2019/20 update funded by

Alcohol Change UK web site. Opens new Window

Alcohol Change UK



Previously also funded by

Society for the Study of Addiction web site Society for the Study of Addiction

Developed with

Skills Consortium web site. Opens new window

Alcohol Treatment Matrix

Effectiveness Bank Alcohol Treatment Matrix

Includes brief interventions

Drug Treatment Matrix

Effectiveness Bank Drug Treatment Matrix

Includes harm reduction

About the matrices

Effectiveness Bank matrices

Articles, slides, and video


Comment/query

Send email for updates



Effectiveness bank home page. Opens new window Matrix cell

Alcohol Treatment Matrix cell A3 Matrix cell logo

Interventions; Medical treatment

Seminal and key studies and reviews on the effectiveness of medical interventions and treatment in medical settings.


S Seminal studiesK Key studiesR ReviewsG Guidancemore Search for more studies

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

S Handing patients responsibility matches extended treatment plus medications (1977). Describes mould-breaking study in London which found a brief session handing responsibility to the couple to cure the husband’s alcoholism worked as well as extended treatment, which generally involved medications. See also this retrospective (2015) from a study researcher. Similar Scottish study later found (1988) more evidence for extended treatment. Discussed in bite’s Where should I start? section.

S Disulfiram only works with compliant patients (1986). Explains that despite overall negative findings, the first rigorous trial of the drug which causes deterrent physical reactions after drinking found that some older and more socially stable disulfiram patients who completed the study did drink less frequently after lapsing. Later the study was seen as confirming the need to supervise the drug’s administration to ensure more patients took more of the pills. Discussion in bite’s Issues section.

S Impressive results from first clinical trial of acamprosate (1985). Three months later the success rate among alcohol-dependent patients detoxified and discharged from a French inpatient unit was 61% if they had been randomly allocated to acamprosate versus 32% on placebo. It was a notable result among severely dependent patients with a record of failed treatments.

S Benzodiazepines best withdrawal treatment (1969). Study which clarified the dangerous confusion over how to prevent the life-threatening complications of alcohol withdrawal.

K Acamprosate unsuitable for some UK patients (2000). Despite positive findings elsewhere, large UK trial found acamprosate did not curb relapse among detoxified alcoholics. The findings highlighted the importance of an accompanying support programme to help keep patients in treatment and taking the pills, and also perhaps the type of patients – steady drinkers rather than the study’s ‘bingers’ – who respond best to the treatment. Discussion in bite’s Highlighted study section.

K Supervised disulfiram works in Britain (1992). In the major UK trial, disulfiram significantly reduced drinking relative to a vitamin pill placebo by nearly 10 UK units a day. The researchers thought their results showed the importance of supervising consumption and making patients aware of the potential consequences of drinking while taking disulfiram. Discussion in bite’s Highlighted study section.

K Naltrexone can’t work unless patients take the pills (2000). In conditions typical of British alcohol treatment clinics, when taken as directed naltrexone reduced post-detoxification drinking – but nearly 60% of patients randomly allocated to naltrexone versus a placebo left treatment early and stopped taking the pills. Discussion in bite’s Highlighted study section.

K Naltrexone boosts primary care treatment (2006). Within the same trial, the large US ‘COMBINE’ study tested acamprosate, naltrexone, and both together against a placebo, and whether adding psychological therapy boosted results from medical care. Therapy elevated placebo drinking outcomes to match the most effective medications; without therapy, naltrexone elevated outcomes to about the same degree. Results supported prescribing naltrexone in primary care-based treatment with relatively compliant patients. Discussion in bite’s Issues section.

R The power of the placebo (2013). Across relevant trials, end-of-treatment improvements in alcohol patients randomly allocated to an inactive placebo on average dwarfed estimates ( reviews below) of additional benefits due to the pill being an active medication. Discussion in bite’s Issues section.

