 (8.9%) African American/Black = 9 (10%)
	Friedmann (2018), US [49]	Friedmann (2018), US [49]	How does pre‐ vs. post‐release commencement of XR‐NTX affect abstinence and antagonist medication coverage in the weeks after release?	Opioids	RCT, pilot study Presumed withdrawal completed (but not stated) 6 (monthly) injections: –Pre‐release group: 1 pre‐release, 5 post‐release –Post‐release group: 6 post‐release FU: Urine drug screen (UDS) to week 25; visits to week 78	n = 15 NI: (XR‐NTX pre‐release) = 9 NC: (XR‐NTX pre‐release) = 6	14 (93%) ‐‐‐‐‐‐‐‐‐ Not reported	Not reported
	Gordon (2015), US [46]	Gordon (2015), US [46]	What is the relationship between treatment adherence, to XR‐NTX and recidivism, and opioid use?	Opioids	Prospective cohort (phase 4 open label trial) Withdrawal status: already completed 7 (monthly) injections (1 pre‐release, 6 post‐release) FU: Not clear, >7 months	n = 27 NI: (‘completers’ i.e. 6 post‐release XR‐NTX injections) = 10 NC: (‘non‐completers’ i.e. <6 post‐release XR‐NTX injections) = 17	16 (59.3%) ‐‐‐‐‐‐‐‐‐ 39.9 (9.3)	White = 4 (14.8) African American/Black = 23 (85.2)
	Lee (2015), US [36]	Lee (2015), US [36]	What is the feasibility and effectiveness of XR‐NTX for relapse prevention among opioid‐dependent male adults leaving jail?	Opioids	RCT, non‐blinded Withdrawal status: already completed 2 (monthly) injections‐ 1 pre‐release, 1 post‐ release FU: 8 weeks post‐release	n = 34 NI: (XR‐NTX) = 17 (1 drop out, analysis done on 16) NC: (placebo) = 17	34 (100%) ‐‐‐‐‐‐‐‐‐ NI: 40 (26–52) NC: 47 (39–58)	Not reported
	Lincoln (2018), US [50]	Lincoln (2018), US [50]	How does pre‐ vs. post‐release commencement of XR‐NTX affect retention in treatment?	Opioid	Cohort study‐ non‐randomised, prospective Withdrawal status: already completed 6 (monthly) injections –Pre‐release group: 1 pre‐release, 5 post‐release –Post‐release group: 6 post‐release FU: 6 months post‐release	n = 67 NI: (pre‐release XR‐NTX) = 47 NC: (post‐release) = 20	60 (90%) ‐‐‐‐‐‐‐‐‐ XR‐NTX before release: 32.9 (22–60) XR‐NTX planned after release: 34.6 (21–54)	White = 60%–68% African American/Black = 3%
	Lobmaier (2010), Norway [44]	Lobmaier (2010), Norway [44]	What are the effects of XR‐NTX on heroin use, non‐opioid drug use and criminal activity after release?	Opioid	RCT–2‐arm, open‐label trial –1/2 articles from this trial (preliminary results) –XR‐NTX withdrawal status: presumed completed (but not stated) –Methadone arm: unclear, likely completed (methadone arm had it newly commenced, participants were excluded if on methadone before trial) Implant (lasting 5–6 months) FU: 6 months	n = 46 NI: (naltrexone implant) = 24 NC: (methadone) = 22	44 (91.6%) ‐‐‐‐‐‐‐‐‐ Not reported (subsequent report—see below)	Not reported
		Lobmaier (2010), Norway [37]	What are the effects of XR‐NTX on heroin use, non‐opioid drug use and criminal activity after release?	Opioid	RCT–2‐arm, open‐label trial –2/2 articles from this trial (final results) Withdrawal status: presumed completed (not stated) Implant (lasting 5–6 months) FU: 6 months	n = 44 NI: (XR‐NTX [implant]) = 23 NC: (methadone) = 21 Note: one less male from each treatment group, than existing Lobmaier article [6] because of withdrawal of research consent before randomisation	3 (93.5%) ‐‐‐‐‐‐‐‐‐ 35.1 (7.0)	Not reported
XR‐NTX in people with HIV d	‘INSPIRE’ study looking at AUDs	Springer (2017), US [48]	What are the effects of XR‐NTX on post‐release alcohol consumption?	Alcohol	RCT‐ double‐blind, placebo‐controlled Withdrawal status: presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) d FU: 6 months post‐ incarceration	n = 100 NI: (XR‐NTX injection) = 67 NC: (placebo) = 33	77 (77%) Trans = 2 (2%) ‐‐‐‐‐‐‐‐‐ 45.01 (8.35)	White = 16 (16%) African American/Black = 65 (65%)
	‘NEW HOPE’ study looking at OUDs	Springer, (2018), US [51]	Does XR‐NTX affect opioid abstinence and time to relapse vs. placebo?	Opioid	RCT–double‐blind, placebo‐controlled –Subsequent analysis Withdrawal status: presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) f FU: 12 months post‐ release	n = 93 NI: (XR‐NTX) = 66 NC: (placebo)	76 (81.7%) ‐‐‐‐‐‐‐‐‐ 45.8 (8.2)	White = 9 (9.7%) African American/Black = 23 (24.7%) Hispanic = 61 (65.6)
		Lier (2022), US [62]	What is the relationship between XR‐NTX and injecting drug use/HIV risk behaviour?	Opioids	Observational study‐secondary analysis Withdrawal status: Presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) d FU: 12 months post‐ release	n = 88 NI: (≥3 doses XR‐NTX) = 21 NC: (2 or less doses XR‐NTX) = 67	71 (80.7%) ‐‐‐‐‐‐‐‐‐ 45.7 (SD 8.28)	White = 9 (10.2) African American/Black = 20 (22.7) Hispanic = 59 (67.1%)
	Combined results from INSPIRE and NEW HOPE	Biondi (2019), US [52]	What are the sexual risk behaviours among people in the CJS with HIV randomised to receive XR‐NTX or placebo?	Alcohol and/or opioids	RCT–double‐blind, placebo‐controlled –Subsequent analysis Withdrawal status: presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) d FU: 12 months post‐ release	n = 193 NI: (XR‐NTX) = 133 NC: (placebo) = 60	153 (79.3%) Trans = 2 (1%) ‐‐‐‐‐‐‐‐‐ 45.4 (8.3)	African American = 88 (45.6%) Hispanic = 80 (41.5%) White = 25 (13%)
		Biondi (2021), US [55]	What were the baseline characteristics by gender of people enrolled in either of 2 clinical trials who were living with HIV, leaving the CJS and being treated with XR‐NTX (compared with placebo)?	Alcohol and/or opioids	Observational study–secondary analysis Withdrawal status: presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) e FU: 12 months post‐ release	n = 193 NI: (XR‐NTX) = 133 NC: (placebo) = 60	153 (79.3) ‐‐‐‐‐‐‐‐‐ Not reported	White = 25 (12.9%) African American/Black = 88 (45.6%) Hispanic = 80 (41.5%)
		Springer, (2015), US [47]	What is the acceptability of XR‐NTX in CJS settings and the retention post‐release?	Alcohol and/or opioids	Observational study Withdrawal status: presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) d FU: 12 months post‐ release	n = 167 f NI: (XR‐NTX) = study authors report the proportion is ‘two‐thirds’ of the total, but do not give precise figures	h (95.60%) ‐‐‐‐‐‐‐‐‐ 44.7	White = 11.1% African American/Black = 47.50%
