Pharmacological interventions for drug-using offenders: an update to a systematic review and meta-analysis

This updated systematic review assesses the effects of pharmacological interventions for drug-using offenders. Systematic review protocols and conventions of the Cochrane Collaboration were followed to identify eligible studies. Studies were pooled in a meta-analysis to assess the impact of pharmacological interventions on drug use and criminal activity. An economic appraisal was conducted. The search strategies identified 22 studies containing 4372 participants. Meta-analyses revealed a small statistically significant mean difference favouring pharmacological interventions relative to psychological interventions in reducing drug use and criminal activity. When comparing the drugs to one another there were no significant differences between those included (methadone versus buprenorphine, naltrexone and cyclazocine). Overall, the findings of this review suggest that methadone and naltrexone may have some impact on reducing drug use and reincarceration. Individual pharmacological drugs had differing (generally non-significant) effects. One study identified serious adverse events. Three studies reported cost and consequences information sufficient to conduct a full economic analysis but this was not comprehensive enough to be able to make judgements across all treatment options. Full economic analyses should be encouraged. The study findings were limited mainly to male adult offenders.


Introduction
Offenders as a socially excluded group of people demonstrate significant drug use and subsequent health problems. Studies investigating the prevalence of drug dependence in UK prisons report variable figures of 10% (Gunn et al., 1991), 39% (Brooke et al., 1996) and 33% (Mason et al., 1997). In a review article, the estimated prevalence of drug abuse among male and female prisoners was estimated to be 10 to 48% and 30 to 60%, respectively (Fazel et al., 2006). A recent survey carried out on the Swedish criminal justice system found that 70% of all individuals had substance abuse problems (including alcohol, drugs or both), with a slightly lower percentage among females compared with the whole group (67%) (Klientkartlaggningen, 2015). These numbers can be compared with the general population in Sweden where 1% of all men and 0.5% of all women fulfil the criteria for drug dependence or abuse (Ramstedt et al., 2013). Similar trends have been reported elsewhere. In France, 30% of prison inmates are heroin addicted, and in Australia, 59% of prison inmates report injecting (primarily heroin) drug use histories. In 2018 in the USA, an estimated 6,410,000 persons were held in prisons or jails, or were on probation or parole (Maruschak and Minton, 2020). Of the prison population, 85% are likely to have an active substance use disorder or were incarcerated for a crime involving drugs or drug use according to the US National Institute on Drug Abuse (National Institute on Drug Abuse (NIDA), 2020). In the US, it is recognised that many offenders need treatment to tackle their drug use (Lo and Stephens, 2000). The link between drug use, subsequent health, social and criminological consequences is well documented in the literature (Michel and Maguet, 2005), and offenders have a high risk of death from opioid overdose within 2 weeks of release from prison (Binswanger et al., 2007, Bird and Hutchinson, 2003, Kinner et al., 2013, Merrall et al., 2010. Substance use disorders are linked to criminal behaviour and are a significant burden on the criminal justice system. Approximately 30% of acquisitive crime is committed by individuals supporting drug use with the use of criminal acts (Magura et al., 1995).
Internationally, methadone maintenance has been the primary choice for chronic opioid dependence in prisons and jails, including those in the Netherlands, Australia, Spain and Canada. As a result, it is increasingly implemented in the criminal justice setting (Moller et al., 2007, Stallwitz and with the exception of the US which has not generally endorsed the use of methadone treatment. For example, of the 30,000 individuals a year who enter prison or jail nearly 90% of those receiving prescribed methadone are forced to stop or taper off this treatment (Fiscella et al., 2004. Reasons for this lack of expansion suggest that public opinion and that of the criminal justice system providers consider methadone treatment as substituting one addiction for another. In contrast, buprenorphine appears not to carry the same social stigma associated with methadone treatment and has been used in France, Austria and Puerto Rico (Catania, 2003, Garcia et al., 2007, Gordon et al., 2017, Reynaud-Maurupt et al., 2005. This is noted because unlike methadone it can be administered on alternate days, a feature that would make buprenorphine use more efficient in correctional settings than methadone (Garcia et al., 2007). Naltrexone treatment has shown some promising findings, but associated problems surrounding high attrition and low medication compliance in the community and high mortality rates pose concerns (Gibson andDegenhardt, 2007, Minozzi et al., 2011). In 2005 naloxone was added to the UK's exempt list of prescription-only medicines that could be administered to save life in an emergency to reverse heroin/opioid overdose (Strang et al., 2006). Use of extendedrelease naltrexone may also be appealing and beneficial to people who are unlikely to access opioid-agonist maintenance treatment or who prefer a relapse-prevention treatment . Trials conducted in the criminal justice setting are still lacking, and continuity of care is considered crucial in the treatment of drug-using offenders who transition between prison and the community.
A growing body of evidence shows the effects of pharmacological interventions for drug use among the general population (see Table 1 for a summary of the evidence). Existing reviews have focused on naltrexone maintenance treatment for opioid dependence (Amato et al., 2005, Lobmaier et al., 2008, Minozzi et al., 2011, the efficacy of methadone (Faggiano et al., 2003, Marsch, 1998, Mattick et al., 2009) and buprenorphine maintenance (Mattick et al., 2009). Recent guidance has been provided from the National Institute for Health and Clinical Excellence (NICE) on evidencebased use of naltrexone, methadone and buprenorphine for the management of opioid dependence (National Institute for Health and Care Excellence (NICE), 2007a, National Institute for Health and Care Excellence (NICE), 2007b, National Institute for Health and Care Excellence (NICE), 2019). Five Cochrane reviews (including 52 studies) reported on the effectiveness of opiate methadone therapies (Amato et al., 2005). Findings showed that methadone maintenance therapies at appropriate doses were most effective in retaining participants in treatment and in suppressing heroin use, but evidence of effectiveness for other relevant outcome measures such as criminal activity was weak and was not systematically evaluated. Naloxone use is also found to work in a similar manner by blocking the euphoria (reinforcement) sought by the user and thereby reducing the drug-seeking behaviour and diminishing the risk of physical dependence of heroin use . Extended-release naltrexone uses the same mechanism as naloxone but gradually releases a sufficient amount to block the euphoria up to 1 month after an injection . Systematic reviews evaluating treatment programmes for offender populations have focused on evaluating treatments in one setting such as community-based programmes (Mitchell et al., 2012a, Mitchell et al., 2012b, or have based their evidence on literature from one country (Chanhatasilpa et al., 2000, Egg et al., 2000, or a number of specific treatments (Mitchell et al., 2006). Pharmacological systematic reviews of offender treatment appear to be sparse. We identified two previous reviews: one focusing on specific drug-and property-related criminal behaviours in methadone maintenance treatment (Marsch, 1998), and one an evaluation of the effectiveness of opioid maintenance treatment (OMT) in prison and post release (Hedrich et al., 2012). The latter of these two reviews identified six experimental studies up until January 2011 (Hedrich et al., 2012). The authors found that OMT in prison was significantly associated with reduced heroin use, injecting and syringe sharing. The use of pre-release OMT was also found to have important implications for associated treatment uptake after release, but the impact on criminal activity was equivocal. Similar findings have been shown with the use of extended-release naltrexone (XR-NTX) which showed that use of XR-NTX before release might reduce opioid overdose, increase abstinence and antagonist medication coverage during the high-risk initial weeks after release in comparison with referral for XR-NTX treatment in the community after release (Friedmann et al., 2018).
Many people under the care of the criminal justice system have a drug misuse NIDA. https://www.drugabuse.gov/about-nida/legislative-activities /testimony-to-congress/2017/research-on-the-use-and-misuse-of -fentanyl-and-other-synthetic-opioids-

