A systematic review of post-release programs for women exiting prison with substance-use disorders: assessing current programs and weighing the evidence.

BACKGROUND
The rising rates of women in prison is a serious public health issue. Unlike men, women in prison are characterised by significant histories of trauma, poor mental health, and high rates of substance use disorders (SUDs). Recidivism rates of women have also increased exponentially in the last decade, with substance related offences being the most imprisoned offence worldwide. There is a lack of evidence of the effectiveness of post-release programs for women. The aim of this systematic review is to synthesise and evaluate the evidence on post-release programs for women exiting prison with SUDs.


METHODS
We searched eight scientific databases for empirical original research published in English with no date limitation. Studies with an objective to reduce recidivism for adult women (⩾18 years) with a SUD were included. Study quality was assessed using the revised Cochrane Risk of Bias tool for randomized trials (RoB2) and the Risk of Bias in Non-randomized Studies - of Interventions (ROBINS-I) tools.


RESULTS
Of the 1493 articles, twelve (n = 3799 women) met the inclusion criteria. Recidivism was significantly reduced in five (42%) programs and substance-use was significantly reduced in one (8.3%) program. Common attributes among programs that reduced recidivism were: transitional, gender-responsive programs; provision of individualised support; providing substance-related therapy, mental health and trauma treatment services. Methodological and reporting biases were common, which impacted our ability to synthesize results further. Recidivism was inconsistently measured across studies further impacting the ability to compare results across studies.


CONCLUSIONS
Recidivism is a problematic measure of program efficacy because it is inconsistently measured and deficit-focused, unrecognising of women's gains in the post-release period despite lack of tailored programs and significant health and social disadvantages. The current evidence suggests that women benefit from continuity of care from prison to the community, which incorporated gender-responsive programming and individualised case management that targeted co-morbid mental health and SUDs. Future program design should incorporate these attributes of successful programs identified in this review to better address the unique challenges that women with SUDs face when they transition back into the community.


Introduction
Incarcerated women are one of the most vulnerable groups in society who, upon entry into prison exhibit a range of complex and inter-related health and social issues (Dumont, Brockmann, Dickman, Alexander, & Rich, 2012;J. E. Johnson & Zlotnick, 2008;Kinner & Young, 2018;Pelissier, Motivans, & Rounds-Bryant, 2005). Although the proportion of incarcerated women globally is much lower than the proportion of incarcerated men (6.9% compared to 93.1%, respectively) (Walmsley, 2017), the number of women imprisoned since 2000 continues to increase globally at a rate that is double the rate for the imprisonment of men (Australian Bureau of Statistics, 2018b; J. E. Johnson & Zlotnick, 2008;B. E. Salem et al., 2013;Walmsley, 2017). There are considerable variations between countries, for example the latest Australian figures show that around 8% (n = 3587) of the prison population is women (Australian Bureau of Statistics, 2018a) and in the United Kingdom this figure was 5% (n = 7745) (Women in Prison, 2017). The United States has the highest total number of women in prison (n = 211,870, representing 8.7%) in any one country, as well as the highest prison population rate for women (about 65.7 per 100,000 of the national population) (World Prison Brief, 2018). Comparatively, African countries have a much lower total prison population proportion at 3.4% (or 3.2 per 100,000 of the national population) (Walmsley, 2017).
Along with SUDs, women in prison are characterised by extensive histories of trauma and poor mental health (MH) (Covington, 2001;J. E. Johnson & Zlotnick, 2008; B. E. Salem et al., 2013;Schonbrun, Johnson, Anderson, Caviness, & Stein, 2017;Wetton & Sprackett, 2007;World Health Organisation, 2009). The prevalence of emotional, physical, and sexual abuse is reported between 77% and 90% of women in prison respectively (Australian Institute of Family Studies, 2012; . A recent review summarised the literature on sexual abuse and mental illness prevalence among samples of incarcerated women (Karlsson & Zielinski, 2018). Best estimates for sexual abuse were: 50-66% for child sexual abuse, 28-68% for adult sexual abuse, and 56-82% for a lifetime of sexual assault (Karlsson & Zielinski, 2018). The review highlighted that incarcerated women have significantly greater exposure to sexual victimization compared to national standards, incarcerated men and women in community (Karlsson & Zielinski, 2018).
Experiences of trauma predispose women for adverse MH conditions such as post-traumatic stress disorder, depression, anxiety and suicide (Karlsson & Zielinski, 2018;World Health Organisation, 2009). Women who experienced trauma as a child have a 40% increase in odds of developing a MH condition in adulthood . A meta-analysis of the effect of adverse childhood experiences on health describes the findings of 37 studies and presents the pooled risk of various health conditions (Hughes et al., 2017). The risk of adverse MH conditions, such as anxiety, depression, and schizophrenia, was found to be about four times higher, as compared to people who experienced less than four adverse childhood experiences (anxiety OR 3.70; depression OR 4.40,schizophrenia OR 3.60). In addition, people with four or more adverse childhood experiences were at higher risk of SUDs with problematic alcohol use nearly six times higher (OR 5.84) and problematic drug use over ten times as high (OR 10.22) (Hughes et al., 2017). Substance dependency among women in prison is significantly higher among women who have experienced childhood abuse and MH problems (H. Johnson, 2006). Women are more likely than men to start using substances as a means to alleviate the pain of trauma and to manage existing MH conditions (Langan & Pelissier, 2001;Stalans, 2009). Trauma, MH and substance-use are therefore inter-related factors that can result in cumulative and compounding MH issues, addiction, and contact with the criminal justice system (see Fig. 1) (Alleyne, 2008;Australian Bureau of Statistics, 2017b;Covington, 2001;Karlsson & Zielinski, 2018;B. E. Salem et al., 2013).

