Clinical Trials of Baclofen for Alcoholism: 2018 Update

Clinical Trials of Baclofen Treatment for AD:

Assessing the available clinical trials of baclofen for AD is not straightforward. It requires an understanding of how baclofen treatment is very different from current AD pharmacotherapies and the serious flaws of some RCTs.

Factors complicating the assessment of the large RCTs:

We’re only assessing a miniscule fraction of the AD patients on baclofen treatment:

Baclofen came to be used for AD in a complete reversal of the usual process ie a pharmaceutical company develops a medication, produces enough pre-clinical and clinical research to obtain the needed authorisations then markets it to the medical profession to prescribe to patients.

In contrast, baclofen was “marketed” directly to people affected by AD, initially via the book The End of My Addiction (1),  and then by the popular press, Internet forums and social media, completely independently from medical spheres of influence. Even prescribing can be out of medical control; on-line pharmacies allow patients to readily obtain baclofen if they can’t find a prescribing doctor.

This means that the number of patients taking baclofen for AD vastly outstrips the number involved in studies. In France, ~100,000 patients initiated baclofen treatment for AD via health system funded prescriptions between 2007 and 2013 (2). The number of additional patients taking baclofen obtained by other means (non reimbursed scripts, internet purchase or from neighbouring Spain where baclofen is OTC) is not known. In contrast, only 640 French patients have taken part in RCTs assessing baclofen’s efficacy in AD. The Alapidir study (3) took 2 years to recruit 320 patients from 39 addiction treatment centres because AD patients wanted to try baclofen treatment rather than enter clinical trials.

The numbers of patient across Europe, in France and surrounding countries, is estimated at 350,000 patients with growing but unknown numbers worldwide. There’s a vast pool of data out there that contains valuable information; not research but insights into real life experience with baclofen treatment for AD. It’s most easily accessed via Internet forums, the largest being the French sites and but there are many others across the world.

There is a severe lack of funding for large RCTs for Baclofen in AD:

There are only three large RCTs for baclofen in AD and, realistically this will not change in the foreseeable future. It’s because the pharmaceutical industry is the major source of funding for clinical research on medications, particularly for the large, expensive RCTs required for authorisations and marketing. There are very few alternative sources of sufficient money for RCTs studying new uses for cheap, out-of-patent medications, particularly if that drug will  compete against profitable medications. The H Pylori story is a good example: cheap antibiotics threatened the highly lucrative market for H2 blockers for the treatment of peptic ulcers.

Baclofen is in a similar situation: AD is a high prevalence, chronic or relapsing condition with high morbidity and mortality for which current treatments have low efficacy. This makes AD a prime target for the development of novel pharmacotherapies by the pharmaceutical industry, making old cheap baclofen very unwelcome indeed.

It’s instructive to examine the sources of funding of the three large RCTs of baclofen in AD: the Beraha study (4) was entirely funded by a private donation through University of Amsterdam and Bacloville’s (5) 1.2 million euro cost was funded by 800,000 euros from the French Government and 400,000 euros from an anonymous donor. Both donors are believed to be wealthy individuals who were successfully treated for AD with baclofen. A small French pharmaceutical company, Ethypharm, funded the Alpadir RCT (3). They aim to sell high dose (60mg) baclofen tablets in France for a modest profit.

Most studies of baclofen in AD are clinical observation studies produced by baclofen treating clinicians as these require little funding and can be incorporated into routine clinical work (Table 2).

The methodology used in two of the three large RCTs is completely different to the clinical use of baclofen in AD:

To understand this problem requires an explanation of baclofen’s major paradigm shift in AD treatment.

Pharmacotherapy with the currently used medications (naltrexone, acamprosate, disulfiram) involves complete cessation of alcohol via a detoxification (detox) of around 7 days and initiation of the medication at the maximum dose with aim of continuous abstinence. The maximum therapeutic effect is therefore at the start of treatment. If the patient relapses into drinking, the whole process generally needs to be restarted to restore abstinence. The detox phase is generally carried out in an Addiction Medicine inpatient setting for heavily dependent patients and is followed by variable periods of inpatient or outpatient rehabilitation. In Europe and the USA, this is typically via 4-6 week hospital or AD clinic inpatient stays (as in 79% of participants in the Beraha study) but can be via purely outpatient follow up (21% of Beraha participants and all Alpadir participants).

