Substance Use and Recidivism Outcomes for Prison-Based Drug and Alcohol Interventions

Abstract

We conducted a systematic review to examine the substance use and recidivism outcomes of prison-based substance use interventions. We searched public health, criminology, and psychology databases, and conducted forward and backward snowballing methods to identify additional studies. Studies were included if they were published between January 1, 2000 and June 30, 2017; were published in English; and reported substance use and/or recidivism outcomes of prison-based substance use interventions. Studies were reviewed for methodological rigor using the Effective Public Health Practice Project’s Quality Assessment Tool for Quantitative Studies. Our search returned 49 studies: 6 were methodologically strong, 20 were moderate, and 23 were weak. Results suggest therapeutic communities are effective in reducing recidivism and, to a lesser extent substance use after release. There is also evidence to suggest that opioid maintenance treatment is effective in reducing the risk of drug use after release from prison for opioid users. Furthermore, care after release from prison appears to enhance treatment effects for both types of interventions. Results provide evidence that policymakers can use to make informed decisions on best-practice approaches when addressing prisoner substance dependence and improving long-term outcomes. This comprehensive review highlights the difficulties of conducting quality research in the prison setting and suggests innovative study design for future research.

health policy, prisoner, substance abuse, substance addiction, vulnerable populations

INTRODUCTION

Drug misuse and dependence in offending populations present significant challenges for public health and justice systems. There is now compelling evidence for the relationship between drug use and criminal behavior (1–3), with 1 meta-analysis finding drug users were 7–8 times more likely to offend than nondrug users (4). In Australia, 58% of prisoners reported high-risk drinking and 67% admitted to illicit drug use in the 12 months prior to imprisonment (5). They also reported 10 times the rate of methamphetamine use than the general population (5). In the United States, 65% of prisoners are diagnosed as substance dependent in accordance with the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, criteria (6). Not surprisingly, prisoners in Australia and the United States are, respectively, 5 and 7 times more likely than the general population to have a substance abuse disorder (6, 7). Similar disparities have been reported in England (8, 9).

After release from prison, many ex-prisoners return to risky alcohol and drug use (10–12). Hazardous drug use after release increases the risk of infectious disease (13), as well as fatal (14, 15) and nonfatal overdose (16). Ex-prisoners are also at increased risk of reoffending soon after release (17–19). The risk of recidivism increases with the use of particular drugs (i.e., alcohol, cannabis, amphetamines and opioids), and risky patterns of use (e.g., injecting drugs) (20, 21).

In 1993, the World Health Organization recommended the implementation of opiate maintenance treatment (OMT) in prisons to reduce drug injection and related harms such as the spread of blood-borne diseases (22). In OMT, synthetic opioid medication such as methadone, buprenorphine, and naltrexone is used to help manage the adverse effects of withdrawal from opioids (e.g., heroin). Authors of previous reviews generally have supported opioid maintenance programs and their effectiveness in reducing relapse into criminal behavior or drug use for opioid users (23–25).

Several psychological treatment options have also been adopted or trialed by prison systems, including motivational interviewing (MI), therapeutic communities, and cognitive behavior therapy (CBT) (26). MI is a counselling approach that seeks to increase prisoner motivation and readiness for change. Previous studies on the effectiveness of MI in reducing substance use and recidivism in prisoners have produced mixed results (26, 27).

Therapeutic communities accommodate prisoners receiving treatment in distinct treatment units away from nonparticipating prisoners. They are characterized by resident involvement, a supportive and confrontational approach, and focus on the underlying disorders rather than solely on drug misuse. There is some evidence from studies that therapeutic communities are effective in addressing drug use and, to a lesser extent, recidivism in prisoner populations (23, 26, 28, 29).

CBT programs for prisoners are designed to change thinking patterns that may foster criminal behavior and substance use. The aim of using CBT is to provide prisoners with the coping strategies to deal with high-risk situations for drug use. Although CBT was effective in reducing recidivism and drug use in some studies, review findings remain equivocal (26, 30).

Despite the World Health Organization recommendations and evidence of effectiveness, few prisoners receive drug and/or alcohol treatment while in prison. Moreover, opportunities for prison-based intervention are often missed; for example, only 11% of US prisoners receive drug treatment (most of which is not evidence-based) (6) and only 8% of Australian discharged prisoners accessed an alcohol treatment program while in prison (5).

Inconsistent results were reported in several systematic reviews conducted in the 1990s on drug and alcohol treatments for adult prisoners (31, 32). More recently, 2 noteworthy reviews have been conducted on this topic. The first is a Campbell Systematic Review meta-analysis of studies published between 1980 and 2011 (33). The second is a meta-review of 3 meta-analyses including the Campbell Systematic Review and 2 earlier reviews covering studies that date back to 1968 (34). According to both reviews, therapeutic communities have a positive effect on recidivism and drug use; evidence relating to the effect of other treatment programs is less clear. In reviews including publications since 2011, researchers have largely focused on a specific type of substance or treatment, only reported substance use or recidivism outcomes, or included only randomized controlled trials (RCTs) (24, 25, 35–37). These specific foci make it difficult to compare across treatment options and assess the influence of treatments on substance use and recidivism outcomes. Furthermore, approaches to treating substance misuse/dependence in prisoner populations have changed significantly in the past 2 decades, as have the rates and types of substance use reported by prisoners, the types of prisons available, and support available before and after release.

