Alcohol Treatment Matrix cell A5: Interventions; Safeguarding the community

2020/21 update funded by

Alcohol Change UK

Previously also funded by

Society for the Study of Addiction

Developed with

Alcohol Treatment Matrix

Includes brief interventions

Drug Treatment Matrix

Includes harm reduction

About the matrices

Articles, slides, and video

Comment/query

Matrix cell

Alcohol Treatment Matrix cell A5

Interventions; Safeguarding the community

Key studies on the impact of alcohol treatment on the community including families, children and crime. Explores the core contradiction between punishment and rehabilitation, asks whether this accounts for the poor record of criminal justice treatment, highlights the most robust test yet of brief alcohol counselling in probation, asks whether it can ever be safe to leave children with severely dependent drinkers, and recounts the alleged deception at the heart of a recommended treatment method.

S Seminal studies K Key studies R Reviews G Guidance more Search for more studies

Links to other documents. Hover over for notes. Click to highlight passage referred to. Unfold extra text

S No use ordering ‘chronic drunks’ to AA (or to treatment) (1967). Identified by reviews (1; below) as one of just three randomised trials of self-help groups, among arrestees who already had a suspended sentence hanging over their heads it found that a court order to attend an “alcoholism clinic” or AA meetings did not further reduce arrests over the following year compared to no treatment; if anything, the reverse. Originally optimistic about the interventions, the authors ended by concluding that their findings “offer no support for a general policy of forced referrals to brief treatment”. Related study and review below. For discussion click and scroll down to highlighted heading.

S Disulfiram backed by sanctions helps reform repeat alcohol-related offenders (1966). In the early ’60s in Atlanta in the USA, a pioneering trial tested whether instead of another spell in jail, ‘skid-row’ repeat drunkenness offenders would take a drug which generates deterrent physical reactions to alcohol. Most did, belying their supposedly hopeless condition. The Effectiveness Bank commentary describes an early trial (1983) in London which tested a similar programme with similar results. For a related discussion click and scroll down to highlighted heading.

K No significant benefits from court-ordered treatment in England (2011). In the English Midlands court-ordered treatment for problem-drinking offenders on probation could not be shown to have reduced re-offending more (though the reductions were substantial) than previous probation arrangements. Similar evaluations in Lancashire (2010), West Yorkshire (2011) and nationally (original source, [UK] Ministry of Justice, 2009) also found considerable improvements but did not benchmark these against any, or any adequate, comparison groups. For discussion click and scroll down to highlighted heading.

K UK anti-offending programme did not cut crime (2011). The main cognitive-behavioural group therapy programme (ASRO) for problem substance users on probation in the UK could not be shown to have affected reconviction rates. See also similar UK findings (2012) from the same cognitive-behavioural family of interventions applied to drink-driving. For discussion click and scroll down to highlighted heading.

K No crime-reduction dividend from offering brief counselling to drunk arrestees in England ([UK] Home Office, 2012). Government-funded pilot schemes found no crime-reduction benefits from brief alcohol counselling for arrestees under the influence of drink, confounding hopes that these ‘arrest referral’ schemes would help quell late-night alcohol-related disorder. The schemes did, however, uncover many dependent drinkers. Related review below. For related discussion click and scroll down to highlighted heading.

K In Wales no crime-reduction return from brief intervention for young men convicted of violence while drunk (2008). Over the following year a randomised trial of a 15–20 minute counselling session based on motivational interviewing did not find it had significantly reduced offending or drinking, though emergency unit attendance for injuries was reduced. Related review below. For related discussion click and scroll down to highlighted heading.

K In UK probation services brief counselling no better than a basic warning at curbing drinking but crime may have been reduced (2014). The largest alcohol brief intervention evaluation yet conducted in Britain found risky drinking rates fell as much after a minimal warning about excessive drinking as after more sophisticated and longer alternatives, but these might (the researchers were unsure) have further reduced the reconviction rate. A similar Scottish study did not directly test effectiveness. Related review below. For discussions click here and here and scroll down to highlighted headings.

K No significant reduction in offending from alcohol treatment in English prisons (2020). Based on a comparison between prisoners recorded as having an alcohol use disorder and released during 2013–14 who while in prison were treated versus those who were not. Overall the proportion who reoffended in the year after release was virtually identical. The few high-risk prisoners who underwent relatively extensive treatment using recommended psychosocial methods may have been less likely to reoffend than the untreated comparison group, but this difference was not statistically significant, and the comparison was not of like with like. For discussion click and scroll down to highlighted heading.

K No offending dividend from UK AA-based prison programme (2018). On key measures the post-release offending records of prisoners who received an intensive programme based on the 12 steps of AA were virtually identical to those of a matched set of prisoners who did not receive the programme. For discussion click and scroll down to highlighted heading.

Families with children initially at risk of being taken in to public care and actually taken in during follow-up period

K Intensive support for problem drinking parents enabled children to stay at home (Welsh Assembly Government, 2008). Evaluated a service which worked intensively over a few weeks with substance-using parents (mainly involved with alcohol) whose children faced imminent care proceedings. The initiative delayed and shortened their removal from the home; a later evaluation (2012) confirmed this was not at the expense of the children’s welfare chart. Listed below the initial evaluation of a national rollout of similar services and related reviews (1 2) and guidance. For discussion click and scroll down to highlighted heading.

K Lessons from Welsh pilot of integrated support for children affected by substance use in the family (Welsh Government, 2014). Evaluation of the first three local schemes in a nationwide rollout of services based on the evaluations listed above. Documents how the schemes changed in response to experience and strategic and operational contexts. Related reviews (1 2) and guidance listed below. For discussion click and scroll down to highlighted heading.