		Vagenas (2014), US [45]	What contributes to hepatotoxicity among HIV‐infected prisoners with OUD and/or AUD prescribed antiretroviral therapy and enrolled in placebo‐controlled trials of XR‐NTX?	Alcohol and/or opioids	Observational study–secondary analysis Withdrawal status: presumed completed (not stated) 6 (monthly) injections (1 pre‐release, 5 post release) d FU: 12 months post‐ release	n = 85 NI: (XR‐NTX) = 61 NC: (placebo) = 24	67 (79%) ‐‐‐‐‐‐‐‐‐ 46 (median years old)	White = 18.80% African American/Black = 46%
XR‐NTV	Woody (2021), US [59]	Woody (2021), US [59]	Does administering XR‐NTX before release vs. after release result in less relapse within the first 3 months?	Opioids	RCT Withdrawal status: confirmed completed (negative urine test, self‐report, absence of withdrawal symptoms) 4 (monthly) injections –Started both in prison and within 30 days of release FU: 24 weeks (presumably post‐release although not explicitly stated)	n = 146 NI: (pre‐release XR‐NTX) = 74 NC: (post‐release XR‐NTX) = 72	Not reported (see below–subsequent analysis) ‐‐‐‐‐‐‐‐‐ 37.74 (10.01)	Not reported
		Jalali (2022), US [61]	What is the cost effectiveness of XR‐NTX administered to prisoners with OUD pre‐release compared to post‐release?	Opioids	Economic evaluation‐ secondary analysis of Woody, 2021 [15] Withdrawal status: confirmed completed (negative urine test, self‐report, absence of withdrawal symptoms) 4 (monthly) injections –Started both in prison and within 30 days of release FU: 24 weeks	n = 86 NI: (XR‐NTX pre‐release) = 38 NC: (XR‐NTX post‐release) = 48 Note: these numbers differ to the Woody study as t includes only those who commenced treatment, not all who were randomised	63 (73.3%) ‐‐‐‐‐‐‐‐‐ 37.74 (10.01)	White = 52 (60%) African American/Black = 18 (21%)
XR‐BUP	Dunlop (2022), Australia [56]	Dunlop (2022), Australia [56]	What is the safety and tolerability of XR‐BUP in a custodial setting?	Opioids	Cohort study Withdrawal status: –Methadone: currently stable methadone –XR‐BUP: not completed (ongoing, self‐reported OUD and/or illicit sublingual buprenorphine) Injections for 16 weeks (weekly for 4 weeks, then monthly) –Started in prison FU: 16 weeks	n = 129 NI: (XR‐BUP) = 67 NC: (methadone) = 62	108 (83.7%) ‐‐‐‐‐‐‐‐‐ Not reported	Aboriginal = 38% No further breakdown
		Ling (2022), Australia [66]	What are the costs of administering XR‐BUP, methadone and SL‐BUP?	Opioids	Economic evaluation Withdrawal status: –Methadone: currently stable methadone –XR‐BUP: not completed (ongoing, self‐reported OUD and/or illicit sublingual buprenorphine) Injections for 16 weeks (weekly for 4 weeks, then monthly) –Started in prison FU: 16 weeks	n = 324 NI: (XR‐BUP) = 50 NC: (methadone +SL‐BUP) = 279 (239 methadone, 40 SL‐BUP) Note: these numbers differ to the Dunlop 2022 article as the comparator group was extended to include the total no. of methadone and SL‐BUP clients, not just those enrolled in the original trial. The XR‐BUP number reflect those who were on XR‐BUP at the point of analysis	Not reported ‐‐‐‐‐‐‐‐‐ Not reported	Not reported
	Lee (2021), US [57]	Lee (2021), US [57]	What is the acceptability and feasibility of XR‐BUP compared to sublingual buprenorphine (SL‐BUP) in jail?	Opioids	RCT–open‐label Withdrawal status: both arms were prescribed SL‐BUP at baseline Injections for 8 weeks post release –Started in prison (variable no. before release) FU: 8 weeks post‐release	n = 52 NI: (XR‐BUP) = 26 NC: (SL‐BUP) = 26	45 (86.5%) ‐‐‐‐‐‐‐‐‐ 42.7	Hispanic = 23 (44%) White = 10 (19%) African American/Black = 13 (25%)
		Cheng (2022), US [60]	What is the treatment experience, treatment satisfaction for XR‐BUP, and barriers and facilitators to XR‐BUP initiation in jail and retention post‐release?	Opioids	Qualitative study–A subsequent analysis of Lee et al., 2021 Withdrawal status: both arms were prescribed SL‐BUP at baseline Injections for 8 weeks post‐release –Started in prison (likely variable duration depending on participant) FU: 8 weeks post‐ release	n = 16 (all XR‐BUP, no comparator)	13 (81.3%) ‐‐‐‐‐‐‐‐‐ 45 (range 30, 55)	Not reported
	Martin (2022), US [63]	Martin (2022), US [63]	What is the retention for XR‐BUP as well as the side effects and reasons for use?	Opioids	Case series g –retrospective Withdrawal status: On SL‐BUP for at least 8 days prior Variable no. of injections –Started in prison FU: Variable	n = 54 (all XR‐BUP, no comparator)	52 (96%) ‐‐‐‐‐‐‐‐‐ 38.7 (8.8)	White = 43 (80%) African American/Black 4 (7%) Hispanic = 7 (13%)
	Soyka (2021), Germany [58]	Soyka (2021), Germany [58]	What is the experience of initiation and stabilisation of XR‐BUP in prisoners on methadone?	Opioids	Case series (retrospective chart review) Withdrawal status: population was stable on methadone before study. Methadone dose was held, then test dose of SL‐BUP given before XR‐BUP administration No. of injections: variable –Started in prison FU: variable	n = 7 (all XR‐BUP, no comparator)	7 (100%) ‐‐‐‐‐‐‐‐‐ Not reported	Not reported
	Wright (2020), UK [54]	Wright (2020), UK [54]	What are the estimated costs of providing XR‐BUP in prisons vs. methadone?	Opioids	Economic evaluation Withdrawal status: N/A (predictive analysis) No. of injections not directly stated‐ article examined predictive annual cost of monthly injections –In prison costs Annual costs	n = 255 NI: (XR‐BUP) = 105 NC: (methadone) = 150	NA (economic analysis) ‐‐‐‐‐‐‐‐‐ NA (predictive model of individual costs)	NA (theoretical sample)
	Wong (2022), Canada [65]	Wong (2022), Canada [65]	What are the costs of medication preparation, administration, monitoring and personnel for XR‐BUP vs. SL‐BUP?	Opioids	Economic evaluation Withdrawal status: not stated (economic analysis) Not reported if intervention commenced during or before incarceration, but all analysis relevant to in‐prison costs only. No. of injections not mentioned FU: 28‐day costs	n = 16 NI: (XR‐BUP) = 3 NC: (SL‐BUP) = 13	Not reported ‐‐‐‐‐‐‐‐‐ Not reported	Not reported
Both	Will (2022), US [64]	Will (2022), US [64]	What is the impact of providing medication for OUD in jail for individuals on ED visits within 28 days of release?	Opioids	Cohort study‐retrospective Withdrawal status: not stated Intervention commenced in prison FU: 28 days post‐release	n = 4352 encounters (2311 individuals) Encounters –Methadone (n = 105) –Naltrexone (n = 21) –XR‐BUP (n = 114) –None (n = 4112)	52 (96%) ‐‐‐‐‐‐‐‐‐ 38.7 (8.8)	White = 43 (80%) African American/Black = 4 (7%)