The present study

In this study, we aimed to systematically review all studies published since 2000 that evaluated the substance use and/or recidivism outcomes of prison-based drug and alcohol interventions (including psychological and pharmacological). We provide a comprehensive and important update on the evidence base for prison-based drug and alcohol interventions, based on the findings from the review, and identify the most effective treatments for this vulnerable population.

METHOD

Search strategy

We conducted a systematic search of public health, criminology, and psychology databases, including Medline, Cumulative Index of Nursing and Allied Health, PsycARTICLES, PsycINFO, CINCH Australian Criminology Database, and ProQuest Criminal Justice. We searched for studies published between January 1, 2000, and June 30, 2017. We focused only on studies published from 2000 because of the changing nature of prison management, drug use in the prisoner population, and treatment options. Search strings pertained to the intervention and study population and did not place any restriction on study design or outcome. Two search strings were used—one for the title and one for the abstract—using truncation and Boolean and proximity operators. The search string for the title search was: (intervent* OR treatment* OR rehabilitat* OR *therap* OR counsel*) N5 ((substance OR drug OR alcohol OR illicit*) W3 (addict* OR abuse OR depend* OR disorder OR us* OR misuse OR treat*)) N5 (prison* OR jail* OR gaol OR correction* OR imprisonment OR incarcerat*). The title search string was paired with the abstract search string using the OR operator. The search string for the abstract was (intervent* OR treatment* OR rehabilitat* OR *therap* OR counsel*) N30 AB ((substance OR drug OR alcohol OR illicit*) W3 (addict* OR abuse OR depend* OR disorder OR us* OR misuse or treat*) OR (“drive* while intoxicated” OR DWI OR DUI)) AND (*prison* OR jail* OR gaol OR correction* OR imprisonment OR incarcerat*) NOT “literature review” OR (“smoking cessation” or “smoking intervention” or “smoking ban” or “anti-smoking” or tobacco* or nicotine*) OR (“drug court” or probation or “psychiatric hospital*”) OR (child* or adolescent or teen or juvenile) OR homeless* OR (animal* or rat* or mice or mouse) OR (gene* therap*) OR “emergency department”. The search was limited to academic journals in English. We used forward and reverse snowballing techniques to identify additional articles.

Inclusion and exclusion criteria

We included in this review studies that assessed the effectiveness of drug and/or alcohol treatment programs administered or initiated in prison or jail. Specifically, studies were included if they met the following criteria: the study population had been classified (or self-identified) as drug and/or alcohol users or engaged in risky drug-taking practices; substance use after intervention and/or recidivism after prison release (i.e., self-reported reoffending, rearrest, or reincarceration) was reported; published between January 1, 2000, and June 30, 2017; and available in English. Studies were excluded if they were qualitative, were drug-abuse interventions that were not prison-based (e.g., drug courts or diversion processes), and if the population comprised juveniles (i.e., not in adult prison).

Methodological quality assessment

We used the Effective Public Health Practice Project’s Quality Assessment Tool for Quantitative Studies (38) to assess the methodological quality of studies. This tool is considered to have high construct validity, content validity, and inter-rater reliability (38). Each study was rated on 6 criteria: selection bias, study design, confounders, blinding, data collection method, and withdrawals. After rating each category as “strong,” “moderate,” or “weak,” an overall rating was assigned according to set criterion (38). Strong studies could not have a weak rating for any criterion. Moderate studies could have 1 weak rating, and weak studies could have 2 or more weak ratings. A “not applicable” rating was given for any criterion that did not apply to the study design (e.g., blinding in retrospective studies in which administrative data were used). A methodological quality assessment was conducted for each study by 2 of the authors (independently), and any discrepancies were resolved through discussion. A meta-analysis was not conducted, because the few studies that received a strong methodological quality rating were heterogeneous in treatment and outcome (39).

RESULTS

Search results

We initially found 2,017 publications through database searching that met keyword search criteria and an additional 36 using snowballing techniques. Of these 2,053, 707 were duplicates. Of the remaining 1,346, 1,277 were excluded on the basis of abstract review, resulting in 69 publications for full-text review. Of these 69, 5 were excluded because the outcomes did not meet inclusion criteria, 1 was excluded because it was the conference abstract of an included study, and 1 protocol paper was excluded. A total of 62 publications based on 49 studies were included in the review. This included 1 paper published in 1999, the first in a series of papers related to the same study. The 2 follow-up papers were published in 2003 and 2004. The publication retrieval process is detailed in Figure 1, and follows Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (40).

Figure 1.

Flow diagram of the process used to identify, screen, and select studies published from January 2000 through June 2017. Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were used in the selection process.

The 49 included studies are summarized in Web Table 1 (available at https://academic.oup.com/aje). Most studies were conducted in the United States (n = 33). The remaining studies were conducted in Canada (n = 3), Sweden (n = 3), Puerto Rico (n = 2), Australia (n = 2), England (n = 1), China (n = 1), Japan (n = 1), Spain (n = 1), France (n = 1), and Taiwan (n = 1). Twenty-three studies included only male participants, 10 included only women, 12 included men and women, and the sex of participants was not specified in 4 studies. Of the 49 studies, 30 tested psychological interventions, 16 tested pharmacological interventions, 2 tested a combination of psychological and pharmacological interventions, and 1 study tested an acupuncture intervention.

