Lysergic Acid Diethylamide (LSD) for Alcoholism: Meta-Analysis of Randomized Controlled Trials

Abstract

Assessments of lysergic acid diethylamide (LSD) in the treatment of alcoholism have not been based on quantitative meta-analysis. Hence, we performed a meta-analysis of randomized controlled trials in order to evaluate the clinical efficacy of LSD in the treatment of alcoholism. Two reviewers independently extracted the data, pooling the effects using odds ratios (ORs) by a generic inverse variance, random effects model. We identified six eligible trials, including 536 participants. There was evidence for a beneficial effect of LSD on alcohol misuse (OR, 1.96; 95% CI, 1.36-2.84; p = 0.0003). Between-trial heterogeneity for the treatment effects was negligible (I² = 0%). Secondary outcomes, risk of bias and limitations are discussed. A single dose of LSD, in the context of various alcoholism treatment programs, is associated with a decrease in alcohol misuse.

Full text

http://jop.sagepub.com/
Journal of Psychopharmacology
http://jop.sagepub.com/content/early/2012/03/08/0269881112439253
The online version of this article can be found at:
DOI: 10.1177/0269881112439253
published online 8 March 2012J Psychopharmacol
Teri S Krebs and Pål Ørjan Johansen
Lysergic acid diethylamide (LSD) for alcoholism: meta-analysis of randomized controlled trials
Published by:
http://www.sagepublications.com
On behalf of:
British Association for Psychopharmacology
can be found at:Journal of PsychopharmacologyAdditional services and information for
http://jop.sagepub.com/cgi/alertsEmail Alerts:
http://jop.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
What is This?
- Mar 8, 2012OnlineFirst Version of Record >>
by guest on March 19, 2012jop.sagepub.comDownloaded from

Journal of Psychopharmacology
0(0) 1 –9
© The Author(s) 2012
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0269881112439253
jop.sagepub.com
Introduction
Alcohol is said to cause more overall harm than any other drug
(Nutt et al., 2010). Alcohol contributes to about 4% of total mor-
tality and about 5% of disability adjusted life-years to the global
burden of disease (Rehm et al., 2009). Despite the often extreme
individual and social consequences of alcohol misuse, many users
find it challenging to stop drinking. Alcoholism, also called alco-
hol dependence, continues to be difficult to treat, and many
patients do not achieve recovery from existing treatments
(Schuckit, 2009).
Numerous clinical investigators have claimed that treating
alcoholics with individual doses of lysergic acid diethylamide
(LSD), in combination with psychosocial interventions, can help
to prevent a relapse of alcohol misuse, for example, by eliciting
insights into behavioural patterns and generating motivation to
build a meaningful sober lifestyle (Dyck, 2008). LSD is well-
known for inducing spectacular and profound effects on the mind
(Henderson and Glass, 1994; Passie et al., 2008). It has previously
been used in standard treatment programs for alcoholism at many
clinics, but, unfortunately, assessments of the clinical value of
LSD have not been based on formal systematic review and meta-
analysis (Mangini, 1998). Hence, we have performed a quantita-
tive evaluation of the effectiveness of LSD for alcoholism, based
on data from randomized controlled clinical trials.
Methods
Search strategy and selection criteria
We searched the PubMed and PsycINFO databases (1943–2010),
without language restrictions, using the following terms: LSD,
lysergic, lysergide, psychedelic*, or hallucinogen*; and alcohol*,
addict*, or dependence. We independently inspected the search
results by reading the titles and abstracts. We retrieved each
potentially relevant publication located in the search and assessed
it for inclusion, subsequently examining the reference lists of eli-
gible studies and relevant review articles. We supplemented our
search for trials by contacting experts. If publications lacked
important information, we attempted to contact study investiga-
tors and institutions.
We specified inclusion and exclusion criteria and defined pri-
mary and secondary outcomes in the meta-analysis study proto-
col. We included randomized controlled trials of LSD for
alcoholism, in which control condition involved any type of treat-
ment, including doses of up to 50 mcg LSD as an active control. If
a trial included multiple randomized treatment arms, all partici-
pants in the eligible LSD arms and all participants in the eligible
control arms were pooled for analysis. We excluded participants
with schizophrenia or psychosis from analysis, as psychosis is
recognized as a contraindication for treatment with LSD (Johnson
et al., 2008; Passie et al., 2008).
Data extraction
Both reviewers independently extracted data and rated the risk of
bias of each included trial. Differences between the reviewers
were resolved through discussion. The following were recorded
Lysergic acid diethylamide (LSD) for
alcoholism: meta-analysis of
randomized controlled trials
Teri S Krebs1,2 and Pål-Ørjan Johansen1,2
Abstract
Assessments of lysergic acid diethylamide (LSD) in the treatment of alcoholism have not been based on quantitative meta-analysis. Hence, we
performed a meta-analysis of randomized controlled trials in order to evaluate the clinical efficacy of LSD in the treatment of alcoholism. Two reviewers
independently extracted the data, pooling the effects using odds ratios (ORs) by a generic inverse variance, random effects model. We identified
six eligible trials, including 536 participants. There was evidence for a beneficial effect of LSD on alcohol misuse (OR, 1.96; 95% CI, 1.36–2.84;
p = 0.0003). Between-trial heterogeneity for the treatment effects was negligible (I² = 0%). Secondary outcomes, risk of bias and limitations are
discussed. A single dose of LSD, in the context of various alcoholism treatment programs, is associated with a decrease in alcohol misuse.
Keywords
Alcoholism, alcohol-related disorders, hallucinogens, meta-analysis, psychedelics, substance-related disorders
1 Department of Neuroscience, Faculty of Medicine, Norwegian University
of Science and Technology (NTNU), Trondheim, Norway
2 Department of Psychiatry, Harvard Medical School, Boston, MA, USA
Corresponding author:
Pål-Ørjan Johansen, Department of Neuroscience, Faculty of Medicine,
NTNU, N-7489 Trondheim, Norway
Email: pal.johansen@ntnu.no
439253JOP0010.1177/0269881112439253Krebs and JohansenJournal of Psychopharmacology
2012
Review
by guest on March 19, 2012jop.sagepub.comDownloaded from