Buprenorphine
Buprenorphine is a partial agonist that activates opioid receptors but with a diminished response. Buprenorphine is effective in the maintenance treatment of heroin-dependent people, at any dose above 2 mg. It suppresses illicit opioid use (at doses 16 mg or greater). Compared with methadone, buprenorphine retains fewer people when doses are flexibly delivered and at low fixed doses. At fixed medium or higher doses, buprenorphine and methadone are equally effective in terms of retention in treatment and suppression of illicit opioid use. Buprenorphine may be preferred to methadone to avoid unwanted effects such as nausea, vomiting, constipation and drowsiness, mind "clouding" sore leg muscles. Buprenorphine has less overdose potential. Buprenorphine is also available in combination with naloxone (0.5-8 mg). The addition of naloxone reduces the abuse potential.

Naltrexone
Naltrexone is an antagonist which blocks the opioid receptor and reduces cravings and urges to use opioids. It does not produce euphoria and it is not addictive.
There is inadequate evidence to support its use alone as a maintenance treatment for opioid dependence. It may be effective as a therapy for highly motivated and well-supported people who prefer to follow an abstinence programme to stay off opioids entirely.  (Fazel et al., 2015). Whilst previous research has evaluated treatment programmes for offenders more broadly, we know little about the challenges, treatment and rehabilitation opportunities for offenders with pharmacological interventions aimed at reducing drug use and criminal activity. This includes consideration of when and where to offer treatment along the criminal justice pathway to maximise scarce resources and maximise the effectiveness of treatment. We therefore believe that a systematic review evaluating existing evidence from randomised controlled trials (RCTs) might be helpful in identifying treatments for reducing drug use and criminal activity in this vulnerable population. In principle, the RCT design eliminates the threat to internal validity providing there is a sufficiently large number of units assigned as the experimental and control conditions and methods to reduce bias are implemented (Weisburd, 2010). Given the importance of relating economic cost to a reduction in drug use and related offending behaviour, good-quality economic evidence will help inform strategies which represent the best use of limited resource (Higgins and Green, 2011). In this review, we use the Drummond checklist (Drummond et al., 2005) to evaluate and document the availability of resource information within the studies. The Drummond checklist is a widely used tool to assess the quality of economic evaluation studies. It assesses whether the study has a well-defined question, describes the competing programmes, provides information on the effectiveness of the programmes, presents all of the important and relevant costs and consequences for each programmes and assesses whether costs and consequences of the programmes were accurately measured and assigned credible values (and discounted appropriately). The checklist also assesses any incremental cost-effectiveness analysis and sensitivity analysis around the costs and consequences.
This systematic review has five primary research questions: (1) Do pharmacological treatments for drug-using offenders reduce drug use? (2) Do pharmacological treatments for drug-using offenders reduce criminal activity? (3) Does the treatment setting (e.g. court, community, prison/secure establishment) affect outcome(s) of pharmacological treatments? (4) Does one type of pharmacological treatment perform better than one other? (5) What are the comparative costs and cost-effectiveness of the available interventions?
The systematic review was guided by a protocol which followed Cochrane approaches (Higgins and Green, 2011) to producing systematic reviews and an earlier version has been published as a Cochrane review (Perry et al., 2015).