The following types of psychological interventions were tested: therapeutic communities (19, 41–55), CBT (56–58), MI (12, 59–61), screening, brief intervention and referral to treatment (62), acceptance and commitment therapy (63–65), personalized feedback intervention (66), and a meditation program (67); the remainder were generic drug and alcohol programs (e.g., counselling; “drug treatment”) (68–74). All 16 studies that were trials of pharmacological interventions tested OMTs. Specific types included methadone (75–83), buprenorphine-naloxone (84), naltrexone (85), extended-release naltrexone (XR-NRT) (86–88), l-α-acetylmethadol treatment (89), and buprenorphine (including 2 studies comparing buprenorphine with methadone and 1 study comparing buprenorphine with dihydrocodeine) (90–94). Two interventions combined psychological and pharmacological approaches. One trialed a combination of prison-initiated methadone, counselling, and continuity of care with OMT after prisoner release (95–98), and the other trialed a combination of CBT and methylphenidate (typically used to treat attention deficit hyperactivity disorder) (99). A final study trialed an alternative medicine, auricular acupuncture treatment (100).

Twenty-nine of the studies were RCTs or controlled clinical trials, 13 were case-control studies, 3 were cohort analytic studies, and 4 were cohort studies. Substance use outcomes were reported in 36 studies, recidivism outcomes were reported in 35, and both outcomes were reported in 21 studies. Details on the outcomes of each study are provided in Web Table 1. Study outcomes are simplified in Table 1 as either a positive effect, no effect, or a negative effect, based on the significant findings (P < 0.05) of individual studies as presented by authors.

Table 1.

Substance Use and Recidivism Treatment Outcomes for Prison-Based Drug and Alcohol Interventions (n = 49), January 2000–July 2017

First Author, Year (Reference No.)	Rating	Intervention Type	Outcomes
			Substance Use	Recidivism
Bahr, 2013 (68)	Moderate	Psychological		Positive
Bowen, 2006 (67)	Weak	Psychological	Positive	None
Forsberg, 2011 (59)	Weak	Psychological	Positive	None
Hser, 2013 (70)	Moderate	Psychological	None	None
Jensen, 2010 (42)	Weak	Psychological		Positive
Jensen, 2012 (41)	Moderate	Psychological		None
Johnson, 2012 (72)	Strong	Psychological	None
Kopak, 2015 (73)	Weak	Psychological		Positive
González-Menéndez, 2014 (63)	Moderate	Psychological	Positive
Lanza, 2014 (64)	Moderate	Psychological	Positive
Lanza, 2013 (65)	Moderate	Psychological	Positive
McKendrick, 2006 (43)	Weak	Psychological	Positive	Positive
Olson, 2014 (44)	Moderate	Psychological		Positive
Owens, 2016 (61)	Moderate	Psychological	Positive
Pelissier, 2005 (56)	Weak	Psychological	Negative	Positive
Porporino, 2002 (74)	Moderate	Psychological		Positive
Prendergast, 2003 (55)	Moderate	Psychological	Positive	Positive
Prendergast, 2004 (19)	Moderate	Psychological	Positive	Positive
Prendergast, 2017 (62)	Moderate	Psychological	None	None
Sacks, 2004 (48)	Weak	Psychological	Positive	Positive
Sacks, 2008 (46)	Weak	Psychological	Positive	Positive
Sacks, 2012 (47)	Moderate	Psychological		Positive
Sacks, 2012 (45)	Weak	Psychological	Positive	Positive
Stein, 2010 (12)	Strong	Psychological	Positive
Sullivan, 2007 (49)	Weak	Psychological	Positive	Positive
Turley, 2004 (72)	Weak	Psychological		Positive
Van Stelle, 2004 (50)	Weak	Psychological	Positive
Vaughn, 2003 (69)	Weak	Psychological	Negative	Negative
Welsh, 2007 (51)	Moderate	Psychological	None	Positive
Welsh, 2013 (52)	Moderate	Psychological	None	Positive
Welsh, 2014 (53)	Moderate	Psychological	None
Wexler, 1999 (54)	Moderate	Psychological	Positive	Positive
Woodall, 2007 (60)	Weak	Psychological	Positive	None
Yokotani, 2015 (67)	Moderate	Psychological		Positive
Zlotnick, 2003 (58)	Weak	Psychological	Positive
Zlotnick, 2009 (57)	Moderate	Psychological	None	None
Cropsey, 2011 (90)	Strong	Pharmacological	Positive
Dolan, 2003 (75)	Moderate	Pharmacological	Positive	Positive
Dolan, 2005 (76)	Moderate	Pharmacological	Positive	Positive
Farrell-MacDonald, 2014 (79)	Strong	Pharmacological		Positive
Garcia, 2007 (84)	Strong	Pharmacological	Positive	None
Gordon, 2017 (94)	Weak	Pharmacological	None	None
Heimer, 2005 (80)	Weak	Pharmacological	Positive
Heimer, 2006 (81)	Weak	Pharmacological	Positive
Kinlock, 2005 (89)	Weak	Pharmacological	None	None
Larney, 2012 (77)	Moderate	Pharmacological	Positive	Positive
Lee, 2015 (88)	Weak	Pharmacological	Positive	Positive
Lee, 2015 (86)	Moderate	Pharmacological	Positive
Lee, 2016 (87)	Moderate	Pharmacological	Positive
MacSwain, 2014 (78)	Weak	Pharmacological		Positive
Magura, 2009 (92)	Moderate	Pharmacological	None	None
Marzo, 2009 (93)	Strong	Pharmacological		Positive
McKenzie, 2012 (82)	Weak	Pharmacological	Positive	None
McMillan, 2008 (83)	Weak	Pharmacological		None
Sheard, 2009 (91)	Weak	Pharmacological	Positive
Shearer, 2007 (85)	Weak	Pharmacological	Positive
Gordon, 2008 (95)	Moderate	Combination	Positive	Positive
Kinlock, 2007 (97)	Moderate	Combination	Positive	Positive
Kinlock, 2008 (98)	Moderate	Combination	Positive	Positive
Kinlock, 2009 (96)	Moderate	Combination	Positive	Positive
Konstenius, 2014 (99)	Moderate	Combination	Positive
Berman, 2004 (100)	Weak	Other	Negative