Abbreviations: AUD, alcohol use disorder; FU, follow‐up; NC, comparator group; NI, intervention group; OAT, opioid agonist therapy; OUD, opioid use disorder; TAU, treatment as usual.

^a Many studies had multiple aims, with the study of the XR relapse prevention medication sometimes only being a secondary aim. We have only included the relevant aim.

^b The majority of studies reference male and female genders only. Where transgender participants were also included in the results (n = 2 articles), the % for both male and transgender patients is reported. The % of female patients is 100% male (and % transgender when available).

^c Additional intervention group XR‐NTX + patient navigator (n = 50). Data not included in our results because the additional variable (patient navigator) was not being explored in this review

^d The group from Yale conducted 2 RCTs examining the use of XR‐NTX in people in the CJS with HIV. One RCT looked at people with AUDs (‘INSPIRE’ study), the other with OUDs (‘NEW HOPE’ study). The data have been grouped here because the authors also presented data on the combined RCTs. In the manuscript they have been counted as separate studies.

^e The articles published from the 2 parent studies intermittently report either that the initial injection was given pre‐release or that inclusion criteria included pre‐release or within 30 days of release. They did not provide a breakdown of the percentage of participants receiving the initial injection pre‐ vs. post‐release.

^f This is the number of people who completed the baseline interview and were deemed eligible for the study injection on release at the time of analysis in the article.

^g Martin et al.. identified this as a retrospective cohort design, however, the lack of comparator group meant that the review authors felt it was more in keeping with a retrospective case series and have referred to it as such throughout the review.

^h There is a discrepancy in the article's reporting of participants' genders

Sixteen articles (n = 9 studies) explored XR‐NTX (injectable or implantable forms) [36, 37, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 59, 61, 62] and eight articles (6 studies) examined XR‐BUP [54, 56, 57, 58, 60, 63, 65]. One study examined both XR‐NTX and XR‐BUP [64]. Of the XR‐NTX studies, one looked exclusively at alcohol use [48], and the remainder looked at opioid use (four articles combined alcohol and opioid use populations [45, 47, 52, 55]).

Twenty‐four articles (n = 16 studies) used quantitative methods, including 10 RCTs [36, 37, 44, 48, 49, 51, 52, 53, 57, 59] (n = 7 studies), four secondary observational analysis of populations enrolled in RCTs [45, 47, 55, 62], one retrospective cohort study [64], three prospective cohort studies [46, 50, 56], four economic analyses [54, 61, 65, 66] and two case series [58, 63]. One study used qualitative methods [60]. Sample sizes ranged from 15 to 2311 individuals, with 16 in the qualitative study. Table 2 provides a summary with the direction of effect on key outcomes.

TABLE 2.

Overview of results for quantitative articles that used a comparator group for analysis (n = 24).