K No demonstrable benefits from intensive support for “troubled families” ([UK] Department for Education, 2011). Early evaluation of national scheme which financially incentivised providers to ‘turn round’ troubled families in England found substantial remission in substance use problems but could not attribute these to the interventions. Later evaluation (2016) of the programme as implemented from 2012 to 2015 found that relative to comparison families, there were no significant impacts on substance use, employment, job-seeking, school attendance and anti-social behaviour. From 2015 the programme was revised to target families with a much broader range of disadvantages and to help younger children benefit. A series of evaluation reports have been published, but the studies (2017) lacked an adequate comparison group. Related reviews (1 2) and guidance listed below. For related discussion click and scroll down to highlighted heading.

K Problem-solving and collaborative approach improves outcomes of child care proceedings in London (2016). Addressing parents’ entrenched substance use and other problems was at the heart of the first UK family drug and alcohol court in the UK. Compared to ordinary care proceedings, it achieved sustainably improved parental and child outcomes at lower cost ([UK] Home Office, 2012). The courts spread outside London. Observations and interviews with judges showed they had (2016) implemented the intended collaborative, problem-solving ethos and given parents a voice, while still prioritising the child’s welfare. The new courts made parents feel (2018) valued, supported, able to share their difficulties, and fairly dealt with. For discussion click and scroll down to highlighted heading.

K Support the relatives too (2011). Brief counselling by specially trained primary care staff seemed to help relatives in England cope with living with a problem drinker, but without a control group against whom to benchmark the outcomes, we cannot be sure that the benefits were actually due to the interventions. Related guidance below (1 2).

Improvements in violence in the family and child welfare after treatment for drinking problems

K Patient-focused treatment helps partners and children too (2003 and 2006). Study of 301 men living with female partners and seeking treatment at two US outpatient alcohol clinics showed that even when treatment is focused on the man with the drinking problem, families benefit in the form of reduced violence and improved child welfare chart. Related review below. For related discussion click and scroll down to highlighted heading.

K If feasible, families benefit most from couples therapy (2009). Compared to individual therapy, found that anti-violence benefits for partners of people with drinking problems were greatest when both were allocated to couples-based therapy which addressed relationships as well as drinking. See also similar couples-therapy reports focused on men (2004) or women (2009; free source at time of writing) with drinking problems, and UK-based advice (2007; free source at time of writing) on how to avoid the risk that couples therapies might provoke partner abuse. Listed below further couples-therapy report from same research stable and related review and guidance (1 2).

% of children of problem drinking fathers with clinically impaired psychosocial adjustment

K When dad has a drinking problem, couples and children benefit most from couples therapy (2002; free source at time of writing). Compared to individual therapy only or the passive attendance of the female partner, adding couples therapy to individual therapy for men seeking treatment for drinking problems significantly improved the functioning of their children ( chart) and the partners’ relationships. Child welfare may be further improved (2008; free source at the time of writing) by integrating joint mother/father parental skills training with couples therapy. See also UK-based advice (2007; free source at time of writing) on how to avoid the risk that couples therapies might provoke partner abuse. Couples-therapy report from same research stable listed above. Related reviews (1 2) and guidance (1 2 3 4) below.

R Routine alcohol treatment can reduce domestic violence (2009; free source at time of writing). When successful, alcohol treatment in general results in reduced violence between sexual partners; couples therapy has yet greater impacts, but is not always (2007; free source at time of writing) safe or feasible. Related studies above (1 2). For discussions click here and here and scroll down to highlighted headings.

R Family programmes can improve the prospects of children whose parents have substance use problems (2012) Of the reviewed programmes, most effective were those which involved both parents and children, particularly the Strengthening Families Programme (2004). Related review and guidance below. For discussion click and scroll down to highlighted heading.

R Programmes for substance using parents or their children validated in randomised trials (2015). Covers the same territory as review above, but narrowed down to trials where families were allocated at random to the evaluated intervention versus a comparator and where children were shown to have benefited on at least one measure. Just four studies met these criteria; the two most relevant (1, free source at the time of writing; 2) are listed above (1; 2). Related guidance below. For related discussion click and scroll down to highlighted heading.

R Alcohol treatment prevents injuries (Cochrane review, 2004) … and also causes of injury such as violence and accidents; same lead author was responsible for an earlier review (1999) analysed for the Effectiveness Bank. Similar message tentatively emerged from another review analysed (2000) for the Effectiveness Bank. For discussion click and scroll down to highlighted heading.

R No “robust” support for any type of alcohol intervention in the criminal justice system (2019) “No specific model of treatment at any stage of the criminal justice system was supported by a substantial, robust and consistent body of literature,” concluded a UK review of the international literature. Studies were either too few or where there were an appreciable number the results were less promising. Related reviews below (1 2). For discussion click and scroll down to highlighted heading.

R “Very little evidence” for brief interventions in UK criminal justice system (2016). Found “very little evidence of effectiveness of brief interventions … mainly due to the lack of follow-up data”. Similarly a review of the international literature on brief interventions in prisons (2016) concluded, “there is some promise in terms of effects [but] not enough studies have been carried out to ascertain efficacy or effectiveness and adequate methodological rigour in the available literature is questionable”. Related studies (1 2) and review above. For discussions click here and here and scroll down to highlighted headings.

R How to stop drink-drivers reoffending (2006). Broader review of drink-driving and responses to it includes the “encouraging” results from rehabilitation programmes. Related guidance below.