Drug use was reported in studies by urinalysis (58, 61, 63–65, 70, 72, 76, 80, 81, 84–88, 90, 91, 94–100), hair and nail testing (19, 54, 55), breath testing (72), or self-report (usually use in 30 days prior) (12, 19, 43, 45–50, 54–56, 58–60, 62–65, 67, 69, 70, 72, 75–77, 80–82, 85, 86, 92, 94), or in official records on drug-related offending (66).

Methodological quality

Of the 49 studies, 6 were rated strong for methodological quality, 20 were rated moderate, and the remaining 23 were weak (Table 1). A summary of ratings for each study across Effective Public Health Practice Project’s criteria is presented in Table 2, based on the information reported in the relevant publications. Common limitations that reduced quality included small sample sizes, high attrition rates, lack of blinding, selection bias, and the use of self-report recidivism data.

Table 2.

Study Ratings Using Effective Public Health Practice Project’s (36) Criteria (n = 49), January 2000–July 2017

First Author, Year (Reference No.)	Selection Bias	Study Design	Confounders	Blinding	Data Collection	Withdrawals	Global Rating
Bahr, 2013 (68)	Moderate	Moderate	Strong	Weak	Strong	Strong	Moderate
Berman, 2004 (100)	Weak	Moderate	Strong	Weak	Strong	Weak	Weak
Bowen, 2006 (67)	Weak	Moderate	Strong	Weak	Strong	Weak	Weak
Cropsey, 2011 (90)	Strong	Strong	Moderate	Strong	Strong	Strong	Strong
Dolan, 2003 (75)	Weak	Strong	Strong	Moderate	Strong	Moderate	Strong
Dolan, 2005 (76)	Weak	Strong	Strong	Moderate	Strong	Moderate	Strong
Farrell-MacDonald, 2014 (79)	Moderate	Moderate	Strong	N/A	Moderate	Strong	Strong
Forsberg, 2011 (59)	Moderate	Strong	Strong	Weak	Weak	Moderate	Weak
Garcia, 2007 (84)	Moderate	Moderate	Strong	N/A	Strong	Moderate	Strong
González-Menéndez, 2014 (63)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Gordon, 2008 (95)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Gordon, 2017 (94)	Moderate	Strong	Weak	Weak	Strong	Strong	Weak
Heimer, 2005 (80)	Moderate	Moderate	Strong	Weak	Moderate	Weak	Weak
Heimer, 2006 (81)	Moderate	Moderate	Strong	Weak	Moderate	Weak	Weak
Hser, 2013 (70)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Jensen, 2010 (42)	Weak	Moderate	Weak	N/A	Moderate	Moderate	Weak
Jensen, 2012 (41)	Moderate	Moderate	Strong	N/A	Moderate	Weak	Moderate
Johnson, 2012 (72)	Moderate	Strong	Strong	Strong	Strong	Strong	Strong
Kinlock, 2005 (90)	Moderate	Moderate	Weak	Weak	Strong	Weak	Weak
Kinlock, 2007 (97)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Kinlock, 2008 (98)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Kinlock, 2009 (96)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Konstenius, 2014 (99)	Moderate	Strong	Strong	Strong	Strong	Weak	Moderate
Kopak, 2015 (73)	Weak	Moderate	Strong	Moderate	Weak	Weak	Weak
Lanza, 2013 (65)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Lanza, 2014 (64)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Larney, 2012 (77)	Weak	Strong	Strong	Moderate	Strong	Moderate	Strong
Lee, 2015 (86)	Moderate	Strong	Strong	Weak	Strong	Moderate	Moderate
Lee, 2015 (88)	Moderate	Strong	Moderate	Weak	Moderate	Weak	Weak
Lee, 2016 (87)	Moderate	Strong	Strong	Weak	Strong	Moderate	Moderate
MacSwain, 2014 (78)	Moderate	Moderate	Weak	Weak	Moderate	Moderate	Weak
Magura, 2009 (92)	Moderate	Strong	Strong	Weak	Strong	Moderate	Moderate
Marzo, 2009 (93)	Moderate	Moderate	Strong	N/A	Strong	Strong	Strong
McKendrick, 2006 (43)	Weak	Strong	Strong	Weak	Moderate	Moderate	Weak
McKenzie, 2012 (82)	Moderate	Strong	Weak	Weak	Weak	Moderate	Weak
McMillan, 2008 (83)	Moderate	Moderate	Strong	Weak	Strong	Weak	Weak
Olson, 2014 (44)	Weak	Strong	Strong	Moderate	Moderate	Moderate	Moderate
Owens, 2016 (61)	Moderate	Strong	Strong	Weak	Strong	Moderate	Moderate
Pelissier, 2005 (56)	Weak	Moderate	Strong	Moderate	Weak	Weak	Weak
Porporino, 2002 (74)	Weak	Moderate	Strong	Moderate	Strong	Strong	Moderate
Prendergast, 2003 (55)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Prendergast, 2004 (19)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Prendergast, 2017 (62)	Moderate	Strong	Moderate	Weak	Moderate	Moderate	Moderate
Sacks, 2004 (48)	Weak	Strong	Strong	Weak	Moderate	Moderate	Weak
Sacks, 2008 (46)	Weak	Strong	Strong	Weak	Moderate	Weak	Weak