Methods
Studies included in the review had to meet a number of different criteria including the study design, the type of participants and type of interventions. Studies had to report pre-specified outcome measures to avoid the possibility of any subsequent bias in choosing outcomes. The eligibility criteria are summarised below and described in detail in Table 2.
Studies of illicit drug-using offenders were included in the review regardless of the gender, age, psychiatric history or ethnicity of the study participants. Eligible interventions were those designed, wholly or in part, to eliminate or prevent relapse to drug use Table 2 Summary of eligibility criteria   Inclusion criteria  Exclusion criteria Population • Illicit drug-using offenders were included in the review regardless of gender, age, psychiatric history, or ethnicity. • Offenders were defined as individuals who were involved in the Criminal Justice System (CJS). • Individuals could reside in special hospitals, prisons, the community or who were diverted from court or placed on arrest referral schemes for treatment. • The study setting could change throughout the process of the study. For example, people involved in the criminal justice system could begin in prison but progress through a work release project into a community setting. • Illicit drug use referred to individuals using drugs occasionally or those who were considered drug dependent.
Studies where part of the population were not involved in the CJS. Studies where the population were alcohol-using as opposed to drug-using offenders.
Intervention • Eligible interventions were those designed, wholly or in part, to eliminate or prevent relapse to drug use or criminal activity, or both, among participants. We defined relapse as individuals who may have returned to an incarcerated setting, or had subsequently been arrested or had relapsed back into drug misuse, or both. We included a range of different types of interventions in the review. • Experimental interventions included in the review were any pharmacological intervention (e.g. buprenorphine, methadone). • Control interventions eligible for the review were no treatment or waiting list controls or minimal and/or alternative treatment or treatment as usual: Minimal or alternative treatment might include reporting use of a similar intervention but at a less intense level, or using a different theoretical approach but with the same components and/or a different alternative intervention. Treatment as usual included any study that reported a combination and/or component of a (i) psychological based intervention (e.g. anger management, motivational interviewing, counselling, aggression replacement, family therapy), (ii) an educational programme (e.g., health, substance abuse education on risky behaviour) and or (iii) life skills (e.g., financial planning, employment skills, computer skills and interpersonal skills in interviews. Studies of alcohol misuse only were not eligible. Studies of interventions with drug-using offenders that were not focused on reducing drug use. or criminal activity, or both, among participants. We defined relapse as individuals who may have returned to an incarcerated setting, or had subsequently been arrested or had relapsed back into drug misuse, or both. The primary outcomes investigated in the review were drug use (self-reported drug use measures and biological drug use measures), self-reported or official report of criminal activity (dichotomous and continuous data for arrests, reincarceration and parole violations) and adverse effects. Secondary outcomes were resource use, costs and cost-effectiveness related to the interventions, health care activity, criminal activity and productivity. Randomised controlled trials (RCTs) were the eligible study design and no date or language limits were applied to this review.

Search strategy for identification studies
The original searches for this review are reported in Perry (Perry et al., 2006, Perry et al., 2015 and To identify economic studies the topic specific searches of nine databases (ASSIA, CENTRAL, CINAHL, Criminal Justice Abstracts, HMIC, LILACS, PAIS, Science Citation Index, Social Science Citation Index) would have identified all potential RCTs and economic evaluations included in the review. For the Embase, MEDLINE and PsycINFO databases the search strategy incorporated an RCT filter to manage the number of identified and relevant citations. This approach may have excluded potential economic evaluations identified from these three databases and is therefore a limitation of the review.
We developed individual search strategies for each database and made use of any controlled vocabulary available for each database. We combined the subject search

Screening and coding process
A team of reviewers independently inspected the titles and abstracts for potential inclusion in the review. Each record was assessed by two reviewers. The full articles were obtained for each record which seemed likely to be eligible and the full text was assessed for eligibility by two reviewers independently. In the case of disagreement, a third independent reviewer helped to arbitrate. The screening criteria are provided in the Appendix.
We used data extraction forms to standardise the reporting of data from all studies obtained as potentially relevant. Two reviewers independently extracted data and subsequently checked them for agreement. The narrative tables included a presentation of the study details (e.g. author, year of publication and country of study origin), study methods (e.g. random assignment), participants (e.g. number in sample, age, gender and ethnicity), interventions (e.g. description, duration, intensity and setting), outcomes (e.g. description, follow-up period and reporting mechanism) and notes (e.g. country and funding).

Measures of treatment effect
Studies were combined as mean differences (MD) with 95% confidence intervals in the meta-analyses for continuous outcomes measured on the same scale and standardised mean difference (SMD) for outcomes measured on different scales. Higher scores for continuous measures are representative of greater harm. We present dichotomous outcomes as risk ratios (RR), with 95% confidence intervals (CIs).

Statistical issues
To avoid double counting of outcome measures (e.g. arrest and parole violation) and follow-up time periods (e.g. 12, 18 months), we checked all trials to ensure that multiple studies reporting the same evaluation did not contribute towards multiple estimates of programme effectiveness. We followed Cochrane guidance and we combined intervention and control groups to create a single pair-wise comparison. Where this was not appropriate we selected one treatment arm and excluded the others. We attempted to contact the study authors via email where missing data occurred in the original publication.
We assessed heterogeneity using the I 2 statistic and Chi 2 statistic (Higgins and Green, 2011). We regarded heterogeneity as substantial if the I 2 was greater than 50% or the p value was lower than 0.10 for the Chi 2 test for heterogeneity (Deeks et al., 2017). Following specific Cochrane guidance (Deeks et al., 2017), we distinguished the following values to denote no important, moderate, substantial and considerable heterogeneity, respectively: 0 to 40%, 30 to 60%, 50 to 90% and 75 to 100%.

Risk of bias assessment
The review team assessed risk of bias in all included studies using risk of bias assessment criteria recommended in the Cochrane Handbook (Higgins and Green, 2011). Further detail on risk of bias assessment is provided in the Appendix.

Data synthesis
The Revman software package (The Nordic Cochrane Centre (The Cochrane Collaboration), 2014) was used to perform a series of meta-analyses for continuous and dichotomous outcome measures. A random effects model was used to account for participants coming from different underlying populations. We used the transformations as laid down by the Cochrane Handbook (Higgins and Green, 2011) for continuous outcomes. For conversions of standard error into standard deviations and the calculation of standard deviations calculated from 95% confidence intervals (CIs) we used the standard equations set out in Fig. 1.

Economic appraisal
Economic or resource use information was assessed using the Drummond classification scheme (see Appendix) (Drummond et al., 2005). This criterion was applied by an economist to indicate the costs and the consequences of the intervention(s) relevant to various public sectors. These included healthcare, criminality, labour force participation or other public goods. According to Drummond, studies containing information on the economics of the intervention can be classified as full economic evaluation studies or partial economic evaluation studies (Drummond et al., 2005). A full economic evaluation is a comparative analysis of two or more interventions in terms of both resource use (or costs) and outcomes (or consequences) (Drummond et al., 2005). Full economic evaluations can be further classified as cost-benefit analysis, cost-effectiveness analysis and cost-utility analysis based on consequences measured in monetary units, physical units (e.g. mortality) and utilities (e.g. quality-adjusted life years (QALYs)), respectively. Evaluations that are not comparative or do not consider both costs and consequences are classified as partial evaluations (e.g. cost-outcome description or cost analysis). Studies containing economic information using the Drummond criteria were identified, and information related to resource use and perspectives were recorded.