R Attending AA: encourage but don’t coerce (1999). Synthesis of studies concludes that people forced by courts or other means to attend AA do worse than when coerced instead into professionally run treatments or left to their own devices. When participants choose AA or allied treatments overall they do significantly better in terms of drinking reductions than when they choose no treatment and sometimes better than in less intensive alternative treatments, but these non-randomised studies are unable to eliminate bias due to more motivated or otherwise more promising participants opting for AA-based approaches. Related seminal study above.

R Is therapy undermined by a punishment context? (2005). Asks whether in criminal justice settings, the contradictions of helping and punishing at the same time (“motivational arm-twisting”) undermine interventions which might work elsewhere – in particular, the client-centred motivational interviewing style of counselling. For discussion click and scroll down to highlighted heading.

G Offender management guidance for England and Wales ([UK] National Offender Management Service, 2010); Treating prisoners in Scotland (Scottish Prison Service, 2011). Official guidance on the commissioning, management and delivery of interventions for alcohol misusing offenders, dating from before the transfer of responsibility for treatment in prison to the NHS. For discussion click and scroll down to highlighted heading.

G Managing alcohol problems among prisoners (World Health Organization, 2012). Based on UK experience, offers an integrated model of best practice in care for problem-drinking prisoners, including a consideration of specific types of treatments. For discussion click and scroll down to highlighted heading.

G Scottish guidance on working with children, young people and families affected by problematic alcohol and/or drug use (Scottish Government, 2013). Intended for all child and adult services, including drug and alcohol services. Includes what new patients should be asked about children and the role these services should play in a system which (Getting our Priorities Right is the document’s title) prioritises child welfare. Related local toolkit for practitioners listed below. For related discussion click and scroll down to highlighted heading.

G Toolkit to help practitioners safeguard children and families affected by problem substance use (NHS Lothian and partner agencies, 2014). Co-produced by health, social and enforcement authorities in the Edinburgh region. Designed to assist the day-to-day practice of health and social care practitioners working with children and families affected by alcohol and drug problems in the family. Getting it right in the toolkit’s title echoes the Scottish national guidance listed above. For related discussion click and scroll down to highlighted heading.

G Implementing support systems to prevent domestic violence and abuse related to substance use (2017). From Adfam, the national UK charity specialising in drugs and the family, good-practice guidance including intervention principles and specific programmes/approaches. Brings together Adfam’s 30+ years of experience in family support. Related studies (1 2 3) and review listed above and guidance below.

G How to broach and manage domestic violence with clients in substance use services (2012). Australian guidance on identifying and dealing with clients who may be perpetrators as well as victims. Related studies (1 2 3), review and guidance listed above.

G Treating the drink-driver (Health Canada, 2004). Authors reviewed evidence and consulted experts to arrive at recommended education and treatment and rehabilitation approaches to alcohol/drug impaired driving. Related review above.

G US expert consensus on treatment in the criminal justice system ([US] Substance Abuse and Mental Health Services Administration, 2005). Guidance on interventions, matching to the offender, and planning programmes.

more Retrieve all relevant Effectiveness Bank analyses or search more specifically at the subject search page. See also hot topics on supporting families, testing and sanctions and protecting children.

Close Matrix Bite guide and commentary Close Matrix Bite

Links to other documents. Hover over for notes. Click to highlight passage referred to. Unfold extra text

What is this cell about? Rows 1–4 of the matrix focus on what treatment can do for the client or patient. Row 5 moves out to what treatment can do for the rest of us, starting with this cell on the impacts of the interventions themselves. Included are evaluations of treatment funded or ordered to safeguard the wider community, and studies of whether treatment in general has a safeguarding impact.

While ethically treatment must focus on the welfare of the individual patient, it may be funded and organised by authorities whose primary motivation is to safeguard the wider community. In these cases, treatment is offered or imposed not because the substance user has sought it, but because it is thought that treating their substance use could result in benefits to the community. Typically these take the form of reductions in crime including drink-driving and violence, but also improvements in parenting and child welfare and reductions in non-criminal behaviour which the community finds offensive and/or which degrades the local social or physical environment. Treatment not organised primarily for these purposes may nevertheless have these benefits; studies and reviews documenting these effects are also included in this cell.

Also here are interventions which focus on the welfare of the children and families of problem drinkers in their own rights, rather than primarily as a means to promote the welfare of the problem drinker. Among these are peer support initiatives (see this example from the national service supporting families affected by substance use) when families grappling with problem substance use in their midst come together to support each other, though evaluations of such initiatives are rare.

For conventional treatment studies, substance use and related harm are the prime yardsticks of effectiveness, but for this row in the matrix less conventional measures come to the fore including crime, need for child care proceedings, and how well families affected by problem substance use are coping.

Where should I start? A thought-provoking starting point is guidance listed above from the World Health Organization (WHO) on treating problem-drinking prisoners. Though international, it was drafted by a team from Scotland and drew extensively on UK experience, so doubles as a good-practice guide for the UK.

The publication’s cover ( illustration right) poses the key dilemmas. On it we find side by side an optimistic subtitle (“An opportunity for intervention”) seemingly belied by a forbidding concrete wall topped with barbed wire. How could such an environment offer “opportunity” for productive intervention, and even if it did, would the benefits persist beyond the highly controlled and atypical environment created by the walls and the wire? And yet of course, the same walls should create the alcohol-free (and in practice less successfully, drug-free) ‘dry space’ within which productive intervention seems feasible. Discussed below, to a degree the same contradiction is found across the criminal justice system.

In recognising that prison “can be both a help and a hindrance”, the guidance acknowledges the dilemma. Though prison “enforces an environment of abstinence”, this is “however, artificial and does not … enable prisoners to practise their newly acquired knowledge about drinking in moderation or coping skills for preventing relapse”.