Sacks, 2012 (45)	Weak	Strong	Strong	Weak	Strong	Strong	Weak
Sacks, 2012 (47)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Sheard, 2009 (91)	Weak	Strong	Strong	Weak	Strong	Moderate	Weak
Shearer, 2007 (85)	Weak	Moderate	Weak	Weak	Moderate	Weak	Weak
Stein, 2010 (12)	Moderate	Strong	Strong	Moderate	Strong	Moderate	Strong
Sullivan, 2007 (49)	Weak	Strong	Strong	Weak	Moderate	Moderate	Weak
Turley, 2004 (71)	Weak	Moderate	Strong	Weak	Weak	Weak	Weak
Van Stelle, 2004 (50)	Weak	Moderate	Weak	Weak	Moderate	Weak	Weak
Vaughn, 2003 (69)	Moderate	Moderate	Weak	Weak	Moderate	Weak	Weak
Welsh, 2007 (51)	Moderate	Moderate	Strong	Moderate	Strong	Weak	Moderate
Welsh, 2013 (52)	Moderate	Moderate	Strong	Weak	Strong	Moderate	Moderate
Welsh, 2014 (53)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Wexler, 1999 (54)	Moderate	Strong	Strong	Weak	Strong	Strong	Moderate
Woodall, 2007 (60)	Weak	Strong	Strong	Weak	Strong	Strong	Weak
Yokotani, 2015 (66)	Moderate	Strong	Strong	Weak	Moderate	Strong	Moderate
Zlotnick, 2003 (58)	Moderate	Moderate	Strong	Weak	Weak	Strong	Weak
Zlotnick, 2009 (57)	Moderate	Moderate	Strong	Weak	Strong	Strong	Moderate

Abbreviation: N/A, not applicable.

Substance use outcomes

Substance use outcomes were reported in 36 studies: 21 (58%) were psychological interventions, 12 (33%) were pharmacological interventions, 2 (6%) used a combined psychological and pharmacological intervention, and 1 (3%) was an acupuncture intervention.

Psychological

Approximately one-half (n = 12 of 21) of all psychological interventions in which substance use outcomes were reported had a significant positive result, no significant differences were found in 7, and a negative result was reported in 2. Although most psychological studies with positive substance use outcomes were methodologically weak (n = 8 of 12), all 7 studies finding no significant change were of moderate (n = 6) to strong (n = 1) quality. All 4 MI studies were RCTs of mixed quality and the intervention was reported to have increased abstinence significantly (12, 59–61). In the first study, improvements were found in past-month drug abstinence at 10 months after release across all 3 arms (i.e., MI with prison staff counsellors vs. MI with prison staff counsellors who have peer group supervision vs. usual planning interview), but no between-group differences were found (59). MI was used in the remaining 3 studies to address hazardous drinking (12, 60, 61). In the first, a MI drunk-driving intervention had a positive treatment effect (compared with no treatment) that was maintained over a 2-year follow-up (60). In the second study, 245 alcohol-dependent incarcerated women who received MI achieved significantly more days abstinent at 3 months after release compared with those in the assessment-only control condition, but these effects had attenuated by 6 months after release (12). In the third study, MI resulted in significantly greater improvements in past-month abstinence at 1 month after release compared with an educational intervention control (61).

The effectiveness of CBT was examined in 6 studies (45, 46, 56–58, 63–65). Only in one (a weak uncontrolled cohort study) did CBT result in a reduction in drug and alcohol use at 3 months after release (58). In 3 RCTs, CBT had inferior effects compared with acceptance and commitment therapy (63–65) and therapeutic communities (45, 46). Of the 8 studies on therapeutic communities, improvements in drug use outcomes were reported in 4 at 3–12 months’ follow-up after release (43, 45, 46, 48–50). However, the effects of therapeutic communities appear to diminish beyond 12 months, because in all 4 trials of therapeutic communities with follow-up periods between 2 and 5 years, researchers found no longer-term effects (19, 51–55). Several more generic psychological drug and alcohol programs were ineffective (Web Table 1) (62, 68–70, 72). In the only study to trial meditation (Vipassana), researchers found it significantly reduced use of crack cocaine, marijuana, and alcohol over 3 months’ follow-up compared with treatment as usual, including chemical dependency treatment and substance use education (67).