	Author (y)		Aim (brief) Target substance Study type	Direction of change –Relevant outcome
	Study	Article		Retention in treatment	Opioid overdose	Substance use	Recidivism	Adverse events
XR‐NTX	Farabee (2020) [53]	Farabee (2020) [53]	What is the effectiveness of XR‐NTX compared to TAU (fact sheet on overdose, referral list of OAT providers) for OUD? Opioid RCT, open label	Not reported	Not reported	No change (XR‐NTX vs. TAU [no medication]) –In % of participants using opioids at 1, 3, 6 and 12 months –In mean % of days using opioids at 1, 3, 6 and 12 months –In % of participants with OUD at 1, 3, 6 and 12 months	No change [XR‐NTX vs. TAU (no medication)] –Re‐arrest over 1 year follow up post‐ release	No between group comparison
	Friedmann (2018) [49]	Friedmann (2018) [49]	What is the effect of XR‐NTX pre‐ vs. post‐release on opioid abstinence? Opioid RCT, pilot study	Mixed effects (for pre‐vs. post‐release) –Improved in number of participants to receive ≥2 injections –No change at injection 1 or injection 6	Not reported	Reduced (for pre‐ vs. post‐release group) –In days of abstinence in the 4 weeks post release (mean and percentage) No change –In median number of days of abstinence	Not reported	No between group comparison
	Gordon (2015) [46]	Gordon (2015) [46]	What is the relationship between treatment adherence to XR‐NTX and recidivism and opioid use? Opioid Prospective cohort (Phase 4 open label trial)	No between group comparison	No change (0 in both arms)	Reduced (for treatment completers vs. non‐completers) –In % of participants with abstinence	No change (for treatment completers vs. non‐completers) –Rearrest –Reincarceration	No between group comparison
	Lee (2015) [36]	Lee (2015) [36]	Feasibility and effectiveness of XR‐NTX compared to TAU (i.e. no medication) as relapse prevention for OUD Opioid RCT, non‐blinded	No change (XR‐NTX vs. TAU) –In rates of completed study visits vs. dropout at weeks 4 and 8	No change (0 in both arms)	Reduced [for XR‐NTX vs. TAU (no medication)] –Higher % of participants with confirmed abstinence at weeks 4 and 8 –Lower % of participants with opioid relapse at 4 and 8 –% of participants with IDU post release No comparison between groups re post‐ release injecting drug use	No change (XR‐NTX vs. TAU) –Reincarceration rates (stated, no P value included) Reduction [participants who received both injections both injection vs. one or none (from medication or placebo arms)] –Reincarceration rates	No between group comparison
	Lincoln (2018) [50]	Lincoln (2018) [50]	The effects on retention and overdose for pre‐ vs. post‐release injection Opioid Cohort study‐ non‐randomised, prospective	Mixed effects (for pre‐ and post‐release) –Improved at 4 weeks –No change at 24 weeks	No change –Deaths <1 year post release	Not reported	No change (pre‐ and post‐release) –% of population with new arraignment <6 months	Not reported
	Lobmaier, 2010 [37]	Lobmaier, 2010 [37]	The effects on opioid use, non‐opioid drug use and criminal activity of XR‐NTX (implant) vs. methadone Opioid RCT–2‐arm, open‐label trial	No change (XR‐NTX [implant] vs. methadone) –In % participants who attended FU interview at 6 months. Improved [XR‐NTX (implant) vs. methadone] –In % of patients retained on implant‐ Presented same findings as below (in less detail)	Not reported	Note: no new results (raw number for analysis presented in other Lobmaier article)	No change (XR‐NTX vs. methadone) –In mean days per month reincarcerated No test of significance for reported % of participants with 1 or more days in prison during follow‐up	No comparison between groups
		Lobmaier, 2010 [44]	The effects on opioid use, non‐opioid drug use and criminal activity of XR‐NTX (implant) vs. methadone Opioid RCT–2‐arm, open‐label trial	Improved –Higher % of participants still had implant in‐situ at 6 months post‐release than remained on methadone	Not reported	No change [XR‐NTX (implant) vs. methadone] –In mean days of heroin use between NTX and methadone before imprisonment vs. at follow‐up –In rates of heroin relapse (defined as 7 or more days of use) between NTX and methadone based on Kaplan–Meier survival estimates	No comparison between groups	Not reported
XR‐NTX in people with HIVc	‘INSPIRE’ study looking at AUDs	Springer (2017) [48]	The effect of XR‐NTX vs. placebo on post‐release alcohol consumption. Alcohol RCT–double‐blind, placebo‐controlled –Subsequent analysis	No change (XR‐NTX vs. placebo) –In no. of participants to have received all 6 injections Improved –Higher % of XR‐NTX injections administered vs. placebo	Not reported	Reduced (for XR‐NTX vs. placebo) –In composite alcohol score for participants who received ≥4 injections No change (XR‐NTX vs. placebo) –In the average time to first heavy drinking day –Average drinks per drinking day NB: Composite score was calculated based on: –Time to first heavy drinking day –Pre‐ and post‐release alcohol consumption –Average drinks per drinking day –% of heavy drinking days –(Total number of drinking days)	No change (XR‐NTX vs. placebo) –In number with no reincarcerations	No change (numbers not presented)
	‘NEW HOPE’ study looking at OUDs	Springer (2018) [51]	HIV viral suppression levels following XR‐NTX for OUD vs. placebo Opioids RCT–double‐blind, placebo‐controlled –Subsequent analysis	Not reported	Not reported	No change (XR‐NTX vs. placebo) –In time to first opioid use (mean days; ITT) Decreased use (for 3 or more XR‐NTX injections vs. placebo or ≤2 injections) –Longer time to first opioid use (mean days)	Not reported	No change (XR‐NTX vs. placebo) –In any individual drug‐related events
		Lier (2022) [62]	Injecting drug use/HIV risk behaviour following XR‐NTX compared to placebo Opioids Observational study–secondary analysis	No change (XR‐NTX vs. placebo) –In attendance at 6‐month study interview	Not reported	No change (XR‐NTX vs. placebo) –In the mean decline in proportion of opioid injection days (1 month post release) –In the mean decline in proportion of opioid injection days (7‐ and 12‐months post release)—statement but no numbers presented –In injecting drug use from pre‐incarceration period to the 1‐month post‐release period between the XR‐NTX or placebo groups Note: authors also conducted as treated analysis (high treatment ≥3, low treatment <3 injections) –Significantly lower mean proportion of opioid use for high than low treatment group by month one and through month five post‐release –No change in the percentage of days used were similar in both groups between months 6 and 11	Not reported	Not reported
	Combined results from INSPIRE and NEW HOPE	Biondi (2019) [52]	Sexual risk behaviours among people with HIV and OUD and/or AUDs receiving XR‐NTX vs. placebo Alcohol and/or opioids RCT‐ double‐blind, placebo‐controlled ‐ subsequent analysis	No comparison group data	Not reported	Not reported	No between group comparison	Not reported
		Biondi (2021) [55]	To assess differences in baseline characteristics by gender of people with HIV being treated for AUD and/or OUD enrolled in an RCT of XR‐NTX. Alcohol and/or opioids Observational study–secondary analysis	Not reported	Not reported	Not reported	Not reported	Not reported
		Springer (2015) [51]	Acceptability and retention of XR‐NTX among people living with HIV disease with AUD and/or OUD Alcohol and/or opioids Observational study–secondary analysis	No between group comparison	Not reported	Not reported	Not reported	Not reported
		Vagenas (2014) [45]	Hepatotoxicity among HIV‐infected prisoners prescribed ART and XR‐NTX Alcohol and/or opioids Observational study–secondary analysis	Not reported	Not reported	Not reported	Not reported	No change (XR‐NTX vs. placebo) –In elevations of Liver Function Tests at baseline, 3 or 6 months post first injection
XR‐NTX	Woody (2021) [59]	Woody (2021) [59]	Relapse rates following XR‐NTX pre‐release vs. post‐release Opioids RCT	Improved (pre‐ vs. post‐release group) –More weeks in treatment	16 overdoses among 9 study patients –3 in the pre‐release group–6 in post‐release group –p = 0.72 a NB: all were either unmedicated or had missed injections	No change (pre‐ vs. post‐release administration) –In percentage with relapse over the 12‐week study period	Increased (for XR‐NTX pre‐ vs. post‐release) –Reincarceration	Not reported
		Jalali (2022) [61]	An economic evaluation of XR‐NTX pre‐release vs. post‐release Opioids Economic evaluation‐ secondary analysis	Not reported	Not reported	Not reported	Not reported	Not reported
XR‐BUP	Dunlop (2022) [56]	Dunlop (2022) [56]	Safety and tolerability of XR‐BUP (with comparator group of methadone) Opioids Cohort study	No between group comparison	No change across groups –0 in both arms	Reduced –In prevalence of non‐prescribed opioid use from baseline a to 4 and 16 weeks in XR‐BUP group –In prevalence of injecting drug use from baseline a to 4 and 16 weeks in XR‐BUP group Mixed results in –Mean days of opioid use from baseline a to 4 (reduced) and 16 (no change) weeks in XR‐BUP group –In mean days of any injecting drug use from baseline a to 4 (reduced) and 16 (no change) weeks in XR‐BUP group	Not reported	No between group comparison
		Ling (2022) [66]	What are the costs of administering XR‐BUP, methadone and SL‐BUP? Opioids Economic evaluation	Not reported	Not reported	Not reported	Not reported	Not reported
	Lee (2021) [57]	Lee (2021) [57]	To pilot the use of XR‐BUP and examine acceptability and feasibility vs. SL‐BUP Opioids RCT–open‐label	Mixed effects overall (for XR‐BUP vs. SL‐BUP) –Improved retention on any form of community buprenorphine treatment at week 8 –No change in retention on the assigned buprenorphine formulation treatment at week 8	0 in each arm	No change (XR‐BUP vs. SL‐BUP) –In difference between opioid‐free urine samples	No change (XR‐BUP vs. SL‐BUP) –Reincarceration in NY jails	No between group comparison
		Cheng (2022) [60]	What is the treatment experience, treatment satisfaction for XR‐BUP, and barriers and facilitators to XR‐BUP initiation in jail and retention post‐release? Opioids Qualitative review	Not reported	Not reported	Not reported	Not reported	Not reported
	Martin (2022) [63]	Martin (2022) [63]	What are the real‐world conditions of the use of XR‐BUP? Opioids Case series–retrospective	No comparator group	Not reported	Not reported	No comparator group	No comparator group
	Soyka (2021) [58]	Soyka (2021) [58]	What is the experience of patient through initiation and stabilisation of a depot buprenorphine in patients who had been on methadone in the CJS? Opioids Case series (retrospective chart review)	No comparator group –4/7 retained on treatment for 14 months	Not reported	Not reported	Not reported	No comparator group
	Wright (2020) [54]	Wright (2020) [54]	What are the costs of providing XR‐BUP in prisons Opioids Economic evaluation	Not reported	Not reported	Not reported	Not reported	Not reported
	Wong (2022) [65]	Wong (2022) [65]	How do the costs of medication preparation, administration, monitoring, and personnel for XR‐BUP vs. SL‐BUP compare? Opioids Economic evaluation	Not reported	Not reported	Not reported	Not reported	Not reported
Both	Will (2022) [64]	Will (2022) [64]	What is the effect of providing medications for OUD (inc. XR‐BUP) on ED visits within 28 days of release from jail? Opioids Cohort study–retrospective	Not reported	Not reported	Not reported	Not reported	Not reported