Baclofen treatment in AD is radically different. The regime described in the book “The End of My Addiction” and used across France and beyond is called the “Ameisen Method”. Patients are treated entirely in an outpatient setting, usually by their General Practitioner while the patient lives in his/her normal environment. There is no initial detox, the patient starting baclofen while still drinking alcohol. Baclofen must be started at low, ineffective doses to avoid severe side effects  then slowly titrated up over many weeks to months to achieve full therapeutic effect. The baclofen dose required is highly variable between individuals, from 30mg/day to 300mg/day or more. So starting baclofen treatment by doing a detox leaves the patient without either alcohol or effective pharmacotherapy for months and relapse risk is high.

The second major difference is that in baclofen treatment, the aim is safe levels of drinking (WHO criteria) rather than obligatory continuous abstinence. The aim is that alcohol consumption will fall steadily as the baclofen dose rises towards the effective dose. Episodes or relapses into heavy drinking are common as the patient adjusts to their new life but require only reinstatement or continuing upward titration of the baclofen dose.

The Bacloville RCT and the observational studies in Table 2 follow the Ameisen method and show the expected 50-60% success rate at 1 year.

The Alpadir and Beraha RCTs using the traditional detox/abstinence regime simply do not assess baclofen as it is used in “real life” clinical practice and also apply unsuitable outcome measures.

In the Alpadir and Beraha RCTs, the patient is classified as a definitive failure of baclofen treatment if s/he drinks any amount of alcohol (Alpadir) or at first heavy drinking day (Beraha). This does not make sense for baclofen given that continuing to increase the dose will give more therapeutic effect.

The classical detox/abstinence regime is not suited to baclofen in clinical use or in assessing its effectiveness in clinical trials.

The reasons for using this inappropriate treatment regime in Alpadir and Beraha relate to various factors such as the reluctance of Addiction Medicine to adopt the new paradigm (no detox, goal not abstinence), no experience of using baclofen clinically and allowing negative beliefs about baclofen to shape the study design with the aim of obtaining the desired result.

The results of Alpadir and Beraha largely reflect the level of other supports rather than baclofen treatment. This can be seen in the placebo rate – 75% in Beraha where most patients received intensive support and 10.5% in Alpadir where only outpatient support was provided. These placebo rates are very atypical for classical AD pharmacotherapy trials for which a 25% placebo effect is expected (7).

Table 1: Comparison Table of the three large Baclofen for AD RCTs:

Beraha (4) Alpadir (3) Bacloville (5)
Patient Number 151 320 320
Patient source Recruited from five Addiction treatment centres Recruited from 39 Addiction treatment centres Recruited in 60 GP practices

Inpatient or outpatient detox.

Abstinent for 4-21 days prior to study

Inpatient or outpatient detox. Abstinent for 3-14 days prior to study No detox, no abstinence required

Baclofen Titration


6 weeks


79% started with 4-6 week inpatient Rx

21% all outpatient Rx

4 weeks


All outpatient Rx

Over 52 weeks


All outpatient Rx

Study period

10 weeks


All outpatient Rx

20 weeks


All outpatient Rx

52 weeks


All outpatient Rx

Trial Duration 16 weeks 24 weeks 52 weeks
Assessed potential participant exclusion rate 330/481 = 69% No data 7/327 = 2%
Study Beraha Alpadir Bacloville
Physical issue exclusions

-Severe physical illnesses e.g. Parkinson׳s disease, gastric ulcer, duodenal ulcer, cerebrovascular disease, respiratory insufficiency, hepatic or renal insufficiency, and epilepsy

-Anti-hypertensive medication

-Severe renal, cardiac or pulmonary disorders

-Epilepsy or history of epilepsy

-Patient with organic disease serious enough to forbid   inclusion in the study according to the opinion of the investigator.


Psychiatric issue and medication exclusions Current severe axis I disorder (other than depression, anxiety, and bipolar disorder)

Severe psychiatric conditions:

schizophrenia and bipolar disorder

-Concomitant treatment with psychotropic medications, except antidepressants at stable dose for at least 2 months, diazepam and oxazepam

-Need for an inten- sive psychosocial intervention during follow-up

Patient with severe psychiatric pathology (psychosis, including schizophrenia and bipolar disorders) that could compromise the observance.


Suicide exclusion risk of suicide suicidal risk or history of suicide No restriction
Other substance addiction or treatment exclusion All exclusions except nicotine All exclusions except nicotine All permitted
Previous baclofen treatment exclusion

Use of baclofen in the past 30 days.


Previous baclofen intake Patient taking already baclofen or having taken baclofen.