Exiting prison
Compared to men, women generally serve short sentences which is a reflection of the minor, non-violent crimes they have been sentenced for (Baldry, 2010;Balyakina et al., 2014;van den Bergh, Gatherer, & Møller, 2009). When women are released into the community they face many disadvantages including poor continuity of care, inadequate social support, parenting stress, homelessness and poverty, and reduced employment opportunities (Baldry, 2010;Begun, Early, & Hodge, 2016;B. E. Salem et al., 2013). A notable difference between men and women in prison is that half of Fig. 1 Pathways to women's criminal offending incarcerated women (54%) are mothers to dependent children (age < 16) and were the primary carer of one or more children before incarceration (compared to only 36% for men) (Australian Institute of Health and Welfare, 2019; Kilroy, 2016). Maternal stress, coupled with the many disadvantages cited, are often barriers to accessing immediate and affordable healthcare and drug and alcohol treatment services. As a result, women with SUDs who are recently released from prison are at a high risk of experiencing an adverse MH episode, illness and death compared to the general population. The risk of death is especially high in the first month after release, and the causes of death are usually preventable, including suicide, injury, and overdose (Sullivan et al., 2019).
Post-release (also known as re-entry, reintegration, and resettlement) programs are interventions that are delivered in the community. Transitional programs are interventions that start pre-release (in custody) and support people during the transition from prison to community (Baldry, 2010;Borzycki, 2005). Post-release and transitional programs are often evaluated based on a measurement of recidivism. Recidivism is used to measure the proportion of people who go on to reoffend during a pre-defined post-release period (Bartels & Gaffney, 2011;Sullivan et al., 2019;Urban Institute, n.d.;Yukhnenko, Sridhar, & Fazel, 2019). A systematic review of recidivism rates, two years post-release for both men and women across 11 countries found re-arrest rates were between 26% and 60% and reconviction rates ranged from 20% to 63% (Yukhnenko et al., 2019). These recidivism rates suggest that many people with a history of incarceration either do not access, or do not benefit from services and programs during their time in prison, or do not have adequate support or change in social circumstances in the community to prevent reoffendingarrest (Baldry, McDonnell, Maplestone, & Peeters, 2006).
Despite the growth of the women's prison population, and their profoundly different criminogenic profile compared to incarcerated men, the majority of prison programs available have been designed for men and extended to women with little alteration (Armstrong, Chartrand, & Baldry, 2005;Bartels & Gaffney, 2011;Langan & Pelissier, 2001;Lawlor, Nicholls, & Sanfilippo, 2008;Suter, Byrne, Byrne, Howells, & Day, 2002). Emerging evidence indicates that community based programs that are gender-responsive and address criminogenic needs can improve the transition process and minimise recidivism rates post-release (Begun et al., 2016;Borzycki, 2005;Borzycki & Baldry, 2003;Carlton & Segrave, 2016). Gender-responsiveness (or genderinformed) refers to programming that explicitly considers the needs that are particularly salient to women. Gender-responsive approaches are trauma-informed and consider the gendered context (or "pathways") of criminal offending (Covington & Bloom, 2006;Gobeil, Blanchette, & Stewart, 2016). A meta-analytic review of correctional interventions for women in prison examined whether programs, either gender-informed or gender-neutral, were effective in reducing recidivism (Gobeil et al., 2016). The results demonstrated that participation was associated with 22% to 35% greater odds of community success and gender-responsive interventions were significantly more likely to be associated with reductions in recidivism (Gobeil et al., 2016).
Given the proportion of women in prison with SUDs and correlation to reoffending and risk of death postrelease, more research is needed to understand the effectiveness of programs for this population. To-date, there has been no systematic review of the evidence about what is available and "what works" in regard to post-release programs for women with SUDs. The aim of this research is to critically review the available evidence of the effectiveness of community based (post-release and transitional) programs offered to women with SUDs to inform program development to decrease reoffending. Further, as the link between criminal offending and substance-use is well established, we also aim to review the effectiveness of interventions to reduce substance-use outcomes post-release and whether this impacts recidivism. This review addresses the following research questions (RQ): RQ1: Are post-release and/or transitional programs effective in reducing recidivism and/or substance-use for women with SUDs post-release? RQ2: Do those that report a reduction in substance-use also report a reduction in recidivism? RQ3: What program characteristics are common among programs which report improved recidivism and substance-use outcomes post-release?