Search results
Searches for this review were first conducted in Perry (Perry et al., 2006, Perry et al., 2015 and were last updated in February 2019. The searches identified a total of 16,786 records and 988 additional records were identified from other sources. Following deduplication, 9657 records were screened using information in the title and abstract. Of these, 9406 records were excluded. We acquired the remaining 251 full papers for assessment and excluded 237 of these. A total of 22 studies were finally deemed eligible for the review and were included in the qualitative synthesis, with 19 of them included in the meta-analyses. The 22 included studies randomised 4372 participants from research published between 1969 and 2019. Fig. 2 shows the PRISMA flow diagram of the numbers of records included and excluded at each stage of the selection process.
Fourteen of the 22 trials were conducted in the US and three in the UK. The other 5 studies were conducted in Iran, Australia, Norway, Sweden and Mexico. Twenty-one of the 22 studies included male adult drug-using offenders and 1 study recruited only female participants . In 2 studies, the percentage of each gender was not reported .
The age of study participants ranged from 17 to 55 years. In 7/22 (32%) studies, the majority of participants were of white ethnic origin. The studies were divided by setting into community (n = 9) , prison (n = 7) (Bayanzadeh, 2004, Rich et al., 2015 and those that started in a secure setting and continued into the community (n = 6) .
Fifteen of the 22 trials evaluated outcomes at 6 months or less, 4 trials reported outcomes up to 12 months , 1 study reported outcomes at 18 months  and 1 reported outcomes at 4 years .
Different outcome measures were presented for each study, and just over half of all studies reported 4 or more outcome measures. One study reported on criminal activity only  and 8 trials reported on drug use only (Bayanzadeh, 2004. Six studies did not report any information on adverse events, side effects from the drug or drug safety (Bayanzadeh, 2004. Three studies referred to the measurement of adverse events but did not report them . 4 studies reported resource and cost information , Murphy et al., 2017, Rich et al., 2015, Warren et al., 2006.

Meta-analyses
Nineteen of the 22 studies provided data to potentially be included in 1 or more metaanalyses. The interventions report on agonistic pharmacological interventions (buprenorphine, methadone) and antagonistic pharmacological interventions (naltrexone and naloxone) compared with no intervention, other non-pharmacological treatments (e.g. counselling) and other pharmacological drugs (vigabatrin and cyclazocine). The comparisons are shown in Table 3 and in Fig. 3.

Agonist (vigabatrin) versus placebo
One study (103 participants) showed no evidence of a difference in terms of the number of negative urine tests for cocaine (MD 2.1, 95% CI − 0.34 to 4.54) and end of trial abstinence for marijuana (RR 0.64, 95% CI 0.16 to 2.52) but showed evidence of a difference in the end of trial abstinence (RR 1.06, 95% CI 0.28 to 4.01) and 4-week abstinence for cocaine (RR 6.36, 95% CI 1.50 to 27.01) (Bayanzadeh, 2004.

Antagonist (naloxone) versus no intervention (placebo or waiting list control)
Three studies (333 participants) showed no significant reduction in self-reported drug use (RR 1.06, 95% CI 0.87 to 1.28) .

Agonist (methadone) versus forced tapered withdrawal from methadone
One study (283 participants) explored continued use of methadone within a prison setting compared with forced tapered withdrawal of methadone (Rich et al., 2015). The study found no significant difference in reduction in any drug use at 1 month following release from prison (RR 1.05, 95% CI 0.83 to 1.34), opioid use at 1 month following release (RR 0.56, 95% CI 0.26 to 1.22) or use of injectable illegal drugs (RR 0.67, 95% CI 0.40 to 1.15) (Rich et al., 2015). The study also found no difference between the two options in terms of reincarceration at 1 month post release from prison (RR 1.49, 95% CI 0.63 to 3.53) (Rich et al., 2015).
Agonist (methadone) prior to release with financial support versus agonist (methadone) on release with no financial support One study (46 participants

Narrative findings
Narrative findings were presented for three studies where suitable data for extraction and exclusion in the meta-analysis were not available. One study compared methylphenidate or placebo to a non-pharmacological intervention . In this study, male participants with attention deficit disorder receiving methylphenidate had a higher proportion of drug negative urine test results compared with the placebo group (p = 0.047). A second study compared a specialist facility for treatment of methadone versus primary care plus suboxone (buprenorphine and naloxone) versus a specialist facility for suboxone treatment . At 6 months, the authors reported no use of opioids in any of the randomised groups (total of 15 participants). The final study compared methadone and placebo versus lofexidine and placebo . The authors reported no statistical reduction in the amount (p = 0.36) or intensity (p = 0.46) of opioid (heroin) use in the last month.

Adverse events
Seven of the 22 (32%) studies did not report any information on adverse events, side effects of drugs nor referred to drug safety (Bayanzadeh, 2004. Three (3/22, 14%) studies measured adverse events, but did not report the findings . Forty-one percent of the studies (9/22) reported only mild or moderate adverse events . One study reported 10 serious adverse events including nine hospitalisations . The hospitalisations included two cases for heart disease and one each for pneumonia, alcohol detoxification, kidney disease, high blood pressure, psychiatric problems and back pain . Two studies reported a fatal overdose , Rich et al., 2015 and one study reported two fatal overdoses .