Treatment Group

Success Criteria =

Primary Outcome

Continuous abstinence for 10 week study period Continuous abstinence for 20 week study period Abstinence or low risk drinking in last 4 weeks of 52 week study period
Treatment Failure First heavy drinking day (>5/6 drinks for F/M patients First alcoholic drink

Not terminated

All continued to 52 weeks

Placebo group result (expect ~25% in AD studies)

75% abstinence


Wk 7-16 (10 Wks)

10.5% abstinence


Wk 7-16 (10 Wks)

36.5% abstinence or low risk drinking

Wk 1-52 (52 Wks)

Study Beraha Alpadir Bacloville
Baclofen group result:

75% abstinence


Wk 7-16 (10 Wks)

11.9% abstinence


Wk 5-20 (16 weeks)

56.8% abstinence or low risk drinking

Wk 1-52 (52 Wks)

Drop out rate in titration phase (Bac+Plac pts)

47/151 (31%)


Wk 1-6 (6 weeks)

14/320 (4%)


Wk 1-4 (4 weeks)

Not applicable
Drop out rate in study phase (Bac+Plac pts)

26/151 (17%)


Wk 7-16 (10 Wks)

130/320 (41%)


Wk 5-20 (16 weeks)

Not applicable

Drop out rate over whole trial duration

(Bac+Plac pts)

73/151 (47%)


Wk 1-16 (16 weeks)

144/320 (45%)


Wk 1-20 (20 weeks)

102/320 (32%)


Wk 1-52 (52 weeks)

Serious AEs reported Hospitalisation for constipation in one patient “No major safety concerns” reported 7 deaths in baclofen group and 3 deaths in placebo group. No stat signif difference.


The Three Baclofen AD RCTs: Specific Issues:

Beraha RCT (4):

  • The most serious criticism is the poor and inexplicable methodological problem of using a mixed treatment cohort. The majority of participants, 115/151 (79%) patients started the post-detox study period with 4-6 weeks of inpatient rehabilitation treatment with intensive input such as twice daily psychotherapy. The other 36 (21%) had entirely outpatient treatment comprising one weekly support sessions. The Beraha study authors have refused requests to release data for the two groups although the 115 intensive treatment cohort is large enough to examine. Their refusal is hypothesised to be because the placebo success rate for the intensive treatment group is even higher than the Beraha study’s already very elevated 75% (over the 10 week study period), making it completely invalid for assessment of additional treatments such as baclofen. The conjecture is that the outpatient group has been deliberately included to pull down the placebo rate and make the study appear more robust.
  • The multiple exclusion criteria excluded 67% of prospective participants and resulted in a very selective group of AD patients with no suicide risk, no serious physical or psychiatric issues and no other substance abuse issues. This is not representative of the AD population and severely limits the relevance of the results to “real life” AD treatment. It is also likely to have contributed to the unusually high placebo success rate described above.
  • As described above, the Beraha RCT used a classical detox/abstinence alcohol pharmacotherapy regime that is not suited to baclofen treatment.
  • The maximum dose of baclofen permitted, 150mg/day is low based on available observational study data that situates the average dose at around 150-180mg/day, meaning that around half the participants would not achieve an effective baclofen dose. Despite this, the average dose in the Beraha study was only 94mg/day. The authors attributed this to high levels of concern about baclofen side effects within the inpatient treatment units, leading to very cautious titration and a 31% dropout rate in the six week titration phase. The drop out rate over the whole 16 weeks was 47%.

Alpadir RCT (3) :

  • This RCT also studied a low risk cohort via a long list of exclusions, as shown in Table 1 above. Even worse, 30% of their study cohort were also low (14.5%) or medium (16.1%) risk drinkers on WHO criteria and are inappropriate participants in the study, especially for the secondary outcome measures of Total Alcohol Consumption (TAC) and Heavy Drinking Days (HDD).
  • In spite of this cohort of relatively low complexity and baseline alcohol intake, the Alpadir cohort obtained poor results, with only 10.5% of the placebo group achieving the primary outcome of abstinence during the 20 week study period. This is very low; the expected placebo rate in AD pharmacotherapy studies of this classical type is around 25%. The authors mentioned this fact in the paper to explain their sample size: Based on a systematic review reporting an abstinence rate of approximately 25% with placebo (Mann et al., 2004) and anticipating a rate of 45% with baclofen. The very low placebo rate was noted in the discussion but no attempt was made to explain it The abstinence rate during 20 weeks was low in both groups and very far from the initial hypothesis.
  • The participant drop out rate was 144/310 or 45% over 20 weeks although only 14 (4%) dropped out in the first 4 weeks with a further 130 (41%) being lost in the subsequent 16 weeks. This suggests a fundamental issue with the study protocol or study’s acceptability to patients that the authors have not explained in the publication.