Pharmacological

OMT significantly reduced drug use by opioid users after release in 2 of the 3 high-quality pharmacological studies in which a drug use outcome was reported (84, 90), and no significant difference was found between treatment group and counselling during follow-up in 1 study (94). Significant reductions in heroin use among OMT participants compared with controls or participants receiving an alternative pharmacological treatment were reported in 8 of the 11 studies reporting opioid use in the 12 months after release from custody (75, 79–82, 84–88, 90, 91). Similar reductions in cocaine use in 2 of 4 OMT studies were reported (82, 84). XR-NRT was studied in 2 RCTs; OMT participants received one injection before release and one 6 months after release (86–88). Compared with a control group (no medication), significant reductions in opioid use and overdose (86–88) were reported for XR-NRT, although treatment effects were attenuated by 18 months. Conflicting results were found for cocaine use (86–88). Finally, in 1 RCT, participants who received prison-based methadone maintenance treatment (MMT) plus long-term community-based MMT after release had significantly lower risk of heroin use and fatal overdose than those who received prison-based MMT only or waitlist control (76, 77).

Three studies in which significant findings were not reported include a study trialing l-α-acetylmethadol treatment (89) and 2 RCTs. Buprenorphine maintenance treatment was compared with MMT in the first study (92). In the second study (a 2 × 2 × 2 factorial design), researchers tested combinations of in-prison treatment (i.e., buprenorphine treatment or counselling) with postrelease treatment (i.e., at an opioid treatment program or community health center) by sex (94).

Combined psychological and pharmacological interventions

In 2 RCTs of moderate quality, researchers compared a psychological intervention delivered with and without a pharmacological intervention and found participants in the combined intervention had significantly fewer positive urinalysis drug test results (95–99). Three interventions were evaluated in one of these studies: counselling versus counselling with transfer to methadone in the community versus counselling with methadone in prison and in the community. The authors reported the last condition was associated with the largest reductions in heroin and cocaine use, immediately after release and during the 12-month follow-up period (95–98). In the second of these studies, a combination of CBT and methylphenidate versus CBT alone was evaluated for the treatment of substance use disorder in prisoners with attention deficit hyperactivity disorder. Participants in the multicomponent arm were significantly more likely to have a negative urinalysis test (for amphetamines or other drugs) than participants in the control arm at the 6-month follow-up (99).

Other

The efficiency of auricular acupuncture (National Acupuncture Detoxification Association–Acudetox protocol) versus a placebo (nonspecific helix protocol) was compared in an RCT of weak quality. The National Acupuncture Detoxification Association–Acudetox group engaged in significantly more drug use than did the placebo group (100).

Recidivism

Psychological

Fifteen of the 23 psychological interventions resulted in a significant reduction in recidivism. Seven of the effective interventions were RCTs. Increased recidivism was reported for only 1 generic drug and alcohol intervention at follow-up (69). However, this finding should be treated with caution because the study suffered from selection bias and high attrition at 12-month follow-up (69). There were 11 studies that included a therapeutic communities intervention; in 10 of these, there were significantly lower rates of recidivism (i.e., rearrest or reincarceration) with follow-up periods between 6 months to approximately 6.9 years (19, 41–49, 51, 52, 54, 55). In 2 of these successful studies with long-term follow-up, those in the treatment condition who engaged in aftercare after release from prison experienced significantly lower rates of reincarceration (19, 44, 54, 55) and rearrest (19, 54, 55) compared with those in the control and treatment (without aftercare) conditions.

Recidivism outcomes for CBT interventions are reported in only a few studies; based on the findings from those studies, researchers suggest CBT has a limited effect on recidivism when implemented as a stand-alone approach. In one large, multisite CBT study, researchers evaluated 20 CBT programs for men and women: 16 CBT programs with male prisoners had no effect; a positive treatment effect was reported for 4 CBT programs with female prisoners during a 3-year follow-up (56). In 2 CBT studies of female prisoners with comorbid conditions (i.e., substance dependence and post-traumatic stress disorder), the effect of CBT on reincarceration was unknown (due to no comparison group in 1 study) after 3 months’ follow-up (58) and nonsignificant after 6 months’ follow-up (57). In 4 RCTs with follow-up periods of up to 12 months, CBT had inferior effects on recidivism compared with therapeutic communities (43, 45–49). Reductions in recidivism during follow-up were reported for 2 generic or multipronged interventions (nonpharmacological) with CBT elements (73, 74).

Findings from 2 studies in which MI was evaluated indicated this type of intervention had no effect on recidivism outcomes when used as a stand-alone option (59, 60). There were mixed results in recidivism outcomes from several other generic drug and alcohol interventions, with 1 having a positive effect (68, 71), 2 having no effect (66, 72), and 1 having a negative effect (69). Furthermore, a meditation intervention in prison had no effect on recidivism (67).