In your opinion, how well does WHO’s guidance address this core issue? You might also test its suitably tentative recommendations against the evidence presented in this cell. For example, considered promising (but mainly on the basis of non-prison work) were the brief interventions to which row 1 of the matrix is devoted. Results from the sole randomised evaluation in a prison setting cited in the guidelines were not entirely negative, but overall unconvincing. The same can be said of brief interventions in prisons in general, for which a review listed above uncovered insufficient studies to be able ascertain efficacy or effectiveness, while “questionable” methodological rigour undermined such studies as there were. Within the UK, for brief interventions little evidence of effectiveness can be found from criminal justice studies outside prison conducted in England (listed above; more in “Highlighted study” section below) and Scotland (listed above).

Highlighted study WHO’s guidance listed above on dealing with alcohol problems in prison (discussed in the previous section) is not alone in considering brief interventions promising for problem-drinking offenders. Drawing on guidance from outside the criminal justice sector, Britain’s National Offender Management Service also considered (listed above) these suitable for non-dependent problem drinkers.

Neither sets of guidance authors would have had available to them the final published results of the government-funded SIPS alcohol brief interventions trials in England discussed in cell A1. Listed above, one of the three sub-studies was set in 20 probation offices, by far the largest UK randomised trial of alcohol advice or counselling for offenders. Results were similar to those from GPs’ surgeries and emergency departments: there were no great differences between how well the screening methods identified risky drinkers, nor were there in drinking reductions after three interventions of varying intensity. The interventions ranged from a straightforward and very terse warning not intended to be a brief intervention at all, to a five-minute brief intervention and most extensively, an additional 20 minutes of counselling at (in the probation arm) a further appointment with a specialist alcohol worker. In probation as in other settings, these two brief intervention formats recommended (listed above) by Britain’s National Offender Management Service (NOMS) were no more effective at reducing drinking than a straightforward warning.

However, the probation arm of the SIPS study did throw up a partial exception to the failure of the brief interventions to improve on a simple warning. Police records revealed that over the next 12 months offenders allocated to either of the brief interventions were significantly less likely to be reconvicted (36% and 38% v. 50%) than those given only the straightforward warning. With so many outcomes tested, this could have been a chance finding. Given no correspondingly greater impacts on drinking, the authors themselves queried whether the findings reflected a real effect. Assuming they did, it seems possible that more intense drink problems among offenders than among patients in the primary care and emergency department arms of the study afforded greater scope for them to respond better to the brief interventions, which (unlike the straightforward warning) both addressed the risks of offending while under the influence of drink. Remember though, that without a no-intervention comparator, there is no way of knowing whether any of the three interventions were better than doing nothing at all.

Are the SIPS results enough to overturn official support for brief interventions?

Are the results of the SIPS studies – intended after all to inform UK government policy – enough for you to overturn NOMS’s recommendations on brief interventions for offenders and to revert to a simple, no-training-needed, ‘Don’t drink; it’s bad for you’ warning? To be precise, the SIPS trials were generally unable to reject the possibility that research- and theory-informed brief interventions worked no better than such a warning. That is not the same as saying the trials proved they were equivalent; there might have been a difference and one may still be found in other trials. Would abandoning brief interventions for offenders on this basis risk throwing the baby out as well as the bathwater, especially since there was that glimmer of hope from reconviction data in the probation arm of the study? Or perhaps they show that with relatively severe problems among offenders, brief interventions are insufficient and we should escalate to fully fledged therapy? The problem with that, is that therapy too has an unconvincing record – the subject of the next section.

Issues to consider and discuss

Why is the record so poor? Look through the studies in this cell and you might spot an unfortunate trend in the criminal justice studies. It starts with the seminal US study from the 1960s listed above. Assuming arrests reflect crime, this found that ordering “chronic drunk arrestees” to 12-step mutual aid groups or to treatment were in crime-reduction terms at best ineffective and possibly counterproductive; hover here to absorb the researchers’ shock at this finding and their speculations about the reasons.

That was no US-only aberration nor one limited to the ’60s. Working through the British record to date accumulates little or no evidence for court-ordered treatment (listed above), the popular cognitive-behavioural family of criminal justice interventions (1, listed above; 2; listed above), for brief counselling (listed above) of arrestees under the influence of drink, for brief interventions (listed above) with young convicted offenders, for brief counselling (listed above) as opposed to merely giving a health warning to heavy-drinking offenders on probation, for brief interventions (listed above) in general across the criminal justice system, for abstinence-oriented prison treatment (listed above), or for treatment in prison in general (listed above).

Two studies presented as exceptions to the overall poor UK record in fact reinforced it

The last two studies in this list warrant closer examination. Both were presented not as confirming the overall poor UK record, but as exceptions to the rule, exceptions which the strength of the studies means must be taken seriously. Both of prison treatment, they were large scale and carefully controlled, taking the non-randomised, real-life comparisons they made of treatment versus no treatment as close as the data would permit towards the level playing field guaranteed by effective randomisation. Overall neither seems to have found treatment effective, but in both cases a positive gloss was placed on the findings, reinforcing the role given to criminal justice-based treatment in national strategy. In fact, in both cases this inadequately founded optimism was based on failing to compare like with like – admittedly an extremely difficult thing to do without randomisation. To its credit, having been prepared to expose its prison programme to independent scrutiny by submitting data to the Ministry of Justice’s ‘data lab’, after reading our draft commentary the Forward Trust withdrew its claim that the findings supported the programme’s “proven” effectiveness. Now the study is more neutrally reported. The supplementary text (click to unfold Unfold supplementary text ) examines the original claim which seems to have stood unchallenged for two-and-a-half years, as well as the positive gloss put by the researchers themselves on the findings of a still larger-scale UK prison study. A rigorously critical stance to claims such as these is not a negative response, but demanded (1 2) by a scientific approach to testing effectiveness. Reading the supplementary text might seem to justify a degree of scepticism over future claims.