2 Journal of Psychopharmacology 0(0)
from each trial where available: intervention characteristics (LSD
dose, control condition, additional treatments); participant charac-
teristics (number, gender, age, inclusion and exclusion criteria);
information given to the participants on the study and the effects
of LSD; trial characteristics (publication year, location, funding
source); outcomes (primary and secondary outcomes, time of
follow-up, method of outcome assessment); evaluation of each
domain of the Cochrane risk of bias assessment tool (sequence
generation, allocation concealment, blinding, incomplete out-
come data, selective outcome reporting) (Higgins and Altman,
2008). Primary outcomes were alcohol misuse, defined as alcohol
use or consequences of alcohol use, as systematically measured
by interview or self-report at the first reported follow-up.
Secondary outcomes were alcohol misuse at short-term (approxi-
mately 3 months), medium-term (approximately 6 months) and
long-term (approximately 12 months) follow-up. We also
extracted data on abstinence, reports of adverse events and any
other secondary outcomes.
Data analysis
Categorical data on alcohol misuse were dichotomized into
‘improved’ or ‘not improved’. We counted as ‘improved’ outcome
categories indicating clear, substantial improvement in alcohol
misuse. Dichotomous and continuous outcome data were pooled
using the generic inverse variance method with a random effects
model. We calculated the effects of intervention results with esti-
mates of pooled odd ratios (ORs) and 95% confidence intervals
(CI) using Review Manager 5.0 (Nordic Cochrane Centre,
Cochrane Collaboration). The percentage of outcome heterogene-
ity attributable to between-trial heterogeneity was assessed by the
I2 statistic. Participants lost to follow-up were counted as not
improved. In a post hoc analysis of trials with available dichoto-
mized data, we calculated the pooled benefit difference on
improvement in alcohol misuse at first follow-up and also calcu-
lated the number needed to treat. The benefit difference (also
known as the risk difference) for each trial is the percentage of
improved patients in the LSD group minus the percentage of
improved patients in the control group. The number needed to
treat is the inverse of the pooled benefit difference and provides an
estimate of the average number of patients needed to be treated
with LSD rather than without LSD to achieve one additional
patient with improved outcome on alcohol misuse.
Results
Description of studies
We identified six eligible randomized controlled trials (Bowen et
al., 1970; Hollister et al., 1969; Ludwig et al., 1969; Pahnke et al.,
1970; Smart et al., 1966; Tomsovic and Edwards, 1970), including
additional reports on three of the trials (Kurland et al., 1971;
Ludwig et al., 1970; Smart et al., 1967). Details of the search are
shown in Figure 1, details of the included studies are shown in
Tables 1 and 2. Among the excluded studies were five non-
randomized controlled trials (Ables and Eng, 1967; Ables et al.,
1970; Jensen, 1962; Jensen, 1963; Van Dusen et al., 1967), one
quasi-randomized controlled trial (allocation by alternating
assignment) (Osmond et al., 1967), two randomized
controlled trials without any outcome data related to alcohol use
(both measured only general psychological variables) (Denson
and Sydiaha, 1970; Ditman et al., 1970), and one randomized con-
trolled trial without extractable outcome data on alcohol misuse
(this trial reported only ‘no statistically significant difference’
between LSD and control groups on alcohol misuse at 12 months
follow-up) (Johnson, 1969).
The six eligible trials included a total of 536 adults; of these
325 (61%) had been randomly assigned to receive full-dose LSD
and 211 (39%) to a control condition. Participants were male in-
patients, except for two females and a small number of day-care
patients in one of the trials (Smart et al., 1966). All participants
were seeking treatment for ‘alcoholism’ as their primary problem
and had been admitted to alcohol-focused treatment programs
before clinical trial recruitment, see Table 1. Note, the DSM-I
defined alcoholism as a ‘well established addiction to alcohol
without recognizable underlying disorder’ (American Psychiatric
Association, 1952).
Among the reported exclusion criteria, trials excluded poten-
tial volunteers with ‘psychiatric complications’ (Bowen et al.,
1970), with a ‘past history of schizophrenic reaction or severe
affective disorder’ (Hollister et al., 1969), or overt psychosis
(Ludwig et al., 1969; Smart et al., 1966; Tomsovic and Edwards,
1970). One trial included a subgroup of patients with schizophre-
nia (Tomsovic and Edwards, 1970), which we excluded from the
meta-analysis. Two trials included additional non-randomized
control groups or non-randomized sub-studies, which we also
excluded from the meta-analysis (Bowen et al., 1970; Tomsovic
and Edwards, 1970).
Single oral doses of LSD ranged from approximately 210 mcg
(3 mcg/kg) to 800 mcg, with a median dose of 500 mcg, see Table
1. No studies used multiple doses of LSD. The control conditions
included low-dose LSD (25 mcg or 50 mcg), d-amphetamine (60
4275 records identified through
database searching
6 additional records identified
through other sources
4090 records excluded
based on titles or
abstracts
68 records flagged for detailed
assessment
9 records included in meta-analysis
(6 trials)
4158 records screened after
duplicates removed
18 open-label or case
reports
6 non-randomized
(5 trials)
5 quasi-randomized
(1 trial)
7 randomized, but no
extractable outcome
data (3 trials)
23 reviews or other
59 records excluded:
Figure 1. Selection of trials for meta-analysis.
by guest on March 19, 2012jop.sagepub.comDownloaded from