Abbreviations: AUDs, alcohol use disorder; ITT, intention to treat; OAT, opioid agonist therapy; OUD, opioid use disorder; TAU, treatment as‐usual.

^a Above results not compared to comparator group (stable methadone), however inclusion criteria for XR‐NTX group (baseline status) was those not on OAT on entry to custody.

Most studies included men and women (19/25) [37, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 56, 57, 60, 61, 62, 63, 64], with men typically exceeding 75%. Two studies included exclusively men [36, 58], whereas four did not describe gender [54, 59, 65, 66]. Only one study reported on other genders, including transgender [48, 52]. Participants were identified as predominantly: African American in two studies [45, 46, 47, 48, 52, 55], White in four studies [50, 53, 63, 64] and Hispanic in two studies [51, 57, 62]. Race was not reported in 11 articles [36, 37, 44, 49, 54, 58, 59, 60, 65, 66]. One Australian study mentioned Aboriginality, which accounted for 38% of the total sample [56].

Critical appraisal

Using the RoB‐2, the risk of bias was high in eight articles [36, 37, 44, 48, 49, 51, 57, 59] and concerning in two [52, 53] (Table S1). With the ROBINS‐I, one article had a critical risk of bias [46], four serious [45, 50, 56, 62] and one moderate [47] (Table S2). We did not conduct a ROBINS‐I on one included article because it examined baseline differences by gender on two existing RCTs, which was not an outcome of interest in this review [55]. No articles were judged to have a low risk of bias.

We used the article by Evers et al. [67] to conduct risk of bias assessments on the economic analyses [54, 61, 65, 66]. Results varied considerably, with some studies receiving a low risk of bias in only two of 11 areas, and others in eight of 11 (Table S3). The relevant JBI tools [68] were used for the qualitative study [60] (Table S4), which met 70% of the JBI criteria, the two case‐series [58, 63] (Table S5), which met 70%, and the retrospective cohort study [64] (Table S6), which met 45%. We used the Grading of Recommendations Assessment, Development and Evaluation approach to assess the level of certainty of presented evidence.

XR‐NTX for alcohol outcomes

SU

The one study (RCT by Springer et al. [48]) examining alcohol specifically, found a reduction in composite alcohol scores for those participants who received four or more injections of XR‐NTX versus placebo (Table S7). Otherwise there were no changes in alcohol scores for the XR‐NTX group versus placebo [48]. The four articles examining the effects of XR‐NTX on alcohol and/or opioids did not report SU outcomes [45, 47, 52, 55].

RIT

Springer et al. [48] also presented results for RIT, reporting a statistically significant higher number of total injections administered in the XR‐NTX group (202/402, 50.2%) compared to placebo (88/198, 44.4%) (Table S8). However, there was no statistically significant difference in the percentage of participants presenting for all six injections [XR‐NTX 10/67 (14.9%); placebo 6/33 (18.2%); P = 0.061]. Two articles [47, 52] (one study) reported on RIT for a group with alcohol and/or opioid use [47, 52] (Tables S9–S10). Neither article compared treatment arms.

Springer et al. [48] found no difference between XR‐NTX and placebo in participants who were not reincarcerated [52] (Table S11).

XR‐NTX for opioids outcomes

SU

The impact of XR‐NTX on opioid use varied across seven studies. Three studies indicated a reduction [36, 46, 49] and four studies showed no change [37, 51, 53, 59, 62] (Tables S12–S13). This variation in effect was associated with heterogeneity of the methods and comparator groups. For example, the studies showing reduction in use examined XR‐NTX versus placebo, XR‐NTX completers versus non‐completers and pre‐ versus post‐release XR‐NTX, respectively, whereas the studies showing no change compared XR‐NTX to methadone, placebo, treatment as usual (TAU; no medication), pre‐ versus post‐release and three or more versus two or less injections received.