Pharmacological

Of 11 pharmacological studies for which a recidivism outcome was reported, reincarceration outcomes based on official records were reported for 9 (75–79, 82, 83, 92, 93). Reported arrests (self-report) also were reported in 2 of these studies (82, 92) and offending (self-report) was reported in the remaining two studies (84, 94). For 6 studies, the follow-up period was 12 months or less after release (82, 84, 88, 89, 92, 95). Much longer follow-up periods were reported on in the 5 remaining studies, with a maximum of 9 years and 7 months—all using survival analysis (76–79, 83, 93). Four of the 5 pharmacological studies that reported a significant treatment effect were from this group of studies for which substantial follow-ups periods were reported and in which a prison-initiated treatment with continued treatment after release was tested. In 3 of these studies for opioid users, a condition of OMT in prison was tested combined with OMT in the community. Results showed significantly lower reincarceration risk when compared with OMT only in prison or with no treatment (76–78). In another study in which OMT was provided after release, lower rates of reincarceration were reported among participants who received XR-NRT compared with control participants who received no treatment. However, reincarceration risk was only significantly lower for treatment participants if they received both prescribed injections: one prerelease injection of XR-NRT and one at 6 months after release; and the sample size in this study was very small (88). In an RCT in which buprenorphine was compared with methadone, no significant differences in reincarceration rates or self-reported arrest were found during the 3-month follow-up (92).

Combined psychological and pharmacological interventions

In an RCT with 12-month follow-up, researchers found that a combination of prison-initiated methadone, counselling, and continuity of care with OMT after release resulted in superior outcomes compared with only counselling, or counselling with methadone treatment after release (96).

DISCUSSION

This systematic review examined the substance use and/or recidivism outcomes of all studies in which prison-based drug and alcohol interventions were evaluated and that were published between January 2000 and June 2017. A total of 62 publications based on 49 studies were included. In the majority of studies (61%), the psychological treatments used were reviewed. Results of this present review concur with those of earlier reviews of therapeutic communities for prisoners (23, 26, 28, 29, 33, 34). Based on these results, therapeutic communities, in the short-term, appear to be effective for reducing recidivism and, to a lesser extent, substance use in ex-prisoners with drug misuse/dependence. In support of prior reviews that reported on OMT effectiveness for opioid users in prisons (24, 25), the current review provides good evidence that supports the use of OMT to treat opioid dependence in the prison setting, with findings of most studies of strong and moderate quality included in this review indicating OMT is effective for reducing drug use and recidivism after release, particularly when there is continuity of care after release.

As noted in other reviews (26, 30), CBT was largely ineffective as a stand-alone treatment in addressing drug use or recidivism after release in 6 studies. However, significant reductions in recidivism were reported when CBT was included as part of a multicomponent approach. Findings on MI add to the existing research (26, 27). We found that although MI shows promise in reducing the risk of drug use after release, it does not appear to reduce recidivism. Existing studies suffer from numerous methodological limitations, including small sample sizes, high attrition, and selection bias. Furthermore, the lack of blinding of outcome assessors and participants has resulted in few RCTs being rated as high, or even moderate, in quality.

Implications for policy and practice

From the findings of this review, it appears many of the drug and alcohol interventions trialed in prison settings result in better substance use and recidivism outcomes than no treatment. Despite this and the high rates of substance use in prisoners, few have access to evidence-based drug and alcohol treatment (101). In this review, we provide additional support for the World Health Organization’s recommendation to implement OMT in all prisons, and to reduce drug relapse and, to a lesser extent, recidivism after release for those with opioid dependence. From the research we reviewed, it also appears that gains made through psychological and pharmacological treatment in prison may be short lived after release without appropriate transitional support and a plan for continuity of health care (44, 75–79, 95–98). Additional investment in aftercare may be an area that can enhance outcomes in prisons with existing OMT and evidence-based psychological treatment practices in place.

Strengths and limitations

This review has 2 key strengths. We used broad inclusion criteria, including a diverse range of psychological and pharmacological interventions as well as other types of alcohol and/or drug interventions. This broad scope allowed us to compare evidence between and within interventions across 2 critical outcomes for drug using prisoners. This review will inform policymakers about current prison-based treatment options and provide the evidence base to make informed, best-practice decisions regarding treatment and support for prisoners before and after their release. Assessment of the methodological quality of the studies was conducted by 2 independent raters; there was limited disparity between ratings.

There were 3 main limitations to this review. First, assessment across studies identified considerable heterogeneity in study design, participant characteristics, treatment type and setting, and definitions and measures of the substance use and recidivism outcomes. As a result, it was not possible to conduct a meta-analysis. Second, the inclusion of studies published only in English was not ideal. Although this may affect the generalizability of the findings, this restriction only resulted in the exclusion of 1 known study. Furthermore, the Campbell Systematic Review (33) on the same topic had no restrictions on language, and still 88% of reviewed studies were from the United States (compared with 67% of studies examined in this review). Third, we only included published journal articles, which makes this review susceptible to publication bias. However, given the substantial proportion of studies in which no treatment effect or a negative treatment effect was reported on substance use outcomes, which were reported in 36% of studies, and recidivism outcomes, which were reported in 43% of studies, publication bias in this review is likely to be minimal. Furthermore, results of statistical testing for publication bias reported in a 2012 Campbell systematic review (a meta-analysis) on prison-based drug treatment indicated no evidence of publication bias in studies that included drug-relapse outcomes or in which counselling treatment was evaluated (33). However, evidence was found that suggested to us publication bias in studies on therapeutic communities (33).