Krebs and Johansen 3
mg), ephedrine sulphate (60 mg), or non-drug control conditions,
see Table 1.
Before the experimental drug session, all participants had
equivalent treatment within each trial; however, between the trials
the preparation for the experimental drug session varied from
minimal to extensive, with most studies providing brief orienta-
tion, often with little or no description of the possible effects of
LSD. During the experimental drug session, the most common
treatment was simple observation with brief reassurance by clinic
staff, only three studies included treatment groups who received
clinical interviews, psychotherapy, or active guidance. In four
studies, the experimental drug session took place in comfortable
surroundings with music available. After the experimental drug
session, only one study included multiple review sessions focused
on discussing the experiences during the drug session, while the
other studies provided only one brief review session or no review
session at all. See Table 2 and the original study publications for
details of the treatment protocols.
Each trial used clearly defined, standardized methods to assess
outcomes on alcohol misuse, although methods varied between
trials, see Table 1. Extracted dichotomous or categorical outcomes
included maintained abstinence from alcohol, alcohol use rating
scales, or composite alcohol use and social adjustment rating
scales; the one continuous outcome was percentage change in
time abstinent from alcohol. Based on examining each categorical
scale, outcome categories labelled ‘slight or questionable’
(Tomsovic and Edwards, 1970), ‘moderate’ (Ludwig et al., 1969),
or ‘fair’ (Bowen et al., 1970) were counted as ‘unimproved’; how-
ever, note that including these outcome categories indicating pos-
sibly trivial improvement as ‘improved’ does not substantially
change the results.
Effect of LSD on alcohol misuse
The pooled odds ratio on improvement in alcohol misuse between
the LSD and control groups was 1.96 (95% CI, 1.36–2.84; p =
0.0003) at the first reported follow-up, see Figure 2. Among the
five trials with dichotomized data, 185 of 315 (59%) LSD patients
and 73 of 191 (38%) control patients were improved at the first
reported follow-up, and the pooled benefit difference was 16%
(95% CI, 8%−25%; p = 0.0003), or, equivalently, the number
needed to treat is six. Including an estimated dichotomized
outcome for the one trial that reported only continuous outcome
Table 1. Included randomized controlled trials of LSD for alcoholism.
LSD (n) Control (n) Blinding of
patients, staff,
outcome assessors
Participant
characteristicsa