Close supplementary text

Forward Trust programme – proven or not?

First let’s dissect the findings and the presentation of findings deriving from an evaluation listed above of an intensive six-week programme run by the Forward Trust. An abstinence-based programme incorporating elements from AA, it was undertaken by 435 men who participated in 117 courses hosted by nine prisons in England and Wales. The aim was to improve outcomes for participants and also to reduce reoffending, but one-year post-release reoffending records of prisoners who participated yielded no reliable indication of any such effect. On key measures their records were virtually identical to those of a matched comparison set of 254,501 prisoners who did not go through the evaluated programme, most of whom will have received no treatment at all. How the figures were interpreted hinged on the perceived adequacy of this matching.

Only randomisation or an equivalent procedure can match both known and unknown characteristics which might affect the impact of an intervention. However, on known characteristics prisoners who received Forward Trust’s programme and their comparators were extraordinarily well matched, though the researchers were modest: “Most variables were well matched.” Matching was based on “offender demographics, criminal history and the following risks and needs: accommodation, employment history, education, family and relationships, drug and alcohol use, mental health, thinking skills and attitudes towards offending” – a fairly comprehensive and relevant set of variables which included indicators of the severity of offending and drinking problems and motivation to address these.

Nevertheless, in the evaluation report the Forward Trust argued that in fact the comparison group was inadequate because there was no information on whether they were actually dependent on alcohol, yet all the treated group had to have been assessed as such in order to get on the programme. In any real-life comparison there will be gaps; on the basis of their experience and their review of the literature, the Trust saw this as an important one. Though by no means eliminated, this concern is lessened by the fact that the data available for both treatment and comparison groups seemed a good proxy for dependence. It included the “frequency and level of alcohol use” and past and current “significant problems” with alcohol or “binge drinking”, plus “problems with motivation to tackle alcohol misuse”. For example, 51% of both sets of prisoners were noted to have significant problems with current use of alcohol and 59% and 58% with “binge drinking”; other indicators of drinking problems were also the same or differed by just 1%. If, for example, the treated group were all dependent but an appreciable proportion of the comparison group were instead episodic heavy (or ‘binge’) drinkers, we would have expected these figures to differ.

Even if a matching on dependence might have altered the picture, there was no getting away from the failure of the available comparison to validate the programme. For the researchers, the things “you cannot say” on the basis of their findings included “that the Forward Trust [programme] decreases the frequency of reoffending”. Far from claiming this, in the face of slightly worse figures for the programme’s clients, they felt the need to caution that it would also be wrong to claim the programme was counterproductive – that it “increases the one-year proven reoffending rate of its participants”. In the report itself, the Trust agreed: “the results are unable to say whether our [programme] has an impact on reoffending”.

In contrast, on their web site the Trust had headed their report on the study, “Forward’s alcohol treatment programme proven to reduce re-offending”. In a page now withdrawn, the justification was the evaluation’s finding that 37% of programme clients reoffended in the year after release, “significantly lower than predicted re-offending rates for alcohol dependent offenders who do not get access to treatment”. The Trust did issue a caveat: “we cannot claim ‘statistical validity’ for the [programme’s] impact on re-offending, due to technical challenges in adequately matching the characteristics of our programme clients with a control group of offenders … any comparison group must also be shown to be alcohol dependent, to allow for a like-for-like comparison”.

In justifying the “proven” claim, reference was made to a “64% re-offending rate for these (sic) sentenced to less than a year in prison” – a peripheral comparison, as 95% of the evaluation’s sample were sentenced for longer. Greater reliance was it seems placed on a study published in 2013 which did present reoffending rates for a presumed substance-dependent set of prisoners. According to the Trust, it found that “a cohort attending low intensity drug or alcohol treatment programmes (and therefore likely to have presented with dependency) is associated with a one-year re-offending rate of 58%”. This comparison with another study abandoned the evaluation’s great strength – its careful matching of offenders who did versus did not go through the Trust’s programme to as far as possible compare like with like. In fact, it introduced new and serious mismatches.

The comparator study limited its sample to offenders sentenced to up to four years, and noted that longer sentences are generally associated with lower proven re-offending rates. However, over a third of the evaluation’s sample were sentenced to four years or more, giving them a head start in registering reduced reoffending. The 58% reoffending figure against which the Trust’s programme was compared concerned prisoners who had been through either alcohol or drug treatment programmes in prison, and derived from a sample who in the four weeks prior to sentence had used drugs or drank daily. In the full sample, 64% were according to these criteria problem drug users and 22% problem drinkers, so it seems likely that those from whom the 58% figure was derived were mainly treated for drug problems, while the evaluation of the Trust’s programme concerned only treatment for drinking problems. If as the Trust said, “any comparison group must also be shown to be alcohol dependent”, this sample did not qualify.

Then there was the time period. The comparator study’s sample would mostly have been released by the end of 2008, but the evaluation of the Trust’s programme sampled offenders released during 2013–14. Finally, there was the prisons – nine in the evaluation, but it seems all prisons in the comparator study. Like was very definitely not being compared with like. If that was the problem with the evaluation report which did not find the Trust’s programme effective, it was even more so with the Trust’s comparisons which it said proved the programme was effective.