Economic outcomes
The economic appraisal using the Drummond criteria identified four studies reporting some cost information , Murphy et al., 2017, Rich et al., 2015, Warren et al., 2006. Table 4 shows the results of the type of resource costs described for each paper. Three of the four studies were full economic evaluations reporting both incremental costs and outcomes , Murphy et al., 2017, Warren et al., 2006. In terms of the overall Drummond classification (Drummond et al., 2005), three studies were coded as 4 (a full economic evaluation) and one study was coded as 3 (partial economic evaluation).
The ) study noted differences in the costs of administering buprenorphine and methadone, but were not sufficient for us to conduct a full cost-effectiveness appraisal. The investigators estimated that about ten times as many inmates can be served with methadone as with buprenorphine with the same staff resources. This cost implication is also endorsed in the community, where physicians have difficulty in obtaining reimbursement for buprenorphine treatment for released inmates, making the continued use of buprenorphine problematic after release.
The Murphy 2017(Murphy et al., 2017 study conducted a full economic analysis reporting incremental cost-effectiveness ratios (ICERs), costs, QALYs and abstinent years of using extended-release naltrexone (XR-NTX) compared with treatment as usual (TAU) at 25 weeks and 78 weeks, respectively. The 25-week ICERs were reported as $162,150/QALY and $46,329/ abstinent year. The ICER values at 78 weeks were $76,400/QALY and $16,371/abstinent year. The study reported that XR-NTX was effective in increasing both QALYs and abstinence rates compared with TAU. However, it was not cost-effective at the normally accepted threshold levels in the US, primarily due to the high price of the injection.
The Warren 2006(Warren et al., 2006 study estimated the cost-effectiveness of the New South Wales (NSW) prison methadone programme compared with no methadone program. The study reported the annual cost of providing prison methadone in NSW as AUD$2.9million (or AUD$3234 per inmate per year) and the ICER as AUD $38 per additional heroin-free day. It also reported that prison methadone is not costlier than community methadone, and it reduces morbidity and mortality through decreasing heroin use in prisons. The final study by Rich 2015(Rich et al., 2015 reported, as a secondary outcome, on healthcare costs (per inmate) associated with continued methadone maintenancetreatment programme compared with usual care (i.e. forced withdrawal from methadone) at 1 month after release from incarceration. Continued methadone treatment resulted in higher (methadone) treatment cost per individual compared with forced withdrawal ($362 vs $225, p < 0.001). The additional treatment costs were offset by savings incurred by reduced cost for physician ($6.60 vs $8.80, p = 0.793) and medical care ($211 vs $372, p = 0.894) after release, resulting in lower total costs ($609 vs $637, p < 0.001). Continued methadone treatment also resulted in a greater probability of attendance at a methadone clinic after release and was more effective than forced withdrawal.

Risk of bias assessment
Although all the included studies were randomised trials, the level of reporting was uneven and many had unclear or high risk of bias for at least one element of the risk of bias assessment. Nearly 50% were rated as unclear risk of bias in terms of random sequence generation and over 50% of the trials were rated as unclear in terms of allocation concealment. 32% (7/22) of the studies were rated as low risk of bias for participant and personnel blinding and 82% (18/22) were rated as unclear on both subjective and objective measures of detection bias. Over 75% of the trials were at high risk of bias because they did not report all of the outcome data they set out to collect.   Over 50% of the studies were rated at low risk for selective reporting. The overall risk of bias judgements of the included studies are shown in Fig. 4, and summary and detailed accounts of the risk of bias assessment can be found in the Appendix.

Conclusions
The results of this systematic review provide some evidence to answer our original review questions. The review identified that we have some evidence to suggest that pharmacological interventions are effective in reducing drug use and criminal activity (Perry et al., 2015). The updated review revealed a total of 22 trials involving 4372 participants. The 22 trials compared pharmacological interventions or compared pharmacological interventions to no intervention and/or a placebo. Most studies were conducted in the USA, with the majority involving male adult offenders. The updated meta-analyses included 19/22 quantitative studies. Ten different pharmacological drugs were evaluated including methadone, buprenorphine, naltrexone, methylphenidate, vigabatrin, suboxone, naloxone, lofexifine, naltrexone and cyclazocine. Methadone in comparison with no treatment was the only comparison to demonstrate a significant reduction in subsequent drug use (using biological measures), self-reported drug use (using continuous measures) and criminal activity. When compared with methadone, no significant differences were found for buprenorphine, naltrexone or suboxone. None of these studies identified a significant reduction in either criminal activity and/or drug use. One study comparing buprenorphine to placebo  was the only study to involve wholly female offenders. More pharmacological trials of female offenders are required to identify comparability between men and women. Studies of naloxone and methylphenidate in comparison with a placebo and nonpharmacological intervention, respectively, showed no significant differences for selfreported drug use. The evidence supporting the use of naloxone was explored using two very old studies  and one more recent study that had to be halted early due to problems with the randomisation procedure . Naltrexone in comparison with a non-pharmacological intervention showed a significant reduction in self-reported drug use and reincarceration. The use of extended naltrexone pre and post prison release has produced some promising results, finding no overdoses during the treatment process . This study also reported that use of extended-release naltrexone was associated with a rate of opioid relapse that was lower than that with usual treatment at 24 weeks, but later comparisons (at 52 and 78 weeks) found no difference between the use of extendedrelease naltrexone and use of a brief counselling intervention . One study investigated methadone continuation treatment compared with forced tapered withdrawal from methadone within a prison setting, but found no significant difference in the reduction of any drug use at 1 month following release (Rich et al., 2015). The study also found no difference between the treatment options where reincarceration occurred at 1 month post release from prison. The impact of providing financial support for methadone treatment following release from prison was compared with no financial support in one study, but no significant difference was observed in terms of drug use or criminal justice outcomes . The study also reported no significant difference between the interventions in relation to arrest or reincarceration at six-month follow-up. We were unable to answer our third review questions about the impact of treatment setting on drug use or criminal activity. This was primarily because of the heterogeneity of the study comparisons and outcome measures. Given the relatively scarce resources for drug treatment evaluation opportunities in this population, future research should seek to build on existing trial data to develop an evidencebase to allow for such comparisons to be made.
The majority of studies included in the review reported only mild to moderate adverse effects, or did not report on drug safety as part of the trial information. One study reported serious adverse effects and three studies reported overdoses. Adverse effects should be measured consistently across any pharmacological drug trial to capture important data about drug safety. The current evidence does not provide a comprehensive overview of drug safety within this population.
Our final review question focused on the comparative costs and cost-effectiveness evidence of using such interventions to reduce drug use and criminal activity. An appraisal of the economic evidence identified studies with costs and outcomes data for pharmacological treatments among drug-using offenders in three studies and costs data in one study. All studies reported data on the direct medical costs associated with the provision of treatment , Murphy et al., 2017, Rich et al., 2015, Warren et al., 2006 and one study also included costs related to criminal justice (Murphy et al., 2017). The findings from the studies revealed the higher effectiveness of pharmacological treatments in terms of increased QALYs and abstinent years (Murphy et al., 2017) and reduced heroin use (Warren et al., 2006). One study reported methadone to be cost-effective at the current willingness to pay thresholds (Warren et al., 2006). Another study reported continued methadone treatment to dominate forced withdrawal in terms of costs and effects (Rich et al., 2015). More research is needed to compare the relative cost-effectiveness of a range of treatments (pharmacotherapies, non-pharmacological interventions and combined treatments) in prison and community-based criminal justice settings.
Additionally, the studies contained mainly male offenders, with only one study containing a 100% sample of female offenders. The majority of studies were conducted in the USA. Together these factors limit the external validity and generalisability of the study findings. For these reasons, the studies are unlikely to reflect different international practice in the criminal justice system. Besides the limitations already discussed, the search methodology was limited to databases that could be accessed via the University of York and extensive website searches were not conducted. As a result, some literature may have been missed from this updated version. Specific search terms were not used to identify cost and cost-effectiveness studies in 3/12 databases which meant that some economic studies might have been missed from this review.
Several research implications can be identified from this review. Generally, better quality research is required to evaluate the effectiveness of interventions with extended long-term effects of aftercare following release into the community. Buprenorphine research in the prison environment requires evidence of the long-term impact and larger studies; currently, an equivalence of buprenorphine and methadone exists. Future clinical trials should collect information from all sectors of the criminal justice system. This would enhance the heterogeneous nature of the included studies and would facilitate generalisation of study findings. Evidence of comparable mortality rates in prisoners using pharmacological interventions (particularly after release) needs to be explored to assess the long-term outcomes of such treatments, perhaps using alternative study designs to randomised controlled trials. The link between dosage, treatment retention and subsequent criminal activity should be examined across all three pharmacological treatment options. Evidence from other trial data suggests that dose has important implications for retention in treatment; in future studies, this should be considered alongside criminal activity outcomes. Cost and cost-effectiveness information should be standardised within trial evaluations; this will help policymakers to decide upon health versus criminal justice costs.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.