Implications for research

In this review, we highlight the difficulties conducting high-quality research in prison settings. Although 29 RCTs were identified, only 6 studies were assigned a strong methodological rating, largely due to the lack of blinding. The adoption of a sophisticated research design (e.g., cluster RCT) in prison-based experiments may better facilitate blinding of participants. The lack of RCTs and quality studies was also noted in a recent systematic review of RCTs of health interventions in prison and after release (38). Authors of that review responded to this challenge by suggesting that the findings of research in other settings and with other populations may be informative. In this instance, applicable research may include studies in residential rehabilitation treatment.

In addition, conducting health and justice data linkage (prospective and retrospective) can greatly improve methodological rigor by reducing reliance on self-reported data, increasing knowledge of confounders and accounting for deaths during follow-up. Data linkage may also enhance understanding of the effect that drug and alcohol treatment has on health and offending trajectories, and highlight opportunities over the life course for preventive intervention. Outcome data can also be improved by collecting more detailed data on relapse patterns (e.g., through timeline follow-back). In many studies, researchers record substance use relapse as either complete abstinence since release or abstinence in the 30 days before follow-up. Neither measure indicates the severity of relapse.

We also identified several promising interventions that require further research in prison settings. Encouraging substance use outcomes from 1 study of acceptance and commitment therapy, and another of meditation require additional investigation. In addition, multipronged interventions that use pharmacological and psychological therapies and/or ensured continuity of care after release showed much promise for opioid-dependent prisoners (78, 79, 96). These promising approaches should be explored further with larger samples and longer follow-up. Evidence has emerged in this review that the good outcomes often resulting from therapeutic communities may be enhanced with engagement in aftercare and should be investigated further.

Increased risk of overdose was found in 2 studies of opioid-using prisoner populations (86, 87), as was death from overdose (95–98) in control (or no treatment) groups, but not OMT groups, during follow-up 1 year after treatment. Similarly, in an Australian methadone maintenance intervention (76), 8 of 382 participants had died of drug overdose by the 4-year follow-up; none died while receiving OMT. Such findings highlight the critical need for targeted, evidence-based treatment for opioid users that withstands the strain of community re-entry and potential disengagement from health services. The findings of this review highlight XR-NRT treatment before and after release for opioid users as a potential protective factor from drug relapse and overdose in the first critical weeks after release, and the need for additional trials with opioid-using prisoner populations.

Finally, with a significant proportion of studies being conducted in high-income countries, there is a clear need for investment in quality research in prison-based substance use treatment in low- and middle-income countries. Such research can address distinct substance use trends and the needs of local populations, while also informing local policy and practice in a meaningful way.

CONCLUSION

Despite the growing need for effective drug and alcohol treatments in prisons internationally, there is a lack of high-quality research available to inform policy and practice. In this review, we provide good evidence that therapeutic communities and OMT (for opioid users) are 2 of the more effective in-prison treatment options for reducing alcohol and drug use and recidivism, particularly when care is extended after release. Additional research using larger sample sizes, unbiased sampling, and sophisticated evaluation designs to overcome the challenges of conducting research with substance users in prison settings (including attrition during follow-up after release) is urgently required. Although there is a large body of literature on prison-based drug and alcohol treatments, there remains a compelling need for more innovative evaluations of prison-based alcohol and/or drug programs to directly inform evidence-based treatment for this vulnerable population.

ACKNOWLEDGMENTS

Author affiliations: Centre for Youth Substance Abuse Research, Institute of Health and Biomedical Innovation, School of Psychology and Counselling, Faculty of Health, Queensland University of Technology, Brisbane, Australia (Dominique de Andrade); Griffith Criminology Institute, Griffith University, Brisbane, Australia (Dominique de Andrade); TC Beirne School of Law, Faculty of Business, Economics and Law, University of Queensland, Brisbane, Australia (Jessica Ritchie); Centre for Accident Research and Road Safety- Queensland, Queensland University of Technology, Brisbane, Australia (Michael Rowlands); School of Public Health and Social Work, Faculty of Health, Queensland University of Technology, Brisbane, Australia (Emily Mann); and Lives Lived Well Research Group, School of Psychology, Faculty of Health and Behavioral Sciences, University of Queensland, Brisbane, Australia (Leanne Hides).

D.d.A. was supported by a Queensland Government Advance Queensland Women’s Academic Fund (2016–2017) and a Strategic Research Support Grant from the Institute of Health and Biomedical Innovation at Queensland University of Technology. L.H. was supported by an Australian Research Council Future Fellowship (2012–2016) and National Health and Medical Research Council Senior Research Fellowship (2017–2021).

Conflict of interest: none declared.

Abbreviations

 
• CBT

  cognitive behavior therapy

 
• MI

  motivational interviewing

 
• MMT

  methadone maintenance treatment

 
• OMT

  opiate maintenance treatment

 
• RCT

  randomized controlled trial

 
• XR-NRT

  extended-release naltrexone