Methods
This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher, Liberati, Tetzlaff, Altman, & the PRISMA Group, 2009). The systematic review was registered in the PROSPERO database CRD42020162036. The databases PubMed and CINAHL (including MESH terms), Cochrane, EMBASE (including EMTREE terms), Scopus, PsycInfo, ProQuest and SOCIndex were originally searched in September 2019, with no date limitation. The search strategy was split into six core concepts using a combination of words related to "Post-release", "Prison", "Women", and "Interventions". The electronic database searches were supplemented with manual searches of the reference lists from relative articles. Due to the limited number of publications found an updated search was conducted in February 2020 following the method by Bramer and Bain (2017), adding search terms related to "Substance use" and "Recidivism" (see Additional File 1).

Eligibility criteria
Studies included were primary reports of effectiveness trials (i.e., studies of an intervention with a comparator) with an objective to reduce recidivism for adult women (⩾18 years) with a known SUD. The program had to be either a post-release or transitional intervention, published in English in a peer-reviewed journal. In this review substance-use included individuals using occasional drugs or alcohol, those who were dependent, or those who had other drug and alcohol related problems prior to their current offence. Studies that included both men and women were included if the results relating to women could be isolated. Due to the limited published studies of women in prison (Baldry, 2010;Borzycki & Baldry, 2003;Segrave & Carlton, 2011) there were no limitations by study design or intervention type to ensure identification of all successful programs. Interventions that were pre-release only (only delivered whilst incarcerated), did not focus on women, were mix gendered and did not report gendered data separately, were excluded. Systematic reviews, meta-analyses, protocol papers and studies that did not evaluate a program were also excluded.

Data extraction and quality assessment
Search results were imported into Endnote X7 software, duplicates removed, and results exported into Covidence online software. Two investigators independently applied eligibility criteria to titles and abstracts and discrepancies identified through the platform were discussed in a face-to-face meeting. Studies that were included were progressed to full-text review where the investigators systematically went through individual articles thoroughly to check eligibility and documented reasons for exclusion. Discrepancies were resolved through face-toface discussion and a third reviewer was approached when needed. The lead reviewer extracted data according to the template for intervention description and replication (TIDieR) checklist and guide (Hoffmann, Glasziou, & Boutron, 2014) into a Microsoft Excel spreadsheet.
Finally, the investigators independently evaluated the risk of bias of studies using the revised Cochrane Risk of Bias tool for randomized trials (RoB2) (Sterne JAC et al., 2019) and the Risk of Bias in Non-randomized Studiesof Interventions (ROBINS-I) tool (Sterne JAC et al., 2016) (See supplementary files 1-2 for full version assessment tools). Each study was scored one point for each criterion that was fully met, half a point (0.5) if a criterion was 'somewhat' met, and zero for criteria that were either not met ('no') or not applicable. Each paper's score was estimated by summing the criteria scores and dividing the total by the number of applicable fields (excluding those criteria which did not apply) and multiplying by 100. Scores < 0.50 were characterised as 'low/ moderate quality' and > 0.50 as 'fair quality'. After the investigators individually assessed studies, they resolved discrepancies through discussion. It should be noted that due to the type of intervention being assessed it was not possible to blind participants, staff, or outcome assessors to participant allocation. We therefore did not score this against the studies performance or detection bias (RoB2 criteria numbers 2.1, 2.2, 4.3; ROBINS-I criteria numbers 6.2).