Search Strategy
The searches for this updated review were conducted in February 2019 in the following databases: ASSIA, CENTRAL, CINAHL, Criminal Justice Abstracts, Embase, HMIC, LI-LACS, MEDLINE, PAIS, PsycINFO, Science Citation Index, Social Science Citation Index.
Initially, there were 4384 records identified. After deduplication, there were 2603 records remaining. After further deduplication against the results of the January 2018 update search, 2445 records remained.

ASSIA search strategy
ASSIA via ProQuest search strategy (search date = 6th February 2019, 247 records identified) (ti(substance* NEAR/2 (addict* OR depend* OR disorder* OR abuse* OR abusing OR misuse* OR misusing OR consumption* OR withdraw* OR withdraw* OR detox*)) OR ab(substance* NEAR/2 (addict* OR depend* OR disorder* OR abuse* OR abusing OR misuse* OR misusing OR consumption* OR withdraw* OR withdraw* OR detox*)) OR ti(drug* NEAR/2 (addict* OR depend* OR disorder* OR abuse* OR abusing OR misuse* OR misusing OR consumption* OR withdraw* OR withdraw* OR detox*)) OR ab(drug* NEAR/2 (addict* OR depend* OR disorder* OR abuse* OR abusing OR misuse* OR misusing OR consumption* OR withdraw* OR withdraw* OR detox*)) OR ti(narcotic* NEAR/2 (addict* OR depend* OR disorder* OR abuse* OR abusing OR misuse* OR misusing OR consumption* OR withdraw* OR withdraw* OR detox*)) OR ab(narcotic* NEAR/2 (addict* OR depend* OR disorder* OR abuse* OR abusing OR misuse* OR misusing OR consumption* OR withdraw* OR withdraw* OR detox*)) OR ti(mdma OR alcohol* OR opiate* OR opioid* OR opium OR heroin OR methadone OR cocaine OR amphetamine* OR marijuana OR cannabis OR crack OR phencyclidine) OR ab(mdma OR alcohol* OR opiate* OR opioid* OR opium OR heroin OR methadone OR cocaine OR amphetamine* OR marijuana OR cannabis OR crack OR phencyclidine)) AND (ti((justice system) OR remand* OR parole* OR probation OR court* OR corrections OR correctional OR revocation) OR ab((justice system) OR remand* OR parole* OR probation OR court* OR corrections OR correctional OR revocation) OR ti(crime OR offend* OR criminal OR convict* OR felon*) OR ab(crime OR offend* OR criminal* OR convict* OR felon*) OR ti(custody OR custodial OR gaol* OR jail* OR prison* OR incarcerat* OR inmate*) OR ab(custody OR custodial OR gaol* OR jail* OR prison* OR incarcerat*or inmate*) OR ti(reoffend* OR reincarcerat* OR recidiv* OR ex-offender*) OR ab(reoffend* OR reincarcerat* OR recidiv* OR ex-offender*)) AND pd(20180101-20190201) [Narcotics] explode all trees #5(substance* or drug* or narcotic*) near/2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw$ or withdraw* or detox*):ti,ab,kw (Word variations have been searched) #6mdma or alcohol* or opiate* or opioid* or opium or heroin or methadone or cocaine or amphetamine* or marijuana or cannabis or crack or phencyclidine:ti,ab,kw (Word variations have been searched) #7#1 or #2 or #3 or #4 or #5 or #6 #8MeSH descriptor: [Crime] explode all trees #9MeSH descriptor: [Criminals] explode all trees #10MeSH descriptor: [Prisoners] explode all trees #11(justice system) or remand* or parole* or probation or court* or corrections or correctional or revocation:ti,ab,kw (Word variations have been searched) #12custody or custodial or gaol* or jail* or prison* or incarcerat* or inmate*:ti,ab,kw (Word variations have been searched) #13reoffend* or reincarcerat* or recidiv* or ex-offender*:ti,ab,kw (Word variations have been searched) #14offend* or criminal* or convict* or felon:ti,ab,kw (Word variations have been searched) #15#8 or #9 or #10 or #11 or #12 or #13 or #14 #16#7 and #15

CINHAL Plus search strategy
CINHAL Plus via EBSCO search strategy (search date = 6th February 2019, 774 records identified) S1 TI (substance* N2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) OR AB (substance* N2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) OR TI (drug* N2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) OR AB (drug* N2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) OR TI (narcotic* N2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) OR AB (narcotic* N2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) S2 TI (mdma or alcohol* or opiate* or opioid* or opium or heroin or methadone or cocaine or amphetamine* or marijuana or cannabis or crack or phencyclidine) OR AB (mdma or alcohol* or opiate* or opioid* or opium or heroin or methadone or cocaine or amphetamine* or marijuana or cannabis or crack or phencyclidine) S3 S1 OR S2 S4 TI (justice system) or crime or remand* or parole* or probation or court* or corrections or correctional or revocation) OR AB (justice system) or crime or remand* or parole* or probation or court* or corrections or correctional or revocation) OR TI (offend* or criminal* or convict* or felon*) OR AB (offend* or criminal* or convict* or felon*) OR TI (custody or custodial or gaol* or jail* or prison* or incarcerat* or inmate*) OR AB (custody or custodial or gaol* or jail* or prison* or incarcerat* or inmate*) OR TI (reoffend* or reincarcerat* or recidiv* or ex-offender*) OR AB (reoffend* or reincarcerat* or recidiv* or ex-offender*) S5 S3 AND S4

Social Science Citation Index via Web of Science search strategy
Social Science Citation Index via Web of Science search strategy (search date = 6th February 2019, 757 records, Indexes = SSCI Timespan = 1900-2019) # 1TOPIC: (substance* NEAR/2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*) ) OR TOPIC: (drug* NEAR/2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*) ) OR TOPIC: (narcotic* NEAR/2 (addict* or depend* or disorder* or abuse* or abusing or misuse* or misusing or consumption* or withdraw* or withdraw* or detox*)) # 2TOPIC: (mdma or alcohol* or opiate* or opioid* or opium or heroin or methadone or cocaine or amphetamine* or marijuana or cannabis or crack or phencyclidine) # 3#2 OR #1 # 4TOPIC: ("justice system" or remand* or parole* or probation or court* or corrections or correctional or revocation) OR TOPIC: (crime or criminal or offender* or criminal* or convict* or felon*) OR TOPIC: (custody or custodial or gaol* or jail* or prison* or incarcerat* or inmate*) OR TOPIC: (reoffend* or reincarcerat* or recidiv* or ex-offender*) #5#4 AND #3

Screening criteria
The screening process was divided into two key phases. Phase one used seven key questions reported in the original new reference review: Prescreening criteria: phase one Is the document an empirical study? [If "no" exclude document.] Does the study evaluate an intervention, a component of which is designed to reduce, eliminate or prevent relapse among drug-using offenders?
Are the participants referred by the criminal justice system at baseline?
Does the study report pre-programme and post-programme measures of drug use?
Does the study report pre-programme and post-programme measures of criminal behaviour?
Is the study a randomised controlled trial?
Do the outcome measures refer to the same length of follow-up for two groups?
Papers included after phase one screening were then scrutinised to assess: Prescreening: phase two

Is the intervention a pharmacological intervention? [if "yes" include document]
3. Drummond (1996)   11. Were the primary outcome measure(s) for the economic evaluation clearly stated?
12. Were the methods used to value health states and other benefits stated?
13. Were the details of the subjects from whom valuations were obtained given?
14. Were productivity changes (if included) reported separately?
15. Was the relevance of productivity changes to the study question discussed?
16. Were quantities of resources reported separately from their unit cost?
17. Were the methods for the estimation of quantities and unit costs described?
18. Were currency and price data recorded?

High risk
The investigators describe a non-random component in the sequence generation process such as odd or even date of birth; date (or day) of admission; hospital or clinic record number; alternation; judgement of the clinician; results of a laboratory test or a series of tests; availability of the intervention.

Unclear risk
Insufficient information about the sequence generation process to permit judgement of low or high risk.

Allocation concealment (selection bias) Low risk
Investigators enrolling participants could not foresee assignment because one of the following, or an equivalent method, was used to conceal allocation: central allocation (including telephone, web-based, and pharmacy-controlled randomisation); sequentially numbered drug containers of identical appearance; sequentially numbered, opaque, sealed envelopes.
High risk Investigators enrolling participants could possibly foresee assignments because one of the following method was used: open random allocation schedule (e.g. a list of random numbers); assignment envelopes without appropriate safeguards (e.g. if envelopes were unsealed or non-opaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.

Unclear risk
Insufficient information to permit judgement of low or high risk. This is usually the case if the method of concealment is not described or not described in sufficient detail to allow a definite judgement.
3. Blinding of participants and providers (performance bias) Objective outcomes Low risk No blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding; blinding of participants and key study personnel ensured, and unlikely that the blinding could have been broken.

High risk
No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding.
Unclear risk Insufficient information to permit judgement of low or high risk; the study did not address this outcome.
4. Blinding of participants and providers (performance bias)-subjective outcomes Low risk Blinding of participants and providers and unlikely that the blinding could have been broken; High risk No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding.

Unclear risk
Insufficient information to permit judgement of low or high risk; the study did not address this outcome.

Blinding of outcome assessor (detection bias)-objective outcomes
Low risk No blinding of outcome assessment, but the review authors judge that the outcome measurement is not likely to be influenced by lack of blinding; blinding of outcome assessment ensured, and unlikely that the blinding could have been broken.

High risk
No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding.
Unclear risk Insufficient information to permit judgement of low or high risk; the study did not address this outcome.
6.Blinding of outcome assessor (detection bias)-subjective outcomes Low risk No blinding of outcome assessment, but the review authors judge that the outcome measurement is not likely to be influenced by lack of blinding; blinding of outcome assessment ensured, and unlikely that the blinding could have been broken.
High risk No blinding of outcome assessment, and the outcome measurement is likely to be influenced by lack of blinding; blinding of outcome assessment, but likely that the blinding could have been broken, and the outcome measurement is likely to be influenced by lack of blinding.

Unclear risk
Insufficient information to permit judgement of low or high risk.
Low risk No missing outcome data; reasons for missing outcome data unlikely to be related to true 7. Incomplete outcome data (attrition bias)for all outcomes except retention in treatment or drop out outcome (for survival data, censoring unlikely to be introducing bias); missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size; missing data have been imputed using appropriate methods; all randomised patients are reported/analysed in the group they were allocated to by randomisation irrespective of non-compliance and co-interventions (intention to treat) High risk Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; for dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate; for continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size; "As-treated" analysis done with substantial departure of the intervention received from that assigned at randomisation.

Unclear risk
Insufficient information to permit judgement of low or high risk (e.g. number randomised not stated, no reasons for missing data provided; number of drop out not reported for each group).

Selective reporting (reporting bias) Low risk
The study protocol is available and all of the study's pre-specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre-specified way; the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre-specified (convincing text of this nature may be uncommon).
High risk Not all of the study's pre-specified primary outcomes have been reported; one or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. sub scales) that were not pre-specified.

Unclear risk
One or more reported primary outcomes were not pre-specified (unless clear justification for their reporting is provided, such as an unexpected adverse effect); one or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta-analysis; the study report fails to include results for a key outcome that would be expected to have been reported for such a study.
9. Other bias Low risk Evidence to suggest other problems identified with the study which might threaten the validity of the random allocation, attrition or data integrity and results of the trial.
High risk Evidence to suggest that the trial might be underpowered/problems with the random allocation process leading to potential self-selection bias/ issues of analysis not conducted using intent to treat analysis or evidence of missing data; concerns of attrition and measurement error including reliance on self-reported measures.

Unclear risk
Insufficient information to permit judgement of low or high risk.

Risk of bias domain summaries
Random sequence Although all the included studies were randomised trials, nearly 50% (10/22) provided an inadequate description and were rated as unclear by the reviewers . Eleven studies were reported at low risk of bias (Bayanzadeh, 2004, Rich et al., 2015 and the remaining study was at high risk of bias .

Allocation concealment
Of the 22 included studies, five reported that the allocation process was concealed and were rated at low risk of bias . Five studies were rated at high risk of bias (Bayanzadeh, 2004, Rich et al., 2015. The remaining 12 studies were all rated as unclear  and the review authors were unable to decide whether allocation concealment had occurred within the studies.

Blinding
Blinding was assessed across four dimensions considering performance and detection bias across subjective and objective measures. Seven studies were rated as unclear risk of bias and provided no information on blinding across all four domains (Bayanzadeh, 2004  Low risk Group allocation was based on block randomisation. A sequential list of case numbers was matched to group allocations in blocks of ten by randomly drawing five cards labelled "control" and five cards labelled "treatment" from an envelope. This procedure was repeated for each block of ten sequential case numbers. The list of case numbers and group allocation was held by a researcher not involved in recruiting or interviewing inmates. The trial nurses responsible for assessing, recruiting and interviewing inmates had no access to these lists. High risk There was substantial cross over from arms 1 and 3 to 2. The authors suggested that the treatment completer analysis is more reliable than the ITT analysis. Data for the completers only are presented. Selective reporting (reporting bias) Low risk Protocol and trial is registered see NCT00142935.
All outcomes are reported as expected. Blinding of participants and personnel (performance bias) subjective outcomes

Low risk
The trial was double-blind prior to release so that, whilst the participant was still in custody and pre-release, neither the participant, prison-based N-ALIVE staff nor prison staff knew the allocation. Participants learned their allocation when they opened the pack at the time of their release.
Blinding of participants and personnel (performance bias) objective outcomes

Low risk
The trial was double-blind prior to release so that, whilst the participant was still in custody and pre-release, neither the participant, prison-based N-ALIVE staff nor prison staff knew the allocation. Participants learned their allocation when they opened the pack at the time of their release.
Blinding of outcome assessment (detection bias) subjective outcomes

High risk
The self-reported questionnaire identified the participant's randomised assignment and the time interval between the preceding N-ALIVE release date and completion date.
Blinding of outcome assessment (detection bias) objective outcomes Unclear risk Not applicable-no objective measures reported in the trial.
Incomplete outcome data (attrition bias) all outcomes High risk Approximately 20% attrition in each group-due to those allocated did not get their allocation as intended.
Selective reporting (reporting bias) High risk Study protocol is reported but not all of the study's pre-specified primary outcomes have been reported due to the trial stopping early. Protocol (ISRCTN34044390) Blinding of outcome assessment (detection bias) subjective outcomes

Unclear risk
No statement regarding blinding of individual who undertook the biochemical urine tests.
Blinding of outcome assessment (detection bias) objective outcomes

Unclear risk
No statement regarding blinding of individual who recorded self-reported or clinical notes.
Incomplete outcome data (attrition bias) all outcomes High risk High levels of attrition. 50% buprenorphine and 45% methadone did not provide urine sample at day 8. 65% and 62% at 1 month, 80% and 85% at 3 months and 86% and 91% at 6 months.
Selective reporting (reporting bias) Low risk Published study protocol identified. All primary and secondary outcomes reported.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.