R Evidence strongest for acamprosate and oral naltrexone (2014). Amalgamates findings from 123 trials of the full range of medications prescribed for at least 12 weeks. Evidence was strongest for acamprosate and oral naltrexone; each prevented 1 patient in 12 or 20 respectively returning to drinking. Naltrexone also prevented 1 in 12 returning to heavy drinking. Health improvements remained unclear. The few head-to-head comparisons yielded no significant differences between the two medications but other reviews amalgamating all relevant data have found minor differences.

R Prescribing in primary care and general medicine (2011). Based on studies offering minimal psychosocial support, recommends oral naltrexone, topiramate or (with abstinent patients) acamprosate – and given supervised consumption and motivated, abstinent patients, also disulfiram. Medication should be accompanied by brief support to promote compliance with treatment. Even when little other support was provided, improvements due to medication were “modest”.

R Who benefits most from naltrexone versus acamprosate? (2013). Amalgamated findings from randomised trials comparing the two drugs or (more often) comparing each to a placebo indicated that on average naltrexone was best for patients who want to reduce heavy drinking, acamprosate for those seeking abstinence. A different kind of analysis (2014) confirmed these conclusions, and another (2015) found no difference in impacts on drinking between European and (mainly) US caseloads. All three reviews found differences slight as were the effects of each medication compared to placebo, and individual trials often failed to find benefits.

R Disulfiram needs supervised consumption and patient awareness (2014). Amalgamated research findings indicate that disulfiram substantially improves on alternatives or placebo when (and on average, only when) compliance is bolstered by supervising administration and patients know they are taking a drug which causes unpleasant reactions if they drink. With the careful patient selection typical of research trials, disulfiram has not been associated with excess deaths or serious adverse events. Different kind of analysis (2011) agreed that given well supervised consumption, disulfiram is more effective than the alternatives. Discussion in bite’s Issues section.

R How to encourage patients to take the pills (2004). Because the reasons why alcohol treatment patients skip their medication are varied, so too must be ways to address this, from reducing side-effects and adjusting dose to compliance-enhancing counselling and enrolling the family. Discussion in bite’s Issues section.

R Benzodiazepines make withdrawal safer and easier (2010). Rigorous review and synthesis of randomised trials indicates the superiority of benzodiazepines for controlling the potentially serious medical consequences of withdrawing from dependent drinking, especially seizures.

G Official English guidelines on treating harmful drinking and alcohol dependence (National Institute for Health and Care Excellence, 2011). Britain’s gatekeeper to the public provision of health care technologies recommends considering acamprosate or naltrexone after withdrawal, but relegates disulfiram to a second-line option. Treatment and care should take into account the patient’s individual needs and preferences. Discussion of disulfiram recommendations in bite’s Issues section. See also Scottish primary care guidelines.

G Treating withdrawal (Royal College of Physicians, 2010). Guidance developed for Britain’s National Institute for Health and Care Excellence on medical care of patients suffering acute withdrawal from alcohol or other alcohol-related medical conditions.

G US consensus clinical guidelines ([US] Substance Abuse and Mental Health Services Administration, 2009). From experts convened by the US health department, how US-approved medications (acamprosate, oral and injectable naltrexone, and disulfiram) can be incorporated in to medical practice, including choosing suitable patients.

G Treating substance use service clients with mental health problems ([Australian] National Drug and Alcohol Research Centre, 2016). Funded by the Australian government; recommends services screen all patients for mental health problems and that mental illness should not be a barrier to treating substance use problems. Research shows these patients can benefit as much as others from routine treatments for problem drinking.

more This search retrieves all relevant analyses.
For subtopics go to the subject search page.

Matrix Bite Open Matrix Bite guide to this cell Open Matrix Bite guide and commentary

Close Matrix Bite guide and commentary Close Matrix Bite Close Matrix Bite guide and commentary


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

What is this cell about? About the treatment of alcohol dependence in a medical context and/or involving medical care, typically by GPs or by alcohol treatment or psychiatric units in hospitals. Clinical staff are responsible for medications, so the centrality of these to an intervention distinguishes it most clearly as medical. Drugs (primarily benzodiazepines) help patients withdraw from alcohol more comfortably and safely, or (primarily in the UK acamprosate, but also naltrexone and disulfiram) are intended to sustain longer term abstinence or moderate drinking.