Two explanations can be hypothesized from information given in the paper:

i. Definitive treatment failure was defined as the first alcoholic drink. In contrast, the Beraha study, like most classical alcohol pharmacotherapy RCTs, defines it as the first heavy drinking day. Alpadir’s very strict definition of abstinence would make failures out of a high number of participants. Alpadir’s protocol involved pushing study participants to a fixed baclofen/placebo dose of 180mg/day. The default was the highest dose achievable without intolerable side effects. The 180mg/day dose was reached in 67% of baclofen group and 89% of the placebo group. They pushed participants to this dose despite considering that at day 28 of the study, the daily dose of 90mg was considered as high enough to be able to maintain abstinence, from which time any intake of alcohol was a failure of treatment.

ii. The target baclofen dose of 180mg/day was chosen because Ethypharm intended to market a 60mg tablet for use as a daily dose of 60mg tds or 180mg/day. Pushing the dose of a trial medication to a maximum tolerated dose rather than therapeutic effect would have contributed to the high drop out rate was well as being unsound, unsafe and ethically dubious.

  • The graphs of Total Alcohol Consumption (TAC) and Heavy Drinking Days (HDD) show a drop between the baseline and 28 day data points, measuring the expected drop in alcohol intake post detox and in the early weeks after. However subsequent to this, the graphs are flat for the next 112 days (D28 to D140) of the study, implying that there was no variation in the data point values in two groups of ~150 patients each over 20 weeks. While only one of these graphs is shown in the Alpadir paper (Figure 2), three more of identical appearance were shown at the ISBRA/ESBRA conference presentation in Berlin on 03/09/2016 (7a, 7b, 7c). There was clearly a lot of missing data in the study given that 41% of participants dropped out between weeks 4-20. This was dealt with by assuming that the alcohol consumption of dropped out patients was the same as those in the placebo group. This will pull the TAC and HDD data from the both treatment groups together and towards the placebo group. There is no explanation of the validity of this assumption, which would normally be that patients who exited the study for drinking or simply stopped participating had returned to heavy drinking.
  • In fact, although poorly designed and executed, the study may have been more positive than suggested. A post hoc analysis of the 215 heavy and very heavy drinkers brought the baseline average daily alcohol intake up to equivalent levels to Bacloville and Beraha, ~120g/day. In this group they reported an “important reduction” in TAC and HDD between the baclofen and placebo groups at 6 months but it did not reach statistic significance. The rate of abstinence in baclofen vs. placebo was not reported for this cohort so presumably was low in both groups for the reasons discussed above.

Bacloville RCT (5):

This is a very straightforward study, in stark contrast to the two others presented above. Bacloville has not yet been published in an English language journal although it has been submitted for publication. The results have already been presented at Addiction Medicine conferences including:

ISBRA/ESBRA in Berlin on 03/09/2016:

( ) : Slides : MP3

The 4th International Medicine in Addiction conference (IMiA) in Sydney on 24/03/2017.


However there is a brief report published in French for the Bulletin of the National Academy of Medicine by Bacloville Principal Investigator (PI), Prof Philippe Jaury, a copy of which can be found as Appendix 1, both as the original French paper and my translation into English.

Bacloville is the only large RCT in which baclofen is used correctly and this is because the PI, Prof Philippe Jaury is one of France’s most experienced baclofen prescribers, involved in daily clinical practice, in training GPs around France and as a major contributor to writing the 2012 and 2017 Prescription Guide for Baclofen in Alcohol Use Disorders (the 2017 version is Appendix 2).

The Bacloville study is a pragmatic risk reduction RCT carried out over 12 months, entirely in general practice (60 practices across France) and in an ambulatory setting, with patients living in their usual environment and being treated by their own GPs. The study accepted almost all the proposed participants (320/327) and had a minimum of exclusions, relating essentially to an inability to participate in the data collection and the likelihood of staying alive over the study duration. The patient cohort reflected the real life complexity of AD patients: often multi-morbid with physical, psychiatric and other substance abuse issues.

There was no detox or rehabilitation program, patients starting baclofen while still drinking alcohol. The dose titration was individualised for each patient up to a maximum of 300mg/day. Patients were asked to fill in a daily diary of alcohol consumption and symptoms (to measure side effects). The treating GPs were free to review patients at whatever interval was required and to refer them for any support programs they thought were needed eg marital therapy, CBT, as per normal management in General Practice. Abstinence was not the goal but rather a choice for participants.