Data synthesis
Tables and text were generated to report study and program characteristics and outcomes. An intervention matrix was created, and descriptive numerical analyses were performed using Microsoft Excel.

Results
The original database search included 1047 citations and the updated search included in 446 citations, resulting in 1493 citations. After the removal of 785 duplicates and 589 articles through title and abstract screening, we reviewed 119 full text articles, of which 105 were excluded as they did not fit the selection criteria. Eleven articles met the criteria with one additional article included following hand-searching, resulting in a total of 12 articles for review (see Fig. 2).

Recidivism
The measure recidivism varied between studies and was used to quantify different crime-related events post-release (Table 2). A return-to-custody (RTC) was the most commonly used measure for recidivism (n = 8) Farrell-Macdonald et al., 2014;J. E. Johnson et al., 2011;Nyamathi et al., 2018;Schram & Morash, 2002;Scott et al., 2017). Recidivism was also a measure of re-arrest rates (n = 4) J. E. Johnson et al., 2011;Needels et al., 2005;Scott et al., 2017), the rate of reoffending (n = 1)  and being conviction-free at follow-up (n = 1) . Six studies used more than one measure for recidivism Guydish et al., 2011;J. E. Johnson et al., 2011;Miller et al., 2016;Needels et al., 2005;Scott et al., 2017), whereas six used a single measure Covington et al., 2008;Farrell-Macdonald et al., 2014;Nyamathi et al., 2018;Schram & Morash, 2002). In total, five of six transitional studies (83%) reported significant reductions in reoffending compared to the control arm Miller et al., 2016;Schram & Morash, 2002). Three post-release programs saw some effects: two had within group effects Scott et al., 2017); and one study reported reduced recidivism but lacked follow-up data to preclude significance . Table 3 visually breaks down study characteristics and the correlation between recidivism outcomes. Of which five/eight (62.5%) incorporated community case management Miller et al., 2016;Nyamathi et al., 2018;Schram & Morash, 2002;Scott et al., 2017); five/seven (71.4%) reported being genderresponsive    RMCs had favourable effects on women in the community who were not on probation but no effect on those on probationnon-probation women (who were assigned to RMC) at the beginning of the quarter were more likely than the control group to engage in any days of substance-use             RMC -Recovery Management Check-ups; Sresults were statistically significant; SAP + aftercare -prison-based substance abuse program and community-based after-care; SAP only -prison-based substance abuse program only (pre-release); WITthe Women's Integrated Treatment model; yyears; 1 -Community case management; 2 -Gender responsive; 3 -Referrals to services; 4 -Cognitive behavioural treatment; 5 -Imbedded substance abuse treatment; 6 -Imbedded MH and/or trauma services; 7 -Health promotion initiatives; 8 -Drug substitution therapy; 9 -Housing support *Farrell-MacDonald reported that follow-up data was collect 27 months post-release but as they were unclear on intervention length we do not know how long post-treatment the follow-up data was collected. For this we reported 27 months for post-treatment effects  and another study  saw reductions but lacked follow-up data to preclude significance. The attributes that supported these programs included SUD, MH and trauma treatment services (100%) J. E. Johnson et al., 2011;Scott et al., 2017); two programs (66.7%) were gender-responsive Scott et al., 2017), two (66.7%) had community case management Scott et al., 2017) and two (66.7%) used cognitive behavioural therapies J. E. Johnson et al., 2011). No correlation between reduced substance-use and recidivism post-release was seen.