But drugs are never all there is to medical care. They help to forge a relatively intoxication-free space during which patients can lose the habit of regular drinking and be supported to find other ways to cope and construct lives incompatible with a return to heavy drinking. Drug-based treatments also include potentially therapeutic interactions with clinical and other staff, and ways to encourage patients to take medications. Sometimes this means enlisting family and others, deepening their involvement in the patient’s recovery. Without these factors there is a risk that any benefits of medication will be lost once the treatment ends.

Though usage has been increasing, in Britain treatment for alcohol dependence usually consists entirely of advice and support. Drugs are almost universally used to ease withdrawal in inpatient units, but in 2013/14 in England, of the 101,782 drinkers treated in non-residential community settings, just 16% were prescribed a medication. As we’ll see later in this bite, relegation of medications to a minority option accurately reflects their generally minor effects relative to the other influences which together constitute the ‘placebo effect’ seen in trials. For this ‘disease’, medications usually add little (but on average, do add a little) to the patient’s impetus to get better, the processes in their life which help them realise and sustain this ambition, and the impact of deciding to enter and get actively engaged in treatment, one manifestation of which is regularly taking medication.

Where should I start? Exemplifying that last point, and helping place medication in context, is a seminal study from England which questioned the orthodoxy that alcoholism requires intensive treatment. Griffith Edwards and colleagues found that male alcoholics (accompanied by their wives) seen by a psychiatrist-led team at an alcohol clinic did as well after a single brief session as after fully fledged treatment, which for about two-thirds of patients included medications in the form of the disulfiram-like drug calcium cyanamide. Looking back over nearly 40 years, a researcher on the study interpreted the results as meaning that formal treatment had been less important than aspects of the process shared by both sets of couples: “all the negotiations and arguments that must have gone on between husband and wife prior to and after the visit to the [GP]; the referral to a psychiatric hospital and the wait for the appointment letter; the whole morning spent going over one’s drinking and one’s marriage with a group of expert strangers; the unequivocal advice, delivered in the presence of one’s spouse; ... knowing that the hospital was keeping a watching brief and that questions were being asked about your behaviour every month and that you would be asked to account for yourself at the year’s end.”

A similar study in Scotland included patients randomly allocated to just five minutes of advice or up to one hour, as well as to extended treatment, which in this case rarely involved medications. Two years later 58% of the extended treatment patients were abstinent or trouble-free drinkers compared to 39% of the advice groups, a difference which verged on the statistically significant. It seemed to be largely due to significantly greater reduction in alcohol-related problems rather than drinking itself, which was substantially reduced in both advice and extended-treatment patients. For several reasons the advantage gained by extended treatment was probably greater than the results suggest, but still the improvements in patients given the briefest of interventions was considerable. Speculating on the causes, the researchers highlighted the “unequivocal diagnosis and ... injunction to abstain” or “the decision of the patient to do something about his drinking” – or both, since the “injunction” explicitly left their recovery up to the patient.

Highlighted study Jonathan Chick led the Scottish study summarised above and also our highlighted study. Published in 1992, it remains the solidest UK foundation for prescribing disulfiram. Prescribing patients a medication which causes very unpleasant reactions to alcohol seems a sure-fire deterrent to repeated drinking, but this and other studies showed that even such powerful drugs are reliant on the patient’s motivation and their social support – for disulfiram, especially the relatives and partners who with the patient’s agreement help ensure the pills are taken. Even then, for many patients that will not be enough. The highlighted study gave disulfiram the advantages of supervising consumption (usually done by husband or wife) and the patients’ awareness that they were taking a deterrent drug, yet still nearly half (but no more so on disulfiram) effectively rejected or dropped out of treatment. Nevertheless, over the entire six-month follow-up, disulfiram patients had reduced their drinking days and amounts drunk by significantly and substantially more. In the final month – during which time many patients must have withdrawn from treatment – drinking reductions remained apparent, but were no longer significant.