Bacloville’s primary outcome measure was simple: success was defined as drinking at or below the WHO defined low risk drinking levels in the twelfth and last month of the study.

The baclofen group achieved this in 56.8% of participants and the placebo group in 35.6%. The baclofen group’s result is in line with the six other studies (all observational) done in baclofen prescribing clinics/practices, shown below in Table 2. The median baclofen dose was 180mg/day.

Bacloville’s placebo rate of 35.6% was higher than the 25% expected for a study of this type which the PI attributed to patients’ motivation, expectations and the better quality and continuity of care when delivered by the patients’ own GP. This likely also contributed to the low drop out rate of participants, 32%, the lowest of the RCTs despite Bacloville being a much longer study of sicker patients.

There were two potential problems with the Bacloville data. The first is that 76 patients came out of their original study group and received unblinded baclofen. This was because patients, usually those who reached the 300mg/day limit of baclofen or placebo without effect, were convinced they were on placebo. They refused to continue in the year long study protocol and demanded to be put on baclofen. As the patients had withdrawn consent to continue in the study, they were exited from it, classified as failures for their group under ITT and unblinded from the trial. Those on placebo who wished to try baclofen were started on it and those on baclofen had their dose increased beyond 300mg/day if wanted. This did not affect the primary outcome results but made it difficult to assess secondary measures under ITT such as amount of alcohol consumed.

The other is the issue of missing data for which multiple imputations were used with a series of independent analyses to verify their validity. For the primary outcome measure of drinking level, it was assumed that any patient lost to follow-up had returned to heavy drinking and was therefore a failure of the treatment.

The positive result of Bacloville, an RCT of long duration with a “real life” cohort of AD patients treated in General Practice has been very poorly accepted by the specialist Addiction Medicine community for reasons which are discussed in the next section. I’ve personally witnessed Bacloville’s PI, Philippe Jaury, a Professor of General Practice and practising GP being angrily told by an Australian Addiction Medicine Specialist that GPs are incapable of managing alcohol dependence and should stay out of it. Bacloville’s results show that the opposite is true but it’s not a “convenient truth”.

Table 2: Observational Trials of Baclofen >60mg/day:


Study No. pts







Primary result at study


Ameisen 2010

France 10

60 Obs 6 months Up to 330mg/day 66% abstinent

Gache, 2010

Switzeland 11*

54 Obs 1 year Up to 275mg/day 47% abstinent

Rigal, 2012

France 12

132 Obs 1 year

Up to 275mg/day

Average 129mg/day

80%: 59% abstinent and 21% low risk drinking (LRD) by WHO criteria

de Beaurepaire, 2012

France 13

100 Obs 2 years

Up to 400mg/day

Average 180mg/day

50% abstinent or LRD

Rigal, 2015

France 14

116 Obs 1 year

Up to 330mg/day

Average 159mg/day

60% abstinent or LRD

Barrault, 2017

France 15

100 Obs 1 year

No limit

Max used 210mg/day

68% abstinent or LRD

*11. Alcoologie et Addictologie, [S.l.], v. 32, n. 2, p. 119-124, june 2010. ISSN 2554-4853. Disponible à l’adresse : >>

Baclofen’s success rate of 50-70%: my clinical perspective:

Pretty much all AD patients who take baclofen at sufficient dose will feel it’s powerful anti-craving and anti-anxiety effects (13) but this translates to only 50-70% success in clinical practice.

The first reason is that baclofen treatment requires a moderate level of engagement and organisation to get to the effective dose via a slow and steady titration. Clearly those who need only low doses and have few side effects are the easiest to treat. Patients at the socially excluded, unstable, chaotic, unreliable end of the AD spectrum will find it difficult to maintain the thrice daily dosing at specific times and the steady dose titration over weeks to months. Compliance is greatly helped by baclofen’s strong anxiolytic effect so patients can often feel it wearing off when their next dose is due but this is not always enough.

The most common reason for treatment failure is that for many AD patients, alcohol acts as an “anaesthetic” which numbs the pain of their lives. This includes the damage caused by abusive, neglectful or poverty stricken childhoods, the trauma of military service or sexual assault or losses such as the death of children, parents or partners. Added to this are the losses due to the AD itself, to careers, marriages, relationships and children. The hardest journey is not stopping the alcohol but learning to live without it.