Discussion
This is the first systematic review to examine postrelease and transitional programs offered to women with SUDs exiting prison to the community. In total we found 12 articles, which examined 11 programs, dating back to 2002, all conducted in North America. The objective of this review was to highlight the evidence about the effectiveness of post-release and transitional programs offered to women with SUDs and reveal what program attributes were common among successful programs.
The preliminary findings suggest that transitional programs had greater effects at reducing recidivism compared to post-release alone (83% compared to 50%). A major benefit for transitional programs is the continuity of care from prison to the community. Transitional support has been previously shown to assist participants in retaining rehabilitative health gains and reducing the risk of injury and death which is high for women with SUDs post-release (Abbot, Magin, Lujic, & Hu, 2017;Feild, 1998;MacDonald, Williams, & Kane, n.d.;Sullivan et al., 2019). Furthermore, as previously discussed, qualitative data also supports the use of transitional programs, as they facilitate pre-release linkage to health and social services in the community.
We were unable to make any correlations between substance-use and recidivism due to a limited pool of studies that reported substance-use as an outcome (RQ2). This is problematic considering all studies included women with SUDs and the direct correlation between substance-use and criminal offending for women is well understood (Fearn et al., 2016;H. Johnson, 2006). A major strength of this study is that it was the first to review and explore a variety of post-release and transitional programs for women with SUDs. As a result, we were able to critically examine the specific attributes of each program and make correlations between those attributes and improved post-release outcomes. Future research in this area can design or incorporate our findings into their interventions to further improve post-release outcomes for women exiting prison.
Five programs reported that allocation to the intervention group significantly reduced recidivism compared to the control group and another three concluded promising effects (RQ1). The attributes that contributed to the success of these programs were transitional, genderresponsive interventions which provided individualised support through community case management, with the use of cognitive behavioural therapies, as well as having substance-use, MH and trauma services available (whether it was imbedded, or women were referred to external services) (RQ3). Six studies reviewed substanceuse post-release, and of those, three reported reduced substance use among program participants (RQ1). Reductions in substance-use was associated with programs that offered gender-responsive support, used cognitive behavioural therapies, and provided substance-use treatment, MH and trauma services (imbedded or referred) (RQ3).
These findings reinforce the existing evidence that the design of transitional programs need to address criminogenic risk factors of women in prison (Borzycki, 2005;Borzycki & Baldry, 2003;Carlton & Segrave, 2016) and indicates the benefit of programs tailored to these characteristics and needs. However, we cannot determine from these studies the specifics of what was delivered to women under the banner of 'individualised support through community case management' or 'gender-responsive' interventions. In this review all studies that incorporated community case management included the role of a case manager who provided individualised links between women and external community-based services. There was no clear identification of what services women prioritised, were referred to, or managed to attend, nor the duration of attendance. Case management is the coordination of health and social services for a particular person. When employed effectively, it can bridge the services received inside prison and connect clients to appropriate community services, improving interagency information-sharing and continuity of care for individual clients (Corrective Services NSW, 2017;Feild, 1998;Warwick, Dodd, & Neusteter, 2012). The flow on effects of improved wellbeing and rehabilitation results in increased survival-time in the community, improved health outcomes including substance-use, which ultimately improves recidivism rates for participants. These preliminary results support the use of community case management. However, further high evidence trials that clearly describe and measure the services women are referred to are needed to continue to build on the evidence pool for women exiting prison with an SUD. Similarly, many programs described their intervention as gender-responsive without any further description of what that involved. It should be noted that gender-responsive programming must include creating an environment through site and staff selection, and program development, content and material that reflects an understanding of the realities of the lives of women in criminal justice settings and addresses their specific challenges and strengths (Covington & Bloom, 2006). In this review, five out of seven genderresponsive studies had an impact on recidivism. In addition, one study  provided community case management but did not state whether it was gender-responsive or not. It could be argued that case management is gender-responsive as it provides individualised support by linking to services based on an individual needs assessment which would target criminogenic needs, which should therefore be based on gender. This highlights two main points: 1) clearer reporting is required on what is provided when an intervention is described as genderresponsive or including case-management; and 2) Gender-responsive approaches are important and we need further research to extrapolate the aspects of gender-responsive programs that are helpful to women.
Findings from quantitative studies have shown specific attributes associated with post-release success, qualitative literature suggests there are other essential program characteristics not discussed in this review. Incarcerated women and service providers who work directly with women exiting prison have reported that stable housing, employment and family-related needs are the most critical attributes to post-release success for women (Kendall, Redshaw, Ward, Wayland, & Sullivan, 2018;O'Brien P. & Leem N., 2007;B. E. Salem et al., 2013). In our systematic review, no studies measured employment and housing status, or child custody in the follow-up periods. Furthermore, qualitative studies have identified the importance of continuity of care, pre-release linkage and emphasised the importance of the relationship between service providers and women participants (J. E. Johnson et al., 2013;Kendall et al., 2018;O'Brien P. & Leem N., 2007;B. E. Salem et al., 2013). Whilst findings from our systematic review reinforce the evidence for transitional programs, they did not measure relational or acceptability aspects of program implementation. We suggest that future interventions involve key stakeholders (e.g. women with SUDs and service providers) in the program design process to get a deeper understanding of what women not only need but what attributes they want to be included in a post-release program.