Even when it was a placebo, there was a striking gap in abstinence between patients who took their pills and those who did not

A striking finding was the huge difference in abstinence rates between patients who largely complied with their treatments and took the pills and those who were less compliant – even when those pills were effectively an inactive placebo. Among the two groups of patients given inactive pills, 50% and 43% of the minority who largely took them were abstinent compared to 9% and 6% of the remainder.

Another study from Dr Chick showed that the same applied to naltrexone, intended to moderate rather than block drinking. With generally low social support and no supervised consumption, before the end of the 12 weeks during which naltrexone or placebo were prescribed nearly 60% of patients had terminated study and treatment early. Despite this degree of loss, across all patients there remained signs that naltrexone had reduced drinking, but the effects were not statistically significant. When the analysis was confined to patients who had completed the study and had largely complied with treatment, there was still no evidence that naltrexone had delayed a return to drinking, but the reduction in the amount subsequently drunk (on average half that in the placebo group) was substantial and statistically significant, seemingly a sign that the experience of drinking on naltrexone was not so ‘moreish’ as it had been without the drug. Dr Chick completed the treble with his acamprosate study. Hampered by high drop-out and non-compliance rates – by the end of the six-month trial fewer than 30% of patients were taking at least 90% of their tablets – it found no significant reductions in drinking relative to a placebo.

Results of these three British studies emphasise measures to keep patients in treatment and taking their medication. Well appreciated for disulfiram, the same applies to less absolutist medications like naltrexone and acamprosate, which the British studies cited in this section suggest may also benefit from supervised consumption. Not only does this help ensure pills are taken, but also more deeply involves family and associates in the patient’s recovery. The next section addresses this issue directly.

Issues to consider and discuss


If ‘compliance’ is crucial, how do we encourage patients to take the pills? No single or best way, concluded a comprehensive review. A basic strategy stressed by UK guidelines is to build a trusting relationship with patients and relate to them in a supportive, empathic and non-judgmental manner. Any qualified practitioner can prescribe anti-drinking medications, but it takes particular skills and qualities to give patients confidence in the treatment, encourage them to take the medication, and to maximise its effects. Improving client-clinician relationships addresses some but not all of the reasons for non-compliance. Patients may skip doses because of side-effects, because they do not feel they have a problem, don’t believe the medication is doing any good, are overcome by cravings, simply forget, are disorganised (perhaps by intoxication), or lack the right kind of support from family or associates – and so on!

Notice that some of these reasons may be ‘legitimate’ – and not just from the patient’s point of view. Side-effects can be distressing or life-diminishing, conceivably some patients’ problems (drugs have been recommended for non-dependent alcohol ‘abusers’) do not warrant taking side-effect producing pills several times a day, and often medications do not help. Another reason to pause before taking strong measures to increase pill-taking is that compliance may be a marker of a pre-existing good prognosis rather than a cause of that prognosis. This is almost certainly part of the reason why even with an inactive placebo, good compliance is associated with good outcomes – as in the major UK disulfiram trial above. Good compliers tend to be diligent, health-conscious and well-organised people, traits likely to facilitate recovery from any illness, condition or misfortune. If this is the case, ‘artificially’ boosting compliance with an ineffective or only minimally effective medication will not improve outcomes to the degree expected from how well naturally good compliers do in research trials. On the other hand, even highly effective medications cannot exert their effects unless they are taken.

Given these considerations, how far should we go to overcome resistance to regularly taking pills? Inducements for complying, sanctions for not? As a last-ditch measure, criminal justice sanctions have been tried with some success in Britain to persuade offenders otherwise facing several months in prison to take disulfiram to reduce their alcohol-related offending. Loss of employment or professional status and ‘tough-love’ pressure are other tactics. As long as the patient is free to refuse and does not suffer sanctions simply for refusing, everyone can benefit: offenders avoid prison, prisons are relieved of some pressure, public money is saved, doctors can practice, their patients benefit, employees can continue to contribute to their work and society, and families are preserved.