The claim of a proven anti-offending impact also ignored the fact that in the evaluation, prisoners who did not go through the programme (and probably – see below – went through none at all while in prison) did just as well, undermining the interpretation that the programme was a causal agent in reducing reoffending.

Based on the scientific literature they cite, the Trust’s revised commentary on the study continues to advance (but in an appropriately tentative manner) the “working hypothesis” that dependence on alcohol is associated with an elevated chance of a return to crime, and that therefore their treated sample of prisoners may have done relatively well to restrict the proportion known to have reoffended to just over a third.

Did treatment work with high-risk, adequately treated prisoners?

At larger scale still was a study listed above of all adult prisoners in England assessed by their offender manager as having an alcohol use disorder and released during 2013–14. It compared the post-release offending of those who while in prison had been treated for that disorder versus those who had not – effectively assessing the national effort to prevent alcohol-related recidivism among prisoners through any intervention which the national monitoring system saw as ‘treatment’, offered in response to what the criminal justice system had identified as severe and/or criminogenic alcohol use.

An incidental finding was how few prisoners identified as having serious drinking problem were recorded as having received treatment before they left prison – just 2,647 or 10% of the 26,654 who left in 2013–14. At issue was whether this treatment had helped prevent offending among the 1 in 10 who received it. Overall, it seemed not – “seemed”, because no test was made of whether across all treated prisoners the proportion who reoffended in the year after release differed to a statistically significant degree from among untreated prisoners. In fact, the article does not even present the total number or proportion of treated prisoners who reoffended. However, it can be calculated: 1,155 of the 2,647 or 43.6% reoffended, virtually identical to the 43.7% among untreated prisoners. It remains possible that adjustment for differences in the risk of offending would yield a statistically significant gap between the two sets of prisoners – but given their similarity, this seems highly unlikely.

Instead of assessing treatment as a whole, the analysis divided treated prisoners into those just prescribed medications versus those who instead or as well were recorded as having some kind of ‘talking’ or psychosocial therapy, the latter constituting 1,914 of the 2,647 treated patients. The psychosocial therapy set was then further subdivided into the 241 – just 9% of all those treated – whose treatment was judged by the researchers to have embodied ‘risk-need-responsivity’ principles (free source at the time of writing) considered to underpin effective treatment in criminal justice settings, versus the remainder. Components of that judgement were that the 9% were classified as at high risk of re-offending, and given this risk had received appropriate treatment in the form of at least one of five “evidence-based” psychosocial interventions, during a treatment episode which was among the longest third recorded. In the light of the evidence presented in this cell, it can be argued whether the psychosocial interventions (which included unspecified “counselling”) truly were “evidence-based” for a prison population, but for the current purposes the key criterion was the duration of the treatment episode.

Out of what was almost certainly no evidence of a reoffending payoff from treatment overall, the researchers nevertheless extracted something positive. Of the 241 high-risk prisoners who underwent treatment thought to embody risk-need-responsivity principles, about 37% reoffended over the year after their release – about 7% fewer than the 44% among the untreated control group. Four methods were used to adjust this gap for differences in the characteristics of these two sets of offenders, of which three proved viable attempts to approximate the ‘level playing field’ which would have been generated by randomisation. What according to the analysts was the most robust and suitable of these methods left the comparison offenders’ reoffending rate at about 44%, but uprated those of the small set of ‘risk-need-responsivity’ offenders (the analysts labelled their treatment “RNR compliant”) to nearly 42%, reducing a gap of just over 7% to just below 2%. Unsurprisingly this difference was not statistically significant, so chance variation could not be ruled out and a real effect could not be claimed. Similarly, neither of the other two methods yielded a statistically significant difference, one widening the gap, the other reducing it.

Nevertheless, the consistency with which ‘risk-need-responsivity’ offenders registered the lowest reoffending rate was enough for the article’s abstract to say, “The outcomes for RNR compliant treatment suggest a lower recidivism rate compared to the control group.” Interpreting their results in the body of the report, the analysts said it was “possible to establish that RNR compliant approaches reported lower offending rates than the control group”, leading to a recommendation to incorporate “RNR principles” into an integrated public health and criminal justice-orientated treatment system.

It would have been equally – and arguably more – justifiable to conclude that the study yielded no support for risk-need-responsivity principles or for the psychosocial treatments considered to have been the vehicle for implementing these principles, the type of conclusion commonly reached when results fail to meet the test of statistical significance intended to sort real effects from chance. Even the existence of a real gap – statistically significant or not – in adjusted offending rates is questionable. After taking into account differences between risk-need-responsivity offenders and their untreated comparators, the most robust estimation method cut the reoffending advantage of the former from 7% to 2%. It seems possible this gap would have disappeared altogether had a more complete adjustment been possible.

One possibly very important factor could not be fully adjusted for – the fact that the risk-need-responsivity offenders were selected to be those who had lasted the longest on their treatment programmes. No such selection could be made for the comparison group, because they were not treated. It meant like was not being compared with like, raising the possibility that a set of offenders selected to be particularly compliant, keen on overcoming their drinking problem, and prepared to stick with their treatment, were being compared with a set not selected for these promising features. If this were the case, these attributes rather than the treatment they received could be the reason why fewer reoffended. Consistent with this interpretation is that the only valid statistically significant difference between the offending rate of these offenders and the other sets of prisoners was in comparison to the higher rate of those who were treated, but only with medications. It seems possible that an opposite selection process was happening, meaning that these offenders were particularly unpromising – that contrary to guidelines, no therapy was offered despite their being known to be problem drinkers and needing medical care, because none was sought or it was rejected by prisoners relatively uninterested in tackling their drinking and/or their criminality. Possibly such a relatively high proportion went on to reoffend because of this disposition, not because of the inadequacy of the treatment.