RIT

Definitions, measurements and comparator groups varied widely across studies looking at RIT with XR‐NTX. Seven studies reported on RIT [36, 37, 44, 46, 49, 50, 59, 62], of which four RCTs [36, 37, 44, 49, 62] and two cohort studies [46, 50] compared treatment arms (Tables S14–S15). One RCT showed longer retention with XR‐NTX implants than methadone, but defined retention as not removing the XR‐NTX implant within 6 months, rather than active treatment participation [37, 44]. Two RCTs revealed no change in retention between XR‐NTX and placebo [62] or TAU (no medication) [36], but used attendance at the study interview (during which the relevant medication was administered) as the retention measure. Of the studies examining XR‐NTX administration pre‐ versus post‐release, two showed mixed effects (with results showing both improved retention and no change for the pre‐release group at various study points) [46, 50] and one showed improved retention in the pre‐release group [59].

Overdose

Four XR‐NTX studies reported on overdoses [36, 46, 50, 59] with rates of 0% (0/16 [36] and 27 [46] participants), 4.5% (3/67 participants) [50] and 10.5% (16 incidents of overdose among 9/86 participants) [59] (Table S16).

Recidivism

Eight of 17 XR‐NTX articles (7 studies) reported on recidivism [36, 37, 44, 46, 50, 52, 53, 59]. Four studies found no significant differences between treatment arms, comparing XR‐NTX against methadone (n = 2) [37, 44] or treatment as usual [53], treatment completers versus non‐completers [46] or pre‐release versus post‐release administration [50] (Table S11). One RCT showed a significant reduction in recidivism if XR‐NTX was received pre‐ versus post‐release [59], whereas another described lower recidivism when comparing partial and full treatment completion among the XR‐NTX arm [36]. One article had insufficient data to assess treatment effect (e.g. no between‐group comparison) [52].

Economic analysis

The one economic analysis of XR‐NTX in opioid use examined outcomes pre‐ versus post‐release [61] (Table S17). Incremental cost‐effectiveness ratios were estimated to be >$500 000 per quality adjusted life‐years for all stakeholder perspectives.

XR‐NTX AEs

Of the 16 articles, eight (7 studies) reported AEs for XR‐NTX (1 article for alcohol, 6 for opioid, 1 for alcohol and/or opioids). Reporting standards and definitions varied widely across studies (Table S18). Eight articles (7 RCTs) reported on AEs for XR‐NTX [49, 53], but only two studies compared XR‐NTX with another arm (placebo) [45, 48, 51]. One study found no statistically significant differences in any individual AEs between the two groups, although the criteria used to define AEs was not specified [51]. The other articles presented results for liver transaminases, but again found no significant difference [45, 48]. Two other XR‐NTX RCTs listed drug‐related AEs like fatigue, lump at injection site, nausea/vomiting etc. [37, 49].

XR‐BUP outcomes

SU

For XR‐BUP, one cohort study showed a reduction in prevalence of opioid and injecting drug use, but compared to baseline rather than to the comparator group of participants stable on methadone [56] (Table S19). The second study, an RCT, found no significant change in SU compared to sublingual buprenorphine (SL‐BUP) [57] (Table S20).

RIT

For XR‐BUP, retention was reported in four of eight studies [56, 57, 58, 63], and overall, there was evidence of reasonable retention (Tables S21–S22). Two of four studies had comparator groups—the first, a cohort study, received methadone [56], whereas the second, an RCT, SL‐BUP [57]. The first reported a 92.3% treatment retention probability at 16 weeks, although this was not compared to the methadone group [56]. The second had mixed findings, revealing no change in participants retained on the assigned formulation at week 8, but improved retention on any form of buprenorphine in the XR‐BUP compared SL‐BUP group [57].

Overdose

Two XR‐BUP studies reported on overdoses (0 among 129 [56] and 52 [57] participants) (Table S16).

AEs

Four studies reported XR‐BUP AEs, although, as with XR‐NTX, reporting standards and definitions varied (Table S18). Of the three XR‐BUP RCTs, one [57] reported eight AEs among 52 participants, primarily injection site tenderness. That study reported no differences in serious AEs between groups. Another XR‐BUP study reported at least one treatment emergent AE in 97% of participants (65/67) with 94% (63/67) deemed related to study drug [56] and most (88%) deemed mild and resolved in a day. The remaining 12% were deemed moderate.

Recidivism

Two of eight XR‐BUP studies reported recidivism outcomes. The RCT found no difference compared to SL‐BUP [57], whereas the case series lacked a comparator group [63] (Table S11).

Economic analysis

There were three economic analyses on XR‐BUP [54, 65, 66] (Table S17). Despite identifying methodological constraints in the XR‐BUP studies, such as all studies reporting total costs rather than cost effectiveness, there was preliminary evidence that initial medication costs are offset by savings in staffing, security and healthcare [65]. One study also demonstrated that as XR‐BUP became increasingly used, the per‐patient costs for methadone and SL‐BUP increased and total costs were most sensitive to treatment administration time.

Qualitative results

Cheng et al. [60] explored the treatment acceptability, barriers and facilitators of XR‐BUP, with some qualitative results also reported in a preliminary (primarily quantitative) article from the same study [57]. Reported benefits of XR‐BUP included no daily attendance for dosing [57], less urgency for follow‐up care immediately post release [57], increased access to employment [60], reduced interactions with corrections officers [60], more anonymity than with daily dosed OAT [60], a perceived sense of ‘normalcy’ (with some attributing that to a lack of cravings and withdrawal symptoms) and avoiding overdose episodes while incarcerated or post release [60]. Perceived barriers to XR‐BUP included apprehension about the novel formulation [57], opposition to needles [57], difficulties accessing XR‐BUP in the community [57], loss of a ‘boost’ feeling from daily sublingual dosing [60] and injection site pain [60].

DISCUSSION

This review summarises the evidence base for the use of XR relapse prevention pharmacotherapy in CJS, which is important given the rapid increase in its use. There was an upward trend in articles published post‐2020, yet high heterogeneity in the studies precluded pooling of results (see Tables S7–S14, Tables S19–S22 for justification of this decision). Moreover, the overall certainty of the evidence in this review was predominantly low to very low (Table S23). However, our results demonstrated increased effort to conduct high‐quality, ethical research in CJS. All XR‐BUP articles were published post‐2020, reflecting its recent regulatory approval (e.g. 2018 in Australia) and suitability for social distancing during the coronavirus disease 2019. Opioids were the primary drug of concern in most studies, indicating limited information on XR medications for AUD.