Such benefits depend on how far medications really do protect patients and others from the serious consequences of continued heavy drinking, and that in turn (especially for disulfiram) depends on patients taking their medication. However, with naltrexone it is possible at least for a month or so to ensure the drug is active and take patient choice out of the equation by irreversibly injecting a long-acting formulation of the drug. It can help some patients cut down their drinking, as in this freely available pilot study. But in this trial, at least a third of full-dose patients experienced side-effects and – not least – if they had needed opiate-based pain relief, it would have been blocked by the long-acting injection. Long-acting naltrexone is one of the measures which by improving compliance also seems to have improved drinking outcomes relative to naltrexone pills taken every day.


Is disulfiram defunct? There was a time when it seemed to be going that way. In the mid-80s overall negative results from the first rigorous trial led the researcher to expect prescribing to wither. That did not happen in the USA and neither has it yet in Britain, where in England in 2015 disulfiram was dispensed 52,466 times. Though still a substantial player, the medication was on a downward trend from a peak of 60,842 in 2012 and well behind the 139,193 recorded for acamprosate in 2015. In hospitals as opposed to GPs’ surgeries, disulfiram used to be dispensed more often than acamprosate, but by 2013 that had reversed and by 2015, even in these specialist settings acamprosate was well ahead. One possibility is that lack of promotion and research in respect of this older medication have let it slip in comparison with newer products. However, in some British alcohol treatment units it has routinely and seemingly successfully been prescribed after detoxification.

Those doctors and patients have support from reviews which found the drug more effective than other medications, as long as consumption was supervised. One review was a particularly sophisticated analysis which divided trials into those which did or did not supervise consumption, and in which patients as well as researchers and doctors did not know whether they were taking disulfiram (‘double-blind’) versus those in which patients knew they were taking a drug which would cause unpleasant reactions if they drank. The ‘gold standard’ double-blind methodology robs disulfiram of its main active ingredient The results showed that the ‘gold standard’ double-blind methodology robbed disulfiram of its main active ingredient – the expectation of a nasty reaction if one returns to drinking – and that as well as this knowledge, patients usually need someone to bolster their resolve by making sure they take the pills. Given these circumstances, disulfiram not only substantially bettered a placebo in reducing drinking at the end of the treatment period but also substantially bettered naltrexone and acamprosate. Then take a look at the other review and at our commentary on its findings. Note also that British trials have found disulfiram effective but acamprosate not.

The drug’s modus operandi has been explained by experienced clinicians. First is the patient’s knowledge that disulfiram will cause an unpleasant reaction after drinking. Secondly, “taking disulfiram ... has certain symbolic connotations. It tells us that here is a patient who is willing ... to surrender some control over his freedom or urge to drink. Such a patient announces both to himself and to the wider world that he is not merely talking about changing his drinking habits ... but is actually doing something about it.” The final element derives from the need for the drug’s administration to be supervised if more than a small minority of patients are to benefit, providing “additional opportunities for involving family members in the broader therapeutic and monitoring enterprise.” Disulfiram is uncompromising in its distinctive effects, yet even for this drug, social, cognitive and symbolic influences mediate its impact on drinking.

With this information at your disposal, do you think the National Institute for Health and Care Excellence (NICE), the UK’s authority on medical treatments, was right to say disulfiram should normally only be considered when acamprosate or oral naltrexone are not suitable. Burrow through their report, and you will see that apart from the risks, this was because the evidence was weaker than for other drugs, which in turn was due to the lack of ‘double-blind’ trials – those in which neither patients nor research assessors know who has taken what. Disulfiram’s reviewers dismissed this as a “meaningless” concern, because the main active ingredient of the treatment is that patients know they have taken a drug which will cause a nasty reaction if they drink.

After making your judgement you may find yourself agreeing with one or other side of a deep divide in academic and expert opinion: one side sees disulfiram as a poorly evidenced and risky option best kept in reserve; another advocates its routine use, arguing that though it needs careful monitoring and usually supervised administration, among patients who will accept this option, it is safe and at least as effective as other medications. Underlying these views can be found differences in understandings of alcohol dependence. According to one view, the treatment merely suppresses drinking while disulfiram is being taken without “target[ing] the core phenomenon of alcohol dependence”. Understanding dependent drinking as a learnt reaction to certain cues, another view is that the treatment precisely targets that core by forcing a dislocation between cues and drinking which with repetition can become cemented, underpinning the construction of an alcohol-free way of living.