In general the division of the treated offenders into those who underwent different forms of treatment risked introducing further unknown differences between these subdivisions and the untreated control group. The 10% who were treated at all may already have been atypical in some way, but yet more atypical would have been the “RNR compliant” group, who formed under 1% of all released problem-drinking offenders. The safest comparison would seem to be between all treated and all untreated problem-drinking offenders – the one that led to virtually identical re-offending rates, not the most encouraging of findings.

Close supplementary text

The poor record of therapy and treatment interventions in the criminal justice system is not unique to the UK, but characterises (review listed above) the international literature. The review which came to this conclusion speculated that the reasons might include a mismatch between the typically high level of need of prisoners and the intensity of extensity of intervention programmes. In particular, “Few interventions followed an individual as they progressed through the [criminal justice system], and even fewer followed them for a considerable period of time. This can create a ‘cliff-edge of support’ on release.” The lack of proven beneficial effect for men was particularly disappointing since they form the vast majority of people whose drinking has contributed to their criminal justice involvement.

Regardless of whether it produces positive or negative findings, rigorous research of any kind is in short supply. In its search for studies published from 2000 to 2019, the UK review cited in the previous paragraph found just three conducted in the UK which reported on alcohol-related outcomes and had a comparison group against whom to benchmark the impact of the evaluated intervention. When a few years before (in 2014) the UK’s Ministry of Justice looked at the evidence, they found “insufficient … to determine the impact on reoffending of alcohol treatment for offenders.” In 2011 a scoping study listed above for the Scottish health service came to similar conclusions: “in the criminal justice setting … there is limited evidence that explores the suitability or effectiveness of alcohol interventions or treatment of any kind”.

There is a relatively bright spot from trials listed above of disulfiram, the drug which causes deterrent physical reactions if someone drinks. Allied with legal pressure to take it, disulfiram seemed to help repeat alcohol-related offenders become abstinent in the USA and England, but in both cases there was no control group against whom to benchmark the results. When this is missing from a study, we cannot know whether the findings were due to the intervention, or would have happened anyway.

Rather than the interventions being ineffective, perhaps absence of evidence reflects the relative absence of research which could produce this evidence – certainly possible, for example, in respect of court-ordered treatment (documents listed above) in the UK. The contradictions of helping and punishing at the same time might undermine treatment Or is it that the contradictions of helping and punishing at the same time (‘motivational arm-twisting’ as we termed it) undermine interventions which might work in other settings? Conceivably, adding a coercive element to treatment makes it seem to the ‘patient’ less like treatment, and they respond less like a patient, reducing effectiveness. That was the clear but not entirely certain implication of a review of evaluations of interventions for offenders, which graded them on the dimension of voluntariness versus coercion. It discovered that the impact of treatment increased in line with the degree to which the offender was free to choose the treatment. Could this be why the record seems stronger for the impact of voluntarily-entered generic treatment programmes than for treatment aimed at offenders (1, listed above; 2, listed above; 3, listed above)? Instead of extending the net of legally coerced treatment, should we seek to maximise the attractiveness and availability of treatment in general so that a higher proportion of offenders choose this rather than being coerced in to it? Or is this an unrealistic ideal which would miss the unmotivated and could never reach enough offenders whose drinking is affecting not just them, but their families and the rest of society?

Research is motivated; and it matters The sudden death (see panel in the linked analysis) of a researcher whose work informed UK guidance after alleged fabrication of research data offers an extreme illustration of the fact that like every other human initiative, research is a motivated endeavour. Nothing entailing this deliberation and effort is undertaken and completed without emotion and motivation to move it. The illustration is unpacked here because the work it cast a shadow over has substantially influenced understandings of a major theme for this cell – which treatments best protect families and children.

Former professor William Fals-Stewart was found dead at his home on 23 February 2010 shortly after his arrest on charges arising from an accusation of scientific misconduct made in 2004 (1 2 3 4). With no evidence of foul play, suicide was a possibility. The accusation related to the alleged fabrication of data in studies undertaken as an employee at the University at Buffalo and Research Institute on Addictions, including his claim of having studied over 200 subjects when there were consent forms for only about 50. The studies were funded by the US National Institute on Drug Abuse, and the fabrication was thought to have been an attempt to keep the funding coming.

However, the arrest was not in relation to the misconduct itself, but to his hiring actors to give false testimony during an investigation of the accusations conducted by the university in 2007. The actors helped Fals-Stewart escape a misconduct verdict due to “insufficient evidence”, but still he was “forced out of his position at the university”. He then sued for wrongful termination, seeking 4 million dollars compensation. While defending the suit, New York State’s legal team discovered that the ‘grant administrators’ who had testified over the phone to the misconduct hearings were professional actors, fed scripts “riddled with inaccuracies regarding his research”. Confronted with this evidence, in 2009 Dr Fals-Stewart withdrew his lawsuit. On 16 February 2010 he was arrested for charges in relation to the attempt to extract money from the university and the state which risked 15 years in prison; a few days later, at age 48 he was found dead.

The deceased was a leading researcher on behavioural couples therapy, the treatment which, if we believe the research, has the most solid evidence base (review listed above) in respect of curbing crime in the home and improving the lives of families affected by problem drinking. He was an author on five of the couples-therapy reports listed in this cell (1, listed above; 2, listed above; 3, listed above; 4, listed above; 5, listed above), and his considerable contribution to that literature in these and other studies will have fed through to reviews and to guidelines which recommend the form of couples therapy (behavioural couples therapy) he was associated with. Among these are guidelines from the UK’s National Institute for Health and Care Excellence, the nation’s most authoritative and influential source of such advice.