The CJS can be a challenging area to conduct research, which may explain why seven of the 14 XR‐NTX articles were subsequent analyses or secondary outcomes from the same US study. This also meant there was a strong United States representation among XR‐NTX studies, although this representation could be because of lower OAT prescribing and/or its large CJS population relative to other developed countries. There are several factors contributing to hesitancy in OAT prescribing, including stigma, concerns about diversion, availability of prescribers and regulatory barriers [69, 70]. Many barriers are less pronounced for naltrexone compared to OAT. Additionally, XR‐NTX has been used extensively in the US drug courts, with some suggesting that it was marketed directly to drug court judges [71]. These unique local issues mean that results from one area may not be as compelling in a different environment. More studies from various jurisdictions are needed to enhance generalisability of XR‐NTX findings.

SU

We found no clear evidence that XR medications led to less SU compared with shorter acting formulations. This is consistent with general population studies where RCTs have indicated that XR‐NTX is not inferior to SL‐BUP in terms of negative urine samples and days of illicit opioid use [72], and weeks with confirmed abstinence and opioid‐free days were improved compared to placebo [73]. However, another study suggested that XR‐NTX was associated with quicker relapses and fewer opioid‐free days than SL‐BUP [74]. By comparison, oral naltrexone has clearly limited evidence for preventing opioid relapse, inferior retention to OAT and is linked to an increased overdose vulnerability when discontinued [25, 27]. Our review showed some evidence supporting sustained treatment with XR‐NTX (≥4 injections) for AUD in the CJS [48]. General population data supports this, showing significant reductions in monthly drinking and heavy drinking days [75]. Nonetheless, given limited prison‐based data, a greater understanding of the feasibility and acceptability of XR‐NTX is warranted.

Only two XR‐BUP trials reported on SU outcomes in the CJS (one RCT showing no significant change vs. SL‐BUP and the other cohort study not comparing study groups), which hindered conclusions. The qualitative findings to date indicate that some XR‐BUP participants experienced a newfound ‘normalcy’ marked by reduced withdrawal symptoms and cravings, whereas others reported feeling less supported [60]. Reasons are not entirely clear, but this may be because of classical and operant conditioning associated with daily dosing routines [76] or changes in receptor occupancy following each new dose. Given the anticipated consistency in pharmacology between XR‐BUP and SL‐BUP, and evidence for some improved outcomes from XR‐BUP in the general population [77], it may be more pertinent to prioritize studies on feasibility, cost efficacy and patient experiences in CJS.

RIT

There was not convincing evidence that XR‐NTX improved retention for AUD or OUD compared to placebo. The single study looking at RIT in AUD found greater partial adherence to XR‐NTX versus placebo (via total number of injections). However, there may be a component of dilution of effect because of the higher attrition in the early months among the placebo group (24/33 [72.7%] received 1 injection compared to 61/67 [91%] of the XR‐NTX arm). Moreover, there was no difference in full retention (with attendance for all 6 injections), and the raw percentages for receiving six injections were quite low, at less than 20%. Further studies would be beneficial to establish if there is a clinically meaningful difference, particularly with oral naltrexone as the comparator. Identification of only one study reporting on RIT for AUD is concerning given that AUDs are likely prevalent in many CJS, with a 2018 survey showing 34% of Australian prison entrants had high‐risk alcohol consumption [78].

For the included opioid studies, the interpretation of RIT was complicated by small to moderate sample sizes, varying definitions, measurements and comparator arms. For instance, some studies used less optimal definitions (e.g. in situ implant at 6 months) and assumed ongoing treatment for those lost to follow‐up [44]. A stronger measure would be active treatment engagement, like returning for subsequent implant insertion. This is important given the heightened risk of overdose associated with ceasing naltrexone stemming from reduced opioid tolerance—a risk that ongoing service engagement can somewhat mitigate. Retention may also be shaped by patient preference and community support. In one trial, 35% (8/23) of XR‐NTX participants dropped out, citing preference for OAT, whereas 29% (6/21) of methadone participants could not begin because of post‐release prescribing barriers (Table S24).

Similarly, retention was reported in half of XR‐BUP articles, yet the lack of head‐to‐head comparison between arms limited firm conclusions. Some effect was suggested, however, for high and/or improved treatment retention [56] for XR‐BUP, which is encouraging. In the general population, the percentage of patients retained on XR‐BUP is higher than for SL‐BUP [73]. However, factors influencing retention are complex and can include inadequate doses, concurrent SU, legal issues and travel distance to treatment facilities [79]. Some of these features were reported on, but others (e.g. dose) were not. Comparison of XR and short‐acting formulations, such as SL‐BUP or methadone, in the CJS would increase our understanding of the factors influencing how best to retain individuals in treatment.

The included qualitative study highlighted both benefits and challenges of XR‐BUP, offering a deeper understanding of retention, including protection from withdrawal during CJS re‐entry, reduced cravings, improved job prospects and higher treatment satisfaction [60]. XR‐BUP lacks the behavioural and ritualistic components of daily dosing, affecting perceived treatment support. Needle phobia and pressure to switch to XR formulations also pose challenges. Similar difficulties have been echoed in qualitative studies conducted in the general population [80]. Exploring patient experiences in relationship to these factors is crucial for understanding the XR treatments and shaping their future direction.

Overdose

There were relatively low numbers of participants in studies with overdose as a reported outcome and no between‐group comparisons. Larger sample sizes could help determine any medication effect on morbidity and mortality.

AEs

Assessing AEs is important for treatment acceptability. However, pharmacologically, both XR‐medications are unlikely to exhibit substantial differences from their daily counterparts given their similar chemical structure and function. No studies were stopped because of serious AEs, and Vivitrol (XR‐NTX) and Sublocade and Brixadi (both XR‐BUP) have US Food and Drug Administration approval. Unfortunately, despite most studies reporting on AE, the measures and definitions varied widely, making it complex to arrive at definitive conclusions. If future research explores AEs of XR formulations, adopting a more rigorous clinical trial approach, with more appropriate comparators, would be beneficial.

Recidivism

Our review found no change in recidivism rates with XR medications. Further research is warranted, however, given the limited studies, variable follow‐up durations and the often low quality of studies. Recognizing the multitude of recidivism‐influencing factors, a wider lens is required. This could involve investigating how XR medications impact broader aspects of an individual's life, including housing and employment situations. In addition, exploring the nature of charges leading to reincarceration could provide valuable insights.

Economic analyses

Certain jurisdictions, including in Australia [81], have been hesitant to adopt XR‐BUP and XR‐NTX because of financial constraints [81]. Given only four studies included an economic analysis, definitive conclusions about the cost efficacy of XR‐NTX in the CJS are not possible, although there was some indication that costs may be offset by savings in staffing, security and healthcare [65]. If this finding is confirmed in subsequent studies, it would allay concerns that the use of XR formulations in CJSs is prohibitively expensive. This potential for cost offsetting is an important finding. Cost may be a major barrier to widespread XR formulation adoption in the CJS and general population.