The placebo effect is the main active ingredient Across randomised trials and across all medications, these have only a modest impact on drinking relative to a pharmacologically inactive placebo. This is generally also true of both acamprosate and naltrexone, though disulfiram is a partial exception; in the special circumstances when doses can be and are effectively supervised to make sure pills are taken, effects verge on what is conventionally considered large.

It means that most patients do almost or just as well if offered a placebo instead of these medications. We can get a feel for how closely the placebo effect matches that of an active medication from an amalgamation of research on naltrexone and acamprosate. It found that six months after treatment started, 65% of placebo patients had lapsed to drinking compared to 77% prescribed acamprosate. Three months after treatment started the corresponding figures for naltrexone were 65% and 71%, while for relapse to heavy drinking they were 50% versus 59%. In each case these raw figures amounted to a statistically significant advantage for the medications, but a slight one which in individual trials often failed to materialise.

Usually research and reviews focus on the medication, but one revealing analysis has focused instead on the placebos against which alcohol medications have been benchmarked. Across relevant trials, improvements in alcohol patients randomly allocated to an inactive placebo on average dwarfed those analyses have found attributable to the additional effects of an active medication. The more severe the dependence, the greater was the improvement among placebo patients, and the greater that improvement, the less was gained by prescribing an active medication.

Unusually, the large US COMBINE study directly tested the power of a placebo by including set of patients not even prescribed an inactive pill, but instead allocated to what should have been a powerful psychosocial therapy. Their abstinence record was significantly worse (67% v. 74% days abstinent) than patients allocated to straightforward medical care but also prescribed a placebo – a testimony to the power of the placebo effect either (as the authors speculated) in positive form as a response to being prescribed medication, or in negative form as disappointment at not being prescribed. “Somewhat unexpectedly,” said the researchers, “we observed a positive effect of receiving placebo medication and medical management over and above that seen with specialist-delivered behavioural therapy alone.” Though large, placebo effects found in randomised trials of this kind might be weakened because research participants are normally told they stand a 1 in 2 chance of being prescribed an inactive placebo, dampening expectations that the pills will work.

What the ‘placebo effect’ consists of has yet to be adequately unpacked, but it is almost certainly far more than a reaction to taking a pill. The ‘placebo effect’ is almost certainly far more than a reaction to taking a pill Some of the apparent effect will be natural remission, but the fact that the more often a placebo is taken, the greater the reduction in drinking, suggests that the social and psychological influences associated with being actively engaged in treatment mediate much of the effects of medications. The impact of engaging in a treatment understood in that culture to help ‘cure’ the condition is one the ‘common factors’ explored in cell A2’s bite – and in Western societies, taking a medicine is perhaps the prime culturally endorsed way to signal the existence of a curable condition and to cure it. Another factor is how the clinician relates to the patient, the possible effect of which emerged in the arm of US COMBINE alcohol treatment trial which evaluated medical care allied with pharmacotherapy in the form of the doctor’s ability to credibly convey optimism about recovery. More on relationship issues in cell B3.

These findings should come as no surprise. Across medicine, when a condition is susceptible to expectations (such as depression and pain for example, but not the cancer which may be causing these) placebo effects rival those of active medications, and do so partly because the context and the expectations of the patient themselves induce biochemical effects which can mimic those of medications, including generation of the body’s natural opiate-type chemicals.

Thanks for their comments on this entry to Colin Brewer of the Stapleford Centre in England and Duncan Raistrick, formerly of the Leeds Addiction Unit in England and now Co-Director of RESULT. Commentators bear no responsibility for the text including the interpretations and any remaining errors.

Close Matrix Bite guide and commentary Close Matrix Bite Close Matrix Bite guide and commentary