There is no specific reason to doubt the research record of behavioural couples therapy overall. But even without the shadow cast by the Fals-Stewart case, it is a concern that most of the studies have been led by the developers of the programmes being evaluated, and that when they have not, though still positive, results have been less convincing. In general, we can have more faith in findings when the researcher has no interest in validating versus invalidating the intervention being tested – in particular, when it is not ‘their’ intervention – than when their reputation, self-esteem, and/or career and income, could hinge on the results.

It follows that we can have even more faith when the results go against their desires and/or expectations, as in the seminal US study listed above, which its creators expected to confirm their earlier observations that court-ordered alcohol treatment cut crime, and the British SIPS trials (discussed above), which failed to confirm expectations (document listed above) that scientific, theory-based counselling would be more effective than a simple health warning. In both the shock was delivered by the ‘randomised controlled trial’ format. It reminds us that done well – and with integrity – this entails the researcher engineering a level playing field and ensuring they can have no hand in which intervention option comes out on top, meaning it can deliver results which force them to think again.

Can it ever be safe to leave children with dependent drinkers? In England in 2014/15 roughly 120,000 alcohol-dependent adults had perhaps 200,000 children living with them in the household. In Scotland in 2008–10 between 36,000 and 51,000 children were estimated to be living with alcohol-dependent parents or guardians. Is the risk of relapse and with it the risk to the children simply too great? Is it simply too risky to leave their children with the most severely affected of these parents, even if they are in treatment, and even if they appear to have successfully completed it? If substance dependence at least behaves like a chronic relapsing condition, and even if that is only broadly valid for treatment populations with their typically low de-addiction resources, relapse is to be expected after treatment, and with it, renewed risk to the child. In recent years in England, about a third of the treatment caseload have been returning to treatment, presumably having relapsed after their previous treatments.

But what if, as well as treatment of dependence, intensive resources were targeted at strengthening the family and improving parenting – an expert family therapist available 24 hours a day, seven days a week, even if only over four to six weeks? This kind of specialist ‘family preservation’ service has been tried and evaluated in Wales (1 listed above; 2 listed above) and in Middlesbrough. Independent researchers found the services prevented the need to permanently place children in care, and reduced time spent in temporary placements. Crucially, over a follow-up period averaging five to six years, one of the Welsh studies listed above was able directly to confirm that reduced resort to care had not been at the expense of the children’s welfare; there was no indication that the service had inadvertently harmed children by helping keep them with their families chart.

The results were convincing enough for the Welsh government to roll out similar services across Wales, initially concentrating on families where there is parental substance misuse and concern over child welfare. Regulations stipulated that teams providing the services must consist of at least five professionals including a social worker, nurse and health visitor, perhaps an attempt to address the need to maintain quality, one highlighted (document listed above) by researchers. An evaluation listed above concluded that at three first-phase sites, the new schemes “appeared to improve short-term outcomes for a good number of families”, though, staff felt, less so for a few with “very chaotic lives and serious multiple issues”. There was evidence directly from the families that their lives and those of the children had improved, but with no comparison group, it is unclear whether the interventions were the cause of the improvements.

Do you find the results to date that convincing? Remember that the parents in the initial trials were not dabblers in drink or drugs, but had problems serious enough to take them to the brink of losing care of their children. Is the risk of relapse to dependent substance use and with it the risk to the children simply too great? Or is the greater risk to unnecessarily blight children’s lives by taking them in to care? Of course, these decisions must be made case-by-case, but still on the basis of an understanding of the general and likely balance of risk and benefit. To help you work through the issues and for more background, read our hot topic on protecting the children of problem substance users.

An alternative is to get the family courts involved (what the family preservation services try to make unnecessary, saving costs), but to use the court’s powers to collaboratively arrange intensive treatment and support from a specialist team allied with the court, and to judicially monitor parents’ progress while the children are under the care of court-appointed guardians. Like other courts specialising in substance use, these ‘family drug and alcohol courts’ reinforce the treatment process with the leverage available to the court. They also provide immediate, legally-backed safeguards for the children. On the evidence listed above from the UK, they seem to have greater success than usual care proceedings in achieving long-term family stability and keeping more children with their parents. In 2019 the UK government announced funding under the “Supporting Families: Investing in Practice” banner to support and evaluate 14 such courts.

It is worth finishing this section with a reminder that in itself, successfully treating substance use problems is likely to improve child welfare because it reduces some manifestations of conflict and violence in the family (1, free source at the time of writing, listed above; 2 listed above) and makes competent parenting more possible. Treatment services have, however, gone further, offering specific parenting and family support, potentially forestalling the need for the more intensive services discussed above intended to help families already at the brink of losing care of their children. Before they reach that point, there seems a strong case (review listed above) for making parenting and child welfare support available to all parents in treatment for problem substance use. Because such support is not predicated on discovering, or the parent admitting to, shortcomings in their child’s upbringing and welfare (which many will be reluctant to do), these programmes can reach families in need of help who would otherwise be missed or feel stigmatised, and can reduce the numbers who reach the point reached by families referred to intensive family rescue services. Examples are given in our hot topic on protecting children, and researchers based in the UK who specialise in substance misuse in families have offered recommendations along these lines for addiction treatment services based on a review of the international literature.

Thanks for their comments on this entry in draft to the Forward Trust based in England. Commentators bear no responsibility for the text including the interpretations and any remaining errors.

Close Matrix Bite guide and commentary Close Matrix Bite