Limitations
The current evidence suggests that women benefit from continuity of care from prison to the community, which incorporated gender-responsive programming and individualised case management. Generalisability is, however, limited by the fact that the majority of studies were conducted in the United States. It remains uncertain whether these programs will be effective with women in countries with a different social structure. Nevertheless, key program attributes are transferable and can inform program development.
The general scarcity of literature meant that we were unable to synthesise the true effectiveness of programs for women exiting prison with SUDs. A meta-analysis was not feasible due to the diverse range of included programs and methodological weaknesses including a lack of stringent study design and various chosen control groups, which in effect has impacted the ability to answer our research question with significance. Very few comparison groups were genuinely usual care or 'no treatment'. In most cases, the control group was receiving another program, thereby making it impossible to isolate the impact of the program under investigation.
Further, understanding the long-term impact of programs is limited due to a lack of appropriate follow-up data. Five programs did not capture data past the completion of the program. Where changes were found, there are limits to how long these changes could be assumed to last due to a lack of proper long-term followup. More research is needed on the effectiveness of post-release programs for women. They need to be of rigorous study design, with appropriate control groups and follow-up to allow evaluation of program effectiveness.
Some studies failed to report program length, frequency of intervention, and follow-up time-points. It is important to clearly describe intervention modalities so that appropriate comparisons can be made. Unexplained lost to follow-up was common among studies, with no detail on important outcomes such as program dropout, accommodation change, homelessness, rearrested/reincarceration, hospitalisation, or death. All critical to understanding the effectiveness of an intervention and recidivism. Follow-up timeframes are also an important indication of how well an intervention was able to influence participant actions post-release such as recidivism and substance-use. Many studies did not follow participants past the completion of the intervention not allowing measurement of long-term impact. A follow-up period of two years has been recommended by a number of researchers as being optimal to understand the longterm effects of a program on participants (Andersen & Skardhamar, 2015; Office of the Inspector of Custodial Services, 2014; Yukhnenko et al., 2019).
Recidivism is one of the most fundamental outcome measures used in criminal justice research (Duwe, 2017;King & Elderbroom, 2014;Leverentz, Chen, Christian, & Maruna, 2020;Urban Institute, n.d.). All studies in this review used recidivism to measure program success however, we found it was inconsistently measured and there was a lack of standardisation across studies. Another limitation relating to recidivism was that most studies used the term recidivism to express a single RTC event. This is a blunt measure, simplifying a complex series of events, failing to account for the legislative and policy context in which a RTC occurs. As a result, readers are given only a partial view of how the criminal justice system operates and the position of women within it. To make recidivism a more meaningful measure we must move beyond a single event that measures success/failure of a program. A series of events such as rearrest, reconviction and reincarceration post-release as well as desistence, time to arrest, offence type and severity (King & Elderbroom, 2014;Urban Institute, n.d.). This suite of measures provides a timeline of events to give readers and policy makers a clearer view of the post-release experience and challenges.
Accounting for the context of health and social disadvantage experienced by women in prison, utilising health and social measures is also required. Almost all women within each study reported having a SUD prior to incarceration, however only six studies reported substanceuse post-release. Furthermore, MH, trauma, child custody, housing and employment outcomes were not analysed. This is concerning, considering the extensiveness of research illustrating these characteristics and their influence on health and recidivism post-release (Baldry, 2010;Carlton & Segrave, 2016;Langan & Pelissier, 2001;Sullivan et al., 2019). Future studies should include, or at least measure, these determinants in any future analysis to give a deeper understanding as to why a program was successful or not.

Conclusion
There is a paucity of literature on the effectiveness of post-release programs for women exiting prison with a SUD and the studies available contain significant methodological and conceptual limitations. There is a breadth of research that outlines the differences of characteristics of men and women within the criminal justice system, however because women make up a small proportion of the total prison population, they have received limited research attention in comparison. Recidivism rates illustrate that remaining in the community after any period in prison is difficult for women with SUDs. The rising rates of women in prison is a serious health and social policy issue in the context of what is already known about the intersecting health and social inequality experienced by women in prison and the barriers to women accessing social determinants of health resulting from disempowerment within broader social structures. The results from this review indicate that transitional, gender-responsive programs that incorporate individualised community case management and target comorbid MH and SUD can have a significant impact on post-release outcomes. Building upon these findings, development of programs for women transitioning back into the community should as a first step incorporate nuanced measures for recidivism and integrate the successful program attributes highlighted by this review.