Limitations

First, comparing two different pharmacotherapy formulations (agonist and antagonist) proved to be challenging and results were presented separately. Second, the studies exhibited substantial heterogeneity across designs, definitions, measures, participant groups and timelines. This diversity restricted our ability to pool data and draw definitive conclusions within or across groups. Third, some studies did not report on all variables of interest, limiting the depth of analysis. Even when relevant variables were reported, the robustness of the findings was often undermined by a high‐risk of bias within studies. Fourth, we opted to exclude grey literature because of practical constraints such as time. Last, most data originated from the United States, which may restrict the applicability of our findings to other jurisdictions because of the unique aspects of the American CJS and its treatment protocols. We note that the available research was conducted before reported rises in fentanyl or nitazene use across many jurisdictions. Notably, the XR treatments in these studies seem to have been commenced in the absence of any acute withdrawal syndrome. However, these changes in drug supply may result in requirements for higher doses of OAT and a more complex withdrawal management process [82].

Recommendations

A lack of evidence is not the same as evidence of non‐efficacy. Our results demonstrate the value of ongoing research, particularly into XR‐BUP given its widespread use and the emerging evidence for effectiveness identified in this review. Future research should focus on increasing rigour while prioritising participants' experiences. This includes consistent definitions, measures and follow‐up periods. Where feasible, preference should be given to comparator groups such as existing short‐acting formulations (naltrexone, methadone and SL‐BUP) or TAU if such medication was unavailable. Prioritising RCTs or—given the challenges of RCTs in CJS—large, well‐conducted observational studies such as those using administrative datasets, would improve the ability to pool results and the confidence in effect estimates. This is especially important for outcomes like opioid overdose, which require large cohorts to be adequately powered. In‐depth qualitative studies are needed to provide meaning to quantitative results. There are ethical concerns around use of placebos among opioid using populations, particularly in CJSs where there are high rates of overdose as well as evidence for the effectiveness of short‐acting formulations. Future research investigating the effectiveness and feasibility of XR‐NTX for AUDs would also be beneficial, especially given AUDs are likely more prevalent than OUD in many CJSs. For instance, in 2018, 34% of prison entrants had high‐risk alcohol consumption, compared to only 7% who had recently used heroin [78].

CONCLUSIONS

This review demonstrated growing evidence for the effectiveness of XR‐BUP in reducing SU and retaining people in treatment, including that initial medication costs may be offset. Although adequate evidence for the use of XR formulations in CJS is yet to be established, our findings are important to the wellbeing and potential for recovery of people in the CJS who experience high rates of stigma and SU relapse. Given the promising trend identified, and the presence of heterogeneity in methods and outcomes, our review demonstrates value in conducting robust studies of extended‐release formulations in prisons. Although there was no clear direction of effect for XR‐NTX, studies of its use for AUD in particular may be indicated because of the current low number of studies despite AUD's high prevalence and associated harms in a CJS population.

AUTHOR CONTRIBUTIONS

Amelia Woods: Conceptualization; data curation; investigation; methodology; project administration; validation; writing—original draft. Catherine Foley: Supervision. Katherine M. Conigrave: Conceptualization; data curation; investigation; methodology; supervision. Winifred Asare‐Doku: Investigation. Anthony Shakeshaft: Conceptualization; methodology; supervision. Stella Settumba‐Stolk: Investigation. Michael Farrell: Conceptualization; methodology; supervision. Michael Doyle: Conceptualization; methodology; supervision.

DECLARATIONS OF INTEREST

There are no contractual constraints on publication. M.D. was supported by a National Health and Medical Research Council investigator grant. GNT1193618. M.F. and C.F. are supported by National Drug and Alcohol Research Centre, which is funded by the Australian Government. M.F. has received untied educational grants from Indivior. C.F. was supported by funding provided by the New South Wales (NSW) Ministry of Health under the NSW Health Prevention Research Support Program.

PROSPERO REGISTRATION

CRD42021276447.

ETHICS

This review involved secondary analysis of published work; ethical approval was not required.

Supporting information

Data S1. Search Strategy

TABLE S1. RoB Quality assessment of included RCTs (n = 10 papers; 8 studies)

Table S2. ROBINS‐I Quality assessment of included studies (n = 7 papers; 5 studies)

Table S3. Quality Assessment of Studies Conducting Economic Analyses (n = 4 studies)

Table S4. Quality assessment of Included Qualitative Studies (n = 1 study)

Table S5. Quality Assessment of Case Series studies (Joanna Briggs Institute) (n = 2 papers)

Table S6. Quality Assessment of Retrospective Cohort studies (Joanna Briggs Institute) (n = 1 study).

NB: Risk of Bias assessments have been completed for prospective cohort studies using the ROBINS‐I tool. While the ROBINS‐I tool can be used for retrospective cohort designs, a number of the questions are more appropriate for prospective research designs, and so the decision was made to use the JBI tool in the case of retrospective cohort studies.

Table S7. Substance use results from RCTs of XR‐NTX for alcohol (n = 1 study)

Table S8. Retention in treatment results for XR‐NTX RCTs for alcohol use (n = 1 study)

Table S9. Retention in treatment results for XR‐NTX RCTs for both alcohol and opioid use (n = 1 study)

Table S10. Retention in treatment results from non‐RCTs of XR‐NTX for both opioids and/or alcohol (n = 1 study)

Table S11. Recidivism results (n = 9 papers, 8 studies)

Table S12. Substance use results from non‐RCT trials of XR‐NTX for opioids (n = 2 studies)

Table S13. Substance use results from RCTs of XR‐NTX for opioids (n = 8 papers, 7 studies)

Table S14. Retention in treatment results from XR‐NTX RCTs for opioids (n = 6 papers, 5 studies)

Table S15. Retention in treatment from non‐RCT trials for XR‐NTX for opioids (n = 2 studies)

Table S16. Opioid overdose results (n = 6 studies)

Table S17. Economic Analyses (all opioids) (n = 4)

Table S18. Adverse events results (n = 12)

Table S19. Substance use results from non‐RCTs of XR‐BUP for opioids (n = 1)

Table S20. Substance use results from RCTs for XR‐BUP for opioids (n = 1)

Table S21. Retention in treatment results from non‐RCT studies of XR‐BUP for opioids (n = 3)

Table S22. Retention in treatment results from RCTs of XR‐BUP for opioids (n = 1)

Table S23. Additional findings (n = 10)

Table S24. GRADE assessment of overall certainty of evidence

Woods A, Foley C, Conigrave KM, Asare‐Doku W, Shakeshaft A, Settumba‐Stolk S, et al. Extended‐release pharmacotherapies for substance use disorders in incarcerated populations: A systematic review. Addiction. 2025;120(5):835–859. 10.1111/add.16766

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.