Residential rehabilitation: literature review

4. Effectiveness of Residential Rehabilitation

4.1 Measuring outcomes

Measuring outcomes has formed the focus of much of the literature on residential rehabilitation. However, it must be noted that measuring the effectiveness of residential rehabilitation in producing specific outcomes is complicated by a number of factors (Godfrey et al. 2004). First, as noted, there are substantial variations in residential rehabilitation provision, including a variety of models, treatment philosophies, programme lengths, and quality. Second, definitions of 'positive outcomes' are similarly diverse, and outcome measurement tools are many and heterogeneous. Positive outcomes in relation to treatment are often primarily defined in terms of individual and public health benefit, in relation to abstinence, reduced drug use, reduced criminal behaviour and improved physical/mental health, but rarely centre on the individual's personal goals. A recent systematic review found that Quality of Life (QOL) indicators have been inconsistently applied in studies of residential rehabilitation, and that QOL is often taken as subjective (Fischer et al. 2019). The review also found that out of 12 different instruments used to assess QOL, only two enabled individuals to identify QOL dimensions important to themselves. Third, with placement in residential rehabilitation typically forming one of multiple episodes of treatment for each individual, it is challenging to delineate which of these multiple treatments has produced specific outcomes. It should also be noted that clinical trials and other study formats investigating substance use issues often exclude certain demographics of people, such as people with severe and complex substance use or comorbid mental health issues, which is likely to affect research findings and their translation into practice (Day et al. 2005).

4.2 Overview of existing evidence of effectiveness of residential rehabilitation

Despite these challenges in measuring efficacy, a relatively robust body of evidence – including systematic reviews, evidence reviews, observational studies, treatment outcome studies and randomised control trials – suggests that residential rehabilitation can be effective in improving a variety of health-related and wider social outcomes. Four systematic reviews relating to residential rehabilitation treatment outcomes were identified as part of this literature review. All four found moderate quality evidence that residential rehabilitation improves outcomes across various domains, including substance use, mental health, social outcomes (including criminal justice involvement) and mortality (Drake et al. 2008; Cleary et al. 2009; De Andrade et al. 2019; AshaRani 2020). An evidence review by the Helena Kennedy Centre for International Justice (HKCIJ 2017) comparing international residential rehabilitation models similarly found that residential treatment broadly increases abstinence success, reduces criminal activity and improves wellbeing. This section explores this evidence base for various outcomes, including substance use, mental health, social (including criminal justice) outcomes, and mortality.

4.2.1 Substance use outcomes

Substance use outcomes have, understandably, formed the predominant focus of the body of research that has explored the effectiveness of residential rehabilitation. Broadly, there is relatively robust evidence that residential rehabilitation is associated with improved substance use outcomes. Previous systematic reviews have identified a range of studies within the UK and international literature which have reported substance use outcomes following residential rehabilitation. De Andrade et al. (2019) found seventeen studies published between 2013 and 2018 which had reported substance use outcomes. Three of these studies explored substance use outcomes for programmes which also deal with mental illness (Bergman et al. 2014; Schoenthaler et al. 2017; McGuire et al. 2018). The study by Bergman et al. (2014) of a cohort of young adults with substance use and mental health comorbidity found significant improvement in substance use for both groups, with no statistically significant differences between the two groups. At 3-month follow-up (81% retention), the percentage of days abstinent was 94% and 92% respectively, and 78% and 86% at 12-months (71% retention). Schoenthaler et al. (2017) found abstinence rates at 6 and 12-months of 68% and 65%, respectively, among their cohort of adults with comorbidity. McGuire et al. (2018) found 100% abstinence on discharge among their cohort of veterans. Another of these 17 studies reported high abstinence rates at 12-months (80%) following residence in a New Zealand therapeutic community, but much lower follow-up rate of 47% (King et al, 2016). These findings were weakened by the fact that 82% of those followed-up at 12 months had completed the full treatment (45% of the total sample). Šefránek and Miovský's (2017; 2018) cohort studies of therapeutic communities in the Czech Republic found abstinence rates of 88% at 12-month follow-up, with a 78% participant retention rate. A US study of a youth residential treatment centre found abstinence rates to be far lower at six months, with 43%, 42% and 41% for opioid dependence, non-opioid use, and opioid dependence, respectively (Schuman-Olivier et al. 2014). These rates were 29%, 32% and 22% on 12-month follow-up (71% retention).

This systematic review (de Andrade et al. 2019) also highlighted two high quality randomised control trials (RCT) which found improvements in both treatment (residential rehabilitation plus adjunctive therapy) and control groups (standard residential rehabilitation), although these improvements were not sustained over follow-up (Daughters et al. 2018; Davis et al. 2018). Compared to baseline, two cohort studies in South Africa (Myers et al. 2018) and New Zealand (Patterson et al. 2018) found statistically significant improvements in abstinence at the end of treatment and three-month follow-up and at discharge, respectively. A Brazilian study judged to be of weak quality (do Carmo et al. 2018), tested urine samples of the 69 study participants twice weekly at discharge, with 43 out of 1283 positive tests. This study reported relapse rates which were much higher than studies using self-reporting, with 49% relapsing at least once during their placement.

AshaRani's (2020) systematic review explored substance use outcomes for residential rehabilitation in the context of stimulants. Four international studies met the criteria, each of which found it to be effective in reducing methamphetamine use, craving and other related outcomes (McKetin et al. 2012, 2013, 2018; Kamp et al. 2019). In comparison with detoxification, residential rehabilitation demonstrated significantly reduced methamphetamine use, an effect that was time sensitive (McKetin et al. 2012), alongside an improvement in abstinence rates. At three months, 33% were continuously abstinent, dropping to 14% and 6% for one and three years, respectively. Another study demonstrated positive outcomes in relation to methamphetamine use, with a significant reduction in past month methamphetamine use at three and six months compared to baseline (McKetin et al. 2013). A further study by the same authors (2018) demonstrated 23% abstinence following residential rehabilitation.

One of the most significant trials undertaken in the UK, the National Treatment Outcome Research Study (NTORS) (Gossop et al. 2000; Gossop et al. 2003), interviewed 418 people from 54 agencies practicing four treatment modalities at intake to treatment and one year, two years and 4-5 years after leaving treatment. Residential rehabilitation was shown to result in substantial reductions in use of heroin, non-prescribed methadone, powder cocaine, benzodiazepines and amphetamines, with heroin use being halved by patients in both the residential and methadone programme cohorts. Positive outcomes were associated with longer time in treatment and completion rates. Age, sex, having friends or partners who used drugs; psychological/physical health; previous treatment contact; histories of using different substances were all found to have no statistically significant outcome on achieving positive outcomes; however, those who experienced lapses and relapses at follow-up were more likely to have used crack cocaine prior to treatment. Those who were undergoing criminal justice procedures at intake were similarly likely to have poorer outcomes. The NTORS study found that various treatments had less longitudinal impact on alcohol and crack cocaine use, noting that future treatment practice should be better adapted to meet the needs of people who use those specific substances.

4.2.2 Mental health outcomes

Numerous studies have explored both the effectiveness of residential rehabilitation in improving mental health outcomes and the effectiveness of an integrated approach to substance use and mental health treatment. The systematic review conducted by de Andrade et al. (2019) identified 17 international studies which reported on mental health outcomes associated with residential rehabilitation. Outcomes included 'psychological distress, post-traumatic stress disorder (PTSD), depression, anxiety, stress and general mental health' (de Andrade et al. 2019: 232). Four studies, including two of weak and two of moderate quality, found statistically significant improvement in Addiction Severity Index (ASI-X or ASI-MV) composite scores for mental health at follow-ups (Deane et al. 2013; Turner and Dean 2016; Schoenthaler et al. 2017; Patterson et al. 2018). A strong quality study by Teesson et al. (2017) observed 'significant' improvements in mental health outcomes in the month prior to follow-ups over 11 years. For those who completed the programme, reductions in depression, anxiety and stress at discharge were similarly found in another strong quality Australian study, based in a therapeutic community with a median stay of 16 weeks. A cohort analytic study by Roos et al. (2018), moreover found that two or more sessions of mindfulness-based relapse prevention within residential settings lead to substantial improvements across a range of mental health outcomes.

De Andrade et al. (2019) also sourced five studies (among the seventeen in total) that examined the effects of integrated treatment for mental health and substance use. The first (Bergman et al. 2014) found that improvements in psychiatric symptoms for people with co-occurring mental and substance use disorder were largely maintained at follow-up, with high retention across follow-ups at three months (81%), six months (73%) and twelve months (71%). The second reported reduced rates of co-occurring mood disorders and substance dependence for individuals with problematic alcohol use at three follow-ups (Schoenthaler et al. 2017). The third study (McGuire et al. 2018) examined comorbid Post-Traumatic Stress Disorder (PTSD) and substance dependency in male veterans receiving integrated residential treatment, and reported substantial improvements in PTSD symptoms and resilience for all 29 participants. Two further studies (Morse and MacMaster 2015; Rome et al. 2017) found statistically significant improvements in ASI and ASI-X composite scores for mental health at follow-ups.

A literature review by Brunette et al. (2004) identified ten controlled studies that suggested a range of outcomes are improved when services integrate treatment for substance use and mental health. The authors note that some of the studies were methodologically limited by attrition; poor engagement; different wait times for the programmes; small sample sizes; high rates of drop-outs and limited individual demographics. There was nonetheless moderate evidence for the effectiveness of integrated services, and particularly their efficacy in addressing comorbid mental health and substance use disorder diagnoses. Planned length of residential stays in these studies ranged from one month to several years, and adopted different treatment models, including therapeutic communities, 12-Step programmes and cognitive behavioural therapy (CBT) models. A study examining a short-term residential programme (Aguilera et al. 1999 in Brunette et al. 2004) reported that the integrated approach was associated with higher completion rates compared with residential rehabilitation for substance use, though not different substance use outcomes. A second study (Anderson 1999 in Brunette et al. 2004) found that integrated dual-disorder programmes predicted higher treatment retention, and that people attending the programme were more likely to enter community programmes and housing upon discharge, and avoid using substances. Burnam et al. (1995 in Brunette 2004), reported improvement in measures of psychiatric symptoms and substance use among people attending an integrated residential programme, though results were not significantly different for individuals attending an integrated outpatient programme or non-integrated outpatient programme.

Lastly, a study evaluating substance use and mental health outcomes for women engaged in sex work found that more provision for mental health during residential treatment was associated with the cessation of sex work at follow-up. The cessation of sex work in turn was associated with lower substance use, higher abstinence and fewer mental health symptoms at 12-month follow-up (Burnette et al. 2009).

4.2.3 Social outcomes

Fewer studies have explored social outcomes from residential rehabilitation. Within those that have, primary focus has been placed on offending, social engagement, and employment. Results of these studies broadly suggest that residential rehabilitation is associated with substantial social improvements. The 2019 systematic review by de Andrade et al. identified 11 studies which have explored social outcomes, all of which reported positive outcomes following residential rehabilitation. A study of Iranian males (Babaie and Razeghi 2013) found that a minimum stay of six months in a therapeutic community was associated with improved quality of life in comparison to standard residential treatment and pharmacotherapy. This study was weakened, however, by high attrition rates in the therapeutic community and pharmacotherapy groups. A New Zealand study (King et al. 2016) found that a far shorter programme (18 weeks, with 12-week average length of stay) was associated with significant improvements in a range of social indicators at three-month follow-up. An Australian study (Harley et al. 2018) on a therapeutic community with a similar length of stay (16 weeks average) found significant improvements in financial satisfaction at discharge. A Brazilian study of residential rehabilitation for crack cocaine dependency (do Carmo et al. 2018) found that 97% of individuals were unemployed on admission, while 80% of those completing the programme were employed on discharge.

A number of studies have also reported statistically significant improvements in Addiction Severity Index (including ASI-X and ASI-MV) composite ratings for social and family relationships (Deanne et al, 2013; Morse and MacMaster, 2015; Turner & Deane, 2016; Rome et al, 2017; Schoenthaler et al, 2017; Patterson et al, 2018). Recent studies specific to Scotland and residential rehabilitation outcomes are relatively few, however an Edinburgh-based study, using a European version of the ASI-X, found significantly improved scores for those who participated in the 12-week abstinence-based residential treatment programme (Rome et al. 2017). Highest composite scores were for economic status and satisfaction with work, followed by family relationships, psychiatric status and other relationships. Economic, medical and psychiatric statuses were the slowest to improve. There were also substantially better outcomes for those who completed the programme than those who dropped out, particularly in medical and psychiatric domains.

4.2.4 Mortality

A number of studies have examined mortality outcomes in relation to residential rehabilitation, although existing studies are largely dated and limited by methodological issues. It is overall difficult to synthesise results from studies examining mortality rates in relation to residential rehabilitation, given the variation in methodology and research contexts. There is similarly a dearth of research exploring mortality rates in relation to different treatment modalities. While so, some existing studies suggest that residential rehabilitation may have a higher rate of mortality when compared with alternative treatment options, though it must be acknowledged that mortality rates are affected by a number of confounding factors difficult to control for, and which in turn affect the ability to draw definitive conclusions. A high quality Australian study (Lloyd et al. 2017), for example, used retrospective data linkage to compare the effectiveness of different treatment modalities. Individuals who had engaged in residential services as their last treatment had an increased risk of death in the first year post-discharge, with only residential detoxification presenting a higher risk of death of the various treatment modalities included in the study. The authors note that this trend may reflect lack of sufficient aftercare or the comparative complexity of need among people attending residential services. They likewise note the need to explore the role of "individual, treatment and social factors that may contribute to mortality following treatment" (2017: 27). An older study explored mortality rates at one to four years following placement in a therapeutic community (Berg 2003). Nine people died during the observation period, giving a mortality rate of 2.28 deaths per 100 observation years. Mortality did not statistically differ between men and women. These findings are comparative with an older Norwegian study cited in Berg (2003) (Anderson et al. 1996 which conducted a ten-year follow-up of 482 individuals attending residential rehabilitation, with an estimated mortality rate of 1.6% in the study population. Rossow and Kielland (1995), also cited by Berg (2003), similarly found a mortality rate of 1.7% per year in the study population (N = 1491), with participants having undergone a psychiatrically oriented drug treatment in a hospital setting.

Other studies have explored factors associated with higher mortality risk following residential placements. A US -based study of 207 veterans attending a 60-day intensive residential rehabilitation programme examined factors which were specifically associated with reduced mortality and lower rates of relapse (Decker et al. 2017). Mortality rates were higher among those who had not completed treatment, those with a nicotine dependence, and those who underwent longer medical rehospitalisation. Comorbid psychiatric disorders were associated with higher risk of mortality in univariate analysis, although not when other variables were considered. Treatment completion and longer psychiatric hospitalisation were associated with reduced mortality. Treatment completion was not found to vary by type of preferred substance, with the majority of study participants engaging in polydrug and alcohol use. The authors theorised that psychiatric hospitalisation reduced mortality, in that participants were able to receive intensive medical attention, although this was noted as an area requiring further research.

A recent Australian study (Bista et al. 2021) similarly sought to explore mortality patterns among a sample of 3,256 young people engaged with a modified therapeutic community. Mortality patterns were assessed in relation to demographic, treatment and pre-treatment substance use, and mental health characteristics. Three treatment groups were included, including those who stayed up to three months in treatment; those who exited treatment after a month or less; and those who were referred but did not attend, or else who attended for less than three days. During follow-up, 64 people died, giving a crude mortality rate of 2.15 deaths per 1000 person-years. Crude mortality rates did not differ significantly between men and women, although standardised mortality ratios were nearly five times higher among women than men. The authors note this is likely due to both lower mortality rates among the female general population, and that a larger proportion of women in the study reported pre-treatment polydrug use, greater substance dependence, suicidal or self-harm attempts and clinically significant distress levels. Mortality rates did not differ significantly between treatment groups, but were closely associated with worse mental health scores on the Global Severity Index T-Score. The authors suggest that aftercare should be offered to more than just those who completed the treatment, and should incorporate targeted interventions to address mental health comorbidities.

Mortality rates were found to differ between men and women by Anderson and Berg (2009), who explored the relationship between length of stay, employment of coping strategies, social integration, sex and type of facility. Six years after treatment entry, four men and three women had died (11.7% of the cohort), all of whom had a low rating on Antonovsky's Sense of Coherence Test (SOC), used to measure coping skills. Women were also 43% less likely to drop out of treatment than men. Results suggest that sex, 'normal' SOC ratings and high degree of social integration at treatment entry were predictive factors of survival.

4.3 Selected factors associated with improved outcomes

A substantial portion of the literature explores specific factors associated with improved outcomes across a range of substance use, social, psychological, behavioural and economic domains. This section explores these studies. There is again a lack of evidence that relates directly to Scotland, although considerable work has been undertaken internationally. Evidence here broadly relates to treatment characteristics (including specific aspects of particular models), characteristics of the person accessing residential rehabilitation (including age, substance use history and motivation) and outcomes from the dynamics between these two (including retention and completion).

4.3.1 Specific Aspects of Treatment Programme

As noted, residential rehabilitation encompasses a diverse range of treatment models, programmes and philosophies, and no one model is robustly evidenced to produce routinely better outcomes. A number of studies, however, do highlight certain aspects of treatment as being effective in, for example, reducing problematic substance use, reducing contact with criminal justice services, improving mental health and wellbeing, as well as a range of other quality of life domains. Robust evidence suggests that programmes which integrate treatment for mental health comorbidities are often better placed to improve outcomes relating to mental health, social functioning and substance use (Brunette et al. 2004; de Andrade 2019). The National Treatment Outcomes Study (NTORS) (Gossop et al. 2002) found that emphasis on cognitive, avoidance and distraction coping strategies during treatment reduced rates of relapse in the cohort. As discussed above, evidence also exists to suggest that 12-Step programmes can effectively prevent instances of relapse when embedded within residential treatment or used concurrently with treatment (Kelly et al. 2013).

Treatment duration has also been found to be associated with substance use outcomes. Across the literature, the effect of programmes having longer durations (lasting three months or more) is noted by several studies as producing positive outcomes and reducing time between relapse (Eastwood et al. 2018; Gossop et al. 2000; HKCIJ 2017). A literature review undertaken by the EMCDDA (2014) alternatively suggests that optimum length of time in treatment is between 12 to 24 months.

A recent qualitative study, centring on semi-structured interviews with six key stakeholders living and working in San Patrignano in Italy, explored the mechanisms by which the programme offered by this rehab facility leads to positive outcomes in its residents (Devlin & Wight, 2020). This study highlighted a range of intertwining mechanisms and contextual factors operating at the individual and organisational levels. Organisational factors included a visionary, entrepreneurial leader; the commitment and dedication of staff, shaped largely by the predominance of people with lived experience in the workforce; the social enterprise model which the rehab followed; and the implementation of evolutionary adaptive learning.

4.3.2 Treatment retention and completion

The association between retention/completion and outcomes has also been explored within the literature. A substantial evidence base indicates that positive outcomes are predicted by higher rates of treatment retention and treatment completion (Gossop et al. 2002; Vandersplasschen et al. 2013; Willey et al. 2016; Rome et al. 2017; Eastwood et al. 2018). Contrasting these, however, a retrospective cohort study by Keen et al. (2001) found no association between positive substance use outcomes and retention or length of stay, with being abstinent at entry the only factor significantly associated with positive outcomes in relation to substance use at follow-up. Across these studies, it has frequently been noted that aiming to ensure the retention of individuals in residential rehabilitation, and the development of swift pathways back into residential rehabilitation or other appropriate services for those who do not complete a placement, is necessary to avoid negative outcomes occurring.

Previous research has also explored factors contributing to the greater likelihood of retention and programme completion. During the NTORS study, Gossop et al. (2002) found an association between higher retention rates and older adults with short histories of substance use. Retention was also found by Johns et al. (2009) to be predicted by social variables, with higher rates of retention among those who were single and those who had higher levels of social support and functioning. Several studies have furthermore examined how therapeutic relationships predict retention rates (Barber et al. 2001; De Weert-Van Oene et al. 2001; Fenton et al. 2001; Meier et al. 2006; Nordheim et al. 2018). Meier et al.'s (2006) study of 187 residents across three residential rehabilitation facilities in the UK found that weaker counsellor rated therapeutic relationships resulted in much earlier drop-outs than stronger ones, even when controlling for variables. The study measured both client and counsellor ratings on the Working Alliance Inventory (WAI-S) during weeks 1-3, alongside a range of other measures such as demographics, substance use, coping skills and treatment expectations. 100 of 187 individuals (53.5%) completed treatment, which lasted 90 days, while the other 87 dropped out prematurely. Of those who dropped out, 28.7% left by the end of the second week, 55.2% by the end of the first month and 80.5% by the end of the second month (day 60). The ratings of the therapeutic relationship by people attending the service were not related to drop-out, which stands in direct contrast to other studies of therapeutic relationships (see Barber et al. 2001 and De Weert-Van Oene et al. 2001). Meier et al. (2006) propose that therapists may have a better sense of problems with treatment goals, whereas the person attending the service may attribute problems to factors other than the individual therapist. Therapists may also, they suggest, lose motivation or confidence in the relationship if they perceive a weaker bond, which may result in lessened quality of their work. The findings of Meier et al. (2006) also echo those of Gossop et al. (2002) in that higher retention and treatment completion were predicted by older age. However it was also found that confidence in treatment, more negative experiences of past treatment, and better education increased retention. People also stayed in treatment longer when treated by counsellors with more experience of drug counselling. In contrast, crack use prior to treatment, more secure attachment styles and more developed coping strategies predicted shorter retention. Regarding the latter two, it was suggested that individuals with better 'psychological resources' might feel ready to leave treatment before its formal completion (Meier et al. 2006: 61).

4.3.3 Characteristics of Person seeking Residential Rehabilitation

There is a dearth of literature relating to the appropriate identification and assessment of candidates for residential rehabilitation. However, some evidence – often conflicting – exists concerning the association between a range of characteristics of the individual seeking rehabilitation and treatment outcomes. An observational cohort study by Eastwood et al. (2018), for example, found that successful completion of treatment and lower likelihood of returning to residential rehabilitation were predicted by a variety of individual demographic factors, such as older age, being in paid employment, and participation in community-based treatment prior to and following residential treatment. Older adults, specifically those with less forensic and psychiatric histories, were similarly found to have better outcomes in an evidence review conducted by the HKCIJ (2017). The review likewise cited several studies which found that stable housing and paid employment lowered the likelihood of relapsing following placements.

Conversely, as discussed previously, Keen et al. (2001) found no association between the characteristics of the person observed at entry and positive outcomes, including age, sex, route of drug administration, polydrug use, length of time addiction or age of first addiction. The only individual characteristic associated with success was being abstinent on entry as opposed to requiring detoxification. The National Treatment Outcomes Study (NTORS) (Gossop et al. 2002) similarly found no statistically significant differences in age, sex, having drug-using friends, partners who used drugs, psychological health and physical health scores, or previous treatment contact between those who maintained abstinence and those who returned to using substances at follow-up.

As noted previously, there is little evidence on the association between substance use histories/profiles and outcomes. Gossop et al. (2002) found that neither duration of heroin use nor severity of heroin dependence were significant predictors of relapse post-treatment, although people who relapsed reported more frequent use of heroin in the 90 days prior to treatment. There was also little difference between those who 'lapsed,' relapsed and maintained abstinence in rates of pre-treatment use of non-prescribed methadone, benzodiazepines, cocaine powder and amphetamines. Those who maintained abstinence were less likely, however, to have been users of crack cocaine at intake. Meier et al. (2006) similarly observed that crack cocaine use was associated with poorer outcomes, although a comparison of outcomes by histories/profiles of substance use was not a primary aim of the study. Meier et al. (2006) controlled for heroin and crack cocaine use, but not other substances, including alcohol.

An Australian study, using the Implicit Association Test, found that identification with heroin and/or alcohol predicted continued drug and alcohol use among people attending or having attended residential rehabilitation (Wolff et al. 2014). Without relying on self-reported measures of identification with substances, it was also found that identification with heroin and alcohol predicted treatment retention rates. Millar et al. (2014) alternatively suggest that people who use opioids are 'more amenable' to residential rehabilitation than those treated with other modalities. The study measured self-reported levels of drug use, offending, social measures and health, finding that people seeking residential rehabilitation for opioid use were better motivated than those seeking other treatment options.

Studies of how motivation affects the outcomes of residential rehabilitation are limited and often conflicting. Gossop et al. (2006) conducted a longitudinal prospective cohort study of 1075 individuals across 54 treatment agencies in England. No statistically significant association was found between readiness to change measures and use of opioids and stimulants at follow-up. The authors conclude that motivation and stated intentions do not translate directly into outcomes and are not sufficient mechanisms for treatment success. Motivation has been found in other studies to relate to engagement and satisfaction, although these in turn are not always directly associated with post-treatment outcomes (Gossop et al. 2003; Callaghan et al. 2005). Alternatively, as outlined above, Millar et al. (2014) argue that motivation may lead to better outcomes for people who have attended residential rehabilitation in comparison with other treatment modalities.

Although not specifically focused on outcomes, Piontek et al. (2017) explored how time in residential rehabilitation affected the 'treatment readiness' of the individual and, in turn, motivation. The study, based in Germany, also assessed whether treatment readiness differed by sociodemographic and substance use profiles. Average motivation was not found to significantly change for all individuals over the course of treatment, although rates of change did differ on an individual level. Treatment readiness was furthermore predicted by higher level of education, being employed, earlier substance use treatments and levels of craving. Alcohol use frequency and use of illicit drugs did not show statistically significant effects on treatment readiness.

Individual level factors identified by the aforementioned recent qualitative study of San Patrignano (Devlin & Wright, 2020) included commitment and motivation to recovery; removal from the formal social environment; being treated with respect; continual socialisation and communal living; the availability of peer mentors with lived experience; highly structured day programmes with rules and routines; and the undertaking of meaningful work in the sector of social enterprise.

Shifts in social identity and participation in social networks have also been found to be important determinants of outcomes following residential rehabilitation. Beckwith et al.'s (2019) analysis, using the Social Identity Mapping in Addiction Recovery (SIM-AR) method, of 155 Therapeutic Community (TC) residents in Victoria, Australia found that substance use severity at follow-up was associated with changes in social identity and network composition. Better substance use outcomes were related to a decreased substance-using identity and an increase in recovery identity, as well as to a decrease in proportion of high-risk groups and an increase in proportion of low-risk groups in one's network. These factors intertwined; a shift towards a stronger recovery identity was positively associated with changes in the proportion of low risk groups in one's network. Similarly, Haslam et al. (2019) investigated the contribution that different types of social relationships in the period prior to residential treatment made to engagement with the treatment community and the subsequent recovery potential. Exploring the relationship between individual and group relationships and outcomes associated with quality of life, they found that group membership prior to treatment was associated with TC identification and quality of life. Further, early identification with TC group membership formed a particularly important platform for improvement in quality of life, functioning as a positive form of influence.

4.3.4 Aftercare

Aftercare is a well-documented and specific form of continuing care, and continues following the individual's placement in residential rehabilitation. There is a strong evidence base around the importance of ongoing support and aftercare as predictors of success following residential rehabilitation, alongside those studies demonstrating the importance of employment and stability in the individual's housing situation as outlined above. Gossop et al. (2003) placed specific focus on the impact of aftercare and continuity of care following residential treatment. Their study explored the influence of the mutual aid fellowship group of Alcoholics Anonymous (AA) prior to, during and following residential rehabilitation in an National Health Service (NHS) inpatient unit for alcohol problems in London. They found significant associations (15%) between frequency of attendance at AA meetings and reduction in alcohol consumption following residential treatment. They found that AA attendance was not significantly associated with quality of life or a reduction in psychiatric symptoms at 6 months follow-up, suggesting that both mutual aid and continuing care in the community following residential rehabilitation are of importance. A two- to three-year follow-up of residents who had completed a residential rehabilitation placement for opiate dependency in Ireland found that abstinence was significantly associated with both programme completion and aftercare (Smyth et al. 2005).

A study in the US by Jason et al. (2015) examined the role of aftercare for a sample of individuals who had (mainly) attended residential rehabilitation facilities. Their study drew on a cohort of 270 people who had been attending either a TC, a usual care setting or a recovery home termed an 'Oxford House'. This latter referred to a residential rehabilitation setting that did not include the formal therapeutic change interventions common to TCs, or on-site access to medical staff or other health professionals. Those who were assigned to the Oxford House following completion of a TC placement demonstrated a range of positive outcomes in relation to those who had gone back to their usual home settings (whether staying with friends of family members, their own house or apartment, homeless shelters or other settings), emphasising the importance of continuity of care and of a protective and stable home environment that can encourage and nurture recovery pathways.

A review of evidence for the EMCDDA undertaken by Vanderplasschen et al. (2014) found that participation in aftercare, post-treatment employment, and old age were the strongest predictors of abstinence following residential rehabilitation placements in a therapeutic community setting. These findings were somewhat contrasted by Vanderplasschen et al. (2010) in their analysis of data collected through the Drug Outcome Research in Scotland (DORIS) study. Comparing outcomes across a 33 month period for 653 individuals, it was found that participation in aftercare did not substantially improve outcomes at 16-month follow-up, following residential treatment (although did improve outcomes following prison-based and community treatment). Aftercare participation was clearly associated with abstinence after eight months, with rates dropping at 16 months, before stabilising again at 33 months. Aftercare was not, however, established as an independent predictor of abstinence. Aftercare participants did score significantly better on various other outcome measures, such as incarceration, drug possession and contacts with non-drug using friends. The study concluded that while aftercare was still an important outcome predictor, treatment modality was a much stronger predictor, with residential rehabilitation in particular associated with substantially improved drug outcomes in comparison with other modalities. The authors highlight that, because programme participation was a stronger predictor of outcomes, and strongest for those in residential rehabilitation, most positive effects may occur during residential treatment and leave little room for further improvement following discharge.

While little data is available, the outcomes from aftercare have in some cases been found to be shaped by their fit with the residential rehabilitation programme which the individual has attended. A study by Best et al. (2020), for example, found that mutual aid engagement had a differential effect on treatment outcomes by the type of residential treatment service attended. Their quantitative study found that, while those who attended two types of residential rehabilitation service typically engaged in mutual aid groups, mutual aid groups were not associated with improved outcomes for those attending TCs. Prior to engagement with aftercare, TCs generally produced better outcomes than other forms of residential treatment, however, post-treatment participation in mutual aid groups produced comparable outcomes between people attending TC and non-TC facilities. Those who had attended TCs therefore received no added value in reducing or abstaining from substances through mutual aid. Best et al. (2020) suggest that the above results can be attributed to philosophical inconsistencies between effective therapeutic community engagement and engagement in 12-Step mutual aid groups in particular. It is also interesting to note that TC residents were more likely to engage with SMART Recovery, a non-spiritual and non-disease based model whose underlying rationale and philosophy may be more compatible with that of a TC. The authors suggest that these findings can also be interpreted from a social identity theory perspective (e.g. Jetten et al. 2015); an approach based on the idea that identity is in part conferred by the groups we belong to and that we positively value. Thus, for the graduate from a TC, their social identity may be that they are 'recovered' and that their dominant social identity is around active re-engagement in the community and the shedding of the substance use identity. However, this is incompatible with embracing a 12-Step philosophy that requires a recognition of 'powerlessness' over substance use and the retention of the 'addict' identity as key to the ongoing recovery process.

4.4 Outcomes of different residential rehabilitation programme models

This section explores the evidence in relation to two different treatment philosophies dominant within the literature; therapeutic communities and 12-Step Programmes, recognising that psychotherapeutic models (including CBT, motivational interviewing and individual and/or group therapy) are often employed across both approaches. While having presented the data in this way, it is acknowledged that these often form 'ideal types'; the aforementioned Scottish Government pathways survey of residential rehabilitation providers (Scottish Government 2021) found that a number of facilities in Scotland offer a programme which incorporates components drawn from these various approaches.

4.4.1 Therapeutic Communities

Therapeutic Communities (TCs) are a form of residential treatment that place emphasis on 'the community,' social relationships and structured activities as forces for psychological and social change (EMCDDA 2014). They largely employ a hierarchical model of care, in that different treatment stages entail different levels of personal and social responsibility, and peer influence is harnessed to both promote behaviour change and demonstrate the TC value system. There are numerous variants of the TC model, which began as longer-term residential treatment programmes and from the 1990s were increasingly modified towards short-term placements.

A body of research examines the effectiveness of TCs in producing positive outcomes; their effectiveness in comparison with other treatment modalities; their structure and everyday operation, and the qualitative experiences of those attending TCs (Agosti and Levin 2007; Albertella and Norberg 2012; Fernández-Calderón et al. 2015; Clarke and Waring 2018; Kaye 2020; González-Saiz and Vergara-Moragues 2021). The evidence base for the efficacy of TCs in comparison with other treatment modalities, including other forms of residential treatment, remains mixed, although there is robust evidence to suggest that they are effective in reducing substance use and criminal activity. The report by the EMCDDA (2014) included a literature review of 28 peer reviewed articles that reported on 16 RCTs and quasi-experiments (all in North America) and a further 21 articles reporting 14 Europe-based observational studies. The report finds moderate quality evidence for the effectiveness of TCs in achieving stated goals, such as reducing substance use, increasing employment and improving mental health in the US context. They note that as European studies are largely observational they are limited in the strength of their conclusions, but that TCs had generally positive outcomes in social and health domains. Retention rates varied across all contexts.

A systematic review by Smith et al. (2006) similarly found that the evidence on TCs was limited by methodological weaknesses and potential biases, but that there was at least some evidence of benefits of TCs. One RCT found that abstinence rates were significantly higher in a TC setting compared with other forms of residential rehabilitation, while others found that, when compared against other treatment programmes in prisons, the TC group had lower rates of re-incarceration and less alcohol and drugs offences 12 months after release. Lower rates of re-incarceration were echoed in the EMCDDA (2014) report, which also found reductions in reoffending and time to reincarceration for TC groups compared against control groups. Another systematic review (Malivert et al. 2012) based on 12 studies of TCs, reported highly variable completion rates for TC programmes (9–56%), and concluded that their effectiveness remains highly contentious, with concerns about completion rates and long-term outcomes outlined in a Cochrane Review by Smith et al. (2006).

Limited works have explored the impact of demographics and individuals' characteristics on TC treatment outcomes. An observational study based in Spain performed a latent class analysis on 4,102 TC attendees' substance use diagnosis, in order to determine the impact of SUD profiles on social, medical and psychiatric outcomes (Fernández-Calderón et al. 2015). The study found that profiles involving heroin, cocaine and benzodiazepines were associated with poorer social and medical function and impulse control disorders, while alcohol and cannabis/cocaine use were associated with cognitive dysfunction and psychotic illness, respectively.

There are similarly a number of studies that have explored TC outcomes in young people, finding positive treatment effects on substance use; health; psychological functioning in the year following treatment (Godley et al. 2001; Dasinger et al. 2004; Morral et al. 2004; Agosti and Levin 2007; Albertella and Norberg 2012). Others have found that positive effects tend to wane after 12 months (Jainchill et al. 2005; Edelen et al. 2007).

Several studies have explored the effectiveness of TCs within prison settings (Lipton et al. 2002; Vandevelde et al. 2004; Stevens 2012). Mitchell et al. (2007) conducted a meta-analysis of 30 prison-based Addictions TCs and found that the mean recidivism rate for the group was 43%, in comparison with an estimated 50% rate for the control group. Lipton et al. (2002) similarly conducted a meta-analysis of 42 studies that considered the effectiveness of TCs and milieu therapy (MT) in reducing recidivism. Both TCs and MT significantly reduced rates of recidivism, with a holistic approach; emphasis on personal and community identity; and longer lengths of stay contributing to positive outcomes. The authors suggest that prison-based interventions followed by community treatment are important aspects in incremental learning and recovery journeys. Ethnographic work by Stevens (2012) on prison-based TCs suggests that processes of identity reconstruction and narrative reframing of the self are connected to lower rates of recidivism in serial offenders. Stevens notes that by committing to 'radically different' cultures, TCs enable the production and reinforcement of individual change and pro-social behaviour.

Other qualitative works have drawn attention to numerous facets of the experience of those attending TCs, in addition to certain practices which impact recovery-oriented outcomes (Chenhall 2008; Gosling 2018). For example, ethnographic works have explored the benefits of a TC in England in developing recovery capital, suggesting that recovery capital in turns offers a practical framework for exploring mechanisms of change (Gosling 2018).

While there is a lack of literature that compares TCs across international contexts, a comparative study by Goethals et al. (2011) explored differences between TCs in Europe and the US. They found that TCs in Europe scored significantly higher on the Survey of Essential Elements Questionnaire (SEEQ) – designed to explore whether residential programmes were faithful to the essential elements of a traditional TC model, as outlined by De Leon (2000) – compared with those in the United States, However, these contrasted existing literature in that they suggested relative congruence between European and US approaches to TCs; a departure from their relative difference previously.

As with much of the literature relating to residential forms of treatment, a substantial portion of the evidence-base concerning TCs is outcomes-focused. Clarke (2017) notes that there is a general lack of research that focuses on gendered experiences of TCs the power dynamics that structure social relationships, including their impact on outcomes.

In summary, TCs are underpinned by a relatively robust body of evidence, some of which suggests that TCs can be more effective in producing positive outcomes than other residential treatment modalities. TCs have been found to be particularly effective when examined in relation to young people and people in prison, although retention and completion rates are subject to high variation across all demographics.

4.4.2 12-Step programmes

Although there exists some variation in the international definition and application of 12-Step programmes, they broadly consist of a highly structured, step-based process, often based on a 'spiritual' understanding of recovery (Galanter 2007). In this sense, 12-Step programmes are often considered 'spiritual fellowships,' premised on a definition of spirituality as that which "gives people meaning and purpose in life" (Puchalski et al. 2014 in Galanter 2007). In comparison with residential TCs, there are generally fewer studies that consider the efficacy of 12-Step programmes in a residential treatment setting. Most research concerns the efficacy of 12-Step on leaving residential rehabilitation, or as an intervention undertaken concurrently with residential treatment (Lopez Gaston et al. 2010).

A systematic review and meta-analysis of spiritual and religious (S/R) treatment interventions reported that 12-Step-oriented interventions produced substantially better results than other interventions and in comparison with no-treatment groups (Hang-Hai et al. 2019). The study specifically analysed 12-Step treatments in the context of formal recovery programmes and not in community settings, and as such excludes Alcoholics Anonymous and Narcotics Anonymous. Comparison interventions included those commonly utilised in substance use treatments, such as CBT and Motivational Enhancement Therapy, among others. Sixteen of the 20 studies identified were 12-Step oriented, and four specifically tested the efficacy of 12-Step interventions. The systematic review concludes that 12-Step oriented programmes are more efficacious than comparison interventions for substance use outcomes. However, it was also noted that the overall quality of the studies sourced was poor, and limited by attrition bias. Similarly, the systematic review by de Andrade (2019) that sought to establish the overarching effectiveness of residential rehabilitation, noted that outcomes were notably better for those who engaged with 12-step support on leaving residential rehabilitation.

Greater attendance of, and active involvement in, 12-Step groups during residential treatment was found by Kelly et al. (2013) to be associated with significantly greater percentages of days abstinent (PDA). The study involved assessment of 303 adults enrolled in residential rehabilitation and a 12-Step programme at intake, three, six and twelve months. Positive outcomes were more strongly associated with active involvement than with attendance alone. A randomised control trial undertaken in California likewise suggests that 12-Step programmes employed in residential rehabilitation settings are effective in both reducing substance use and reducing the adverse effects of substance use (Banerjee et al. 2009). Another US-based study (Decker et al. 2017) found that completion of a 12-Step residential programme was a significant predictor of reduced relapse rates and reduced mortality.

Lopez Gaston et al. (2010) examined engagement and barriers to engaging with 12-Step models of treatment for patients of an inpatient detoxification unit in England. They note that affiliation levels were low and that perceived religious components acted as a barrier to engagement, as did "prior negative experiences in 12-step meetings, failure to identify with group members, and difficulties simultaneously participating in both 12-step meetings and structured treatment" (Lopez Gaston et al. 2010: 313).

4.5 Effectiveness of residential rehabilitation in comparison with non- residential treatment interventions

Relatively little research explores the effectiveness of residential rehabilitation in improving outcomes in comparison with other forms of treatment, such as outpatient/community-based treatment and Medication Assisted Treatment (MAT). The majority of the small existing evidence base suggests that residential rehabilitation produces more positive outcomes in relation to substance use than other treatment modalities, although conflicting evidence exists in some cases. Assessing the comparative effectiveness of residential rehabilitation is again made difficult by the aforementioned factors.

The Drug Treatment Outcomes in Scotland (DORIS) (McKeganey et al. 2009) found that people who had attended residential rehabilitation were more likely to sustain a 90-day period of abstinence than those receiving MAT or other treatment options. There were no significant differences in the severity of dependence among those receiving residential rehabilitation and MAT. It is, however, noted that results may have been affected by other factors not controlled for, such as individuals' motivation.

The Australian Treatment Outcomes Study (ATOS), which included 615 participants, found that residential rehabilitation was 16 times more effective in achieving abstinence at the three-year follow-up when compared with MAT and inpatient withdrawal management (Teesson et al. 2008). At the 11-year follow-up, research participants who had undergone residential rehabilitation were also more likely to have remained abstinent. The Methamphetamines Treatment Evaluation Study (MATES), which included 360 participants, evidenced that residential rehabilitation successfully produced large reductions in the frequency of methamphetamine use at three-months relative to inpatient withdrawal and a non-treatment control group, although rates of reduction slowed over time (McKetin et al. 2012).

Similarly, a prospective, multi-site treatment and outcome study with 796 people from 21 alcohol and drug services in Victoria, Australia (Manning et al. 2017) explored substance use outcomes among people who had accessed a range of treatment interventions, including individual and group counselling, inpatient or home-based withdrawal, pharmacotherapy treatment or residential rehabilitation. With 52.0% of the overall sample having achieved abstinence or reduction in drug use at 12 months, having residential rehabilitation as the primary treatment intervention was a significant predictor of abstinence from all drugs of concern, although this was limited to those with a primary alcohol problem (Manning et al, 2017).

A randomised study of 227 people allocated to three different treatments for substance dependency in Iran found that therapeutic communities produced better mental health and quality of life outcomes than MAT and other forms of residential rehabilitation (Babaei and Razeghi 2013). Therapeutic communities are defined in the study as interventions that emphasise long-term care, and which centre education, personal growth, and self-efficacy, although no definition is provided for residential rehabilitation, so it is unclear how these differed. The study is also limited by the small sample size and the uneven distribution of participants across treatment modalities. A large South African study found similar findings; those at discharge from residential rehabilitation were significantly more likely to be abstinent in comparison with those attending outpatient treatment services (Myers et al, 2018).

Contrasting these findings, a US-based study that considered the impact of drug treatment on recidivism among drug-involved offenders found no significant differences between those receiving residential and non-residential treatment (Krebs et al. 2009). Those receiving residential treatment did, however, take longer to recidivate than those who received no treatment. The study employed a propensity model analysis of administrative data of 129,577 people, observing explanatory factors such as demographics and history of involvement with criminal justice in order to account for baseline differences. The authors note that the study is limited by the lack of data on individual histories of substance use and an inability to strictly define what constituted treatment.

4.6 Cost effectiveness of Residential Rehabilitation

There is a dearth of research exploring the cost-effectiveness of residential rehabilitation. Again, it is acknowledged within the literature that comparative analysis of the cost-effectiveness of different treatment modalities is challenged by the difficulty in terms of ascribing short, medium and, particularly, longer-term outcomes to a single episode of treatment given that people likely draw on multiple treatment episodes (Godfrey et al. 2004).

One study of the cost-effectiveness of drug treatment services in the UK demonstrated that residential rehabilitation and other treatment modalities (specialist inpatient treatment, community treatment and MAT) were associated with substantial economic benefits over the two year period post-follow up (Godfrey et al. 2004). These researchers measured both healthcare and other expenditure, such as criminal justice, on a cohort of 549 individuals both two-years prior to their index treatment, and two-years post-treatment. The study found a ratio of potential consequences to net treatment investment ratio of 18:1 across all treatment modalities, although it was not possible to break down this analysis by treatment type given the aforementioned complexity of relating outcomes to any single episode of treatment. The researchers suggest that these estimates were conservative, in that many potential benefits were not estimated.

A US study by Mojtabai and Graff Zivin (2003) explored the cost effectiveness of four treatment modalities for substance use disorders using propensity score analysis. The study found that all four modalities – inpatient programmes, residential rehabilitation programmes, outpatient detox and methadone programmes, and outpatient drug-free programmes – were cost-effective in comparison with other, non-substance related health interventions. Of the four modalities, this study found that the outpatient drug-free treatment programmes were the most cost effective, even for those who were more likely to choose (or be referred to) treatment in other modalities. However, this study was limited in that it did not consider wider costs mitigated by treatment, such as reductions in criminal activity, and cautioned that the data on which it drew was now relatively outdated (with the rate and length of residential treatment having reduced in the period since). These authors noted that their study contrasted with previous analysis by Hubbard et al. (1989) which found that residential treatment provided the greatest cost-effectiveness in comparison with MAT and outpatient drug-free programmes when taking wider costs and benefits (including employment, criminal activity and mental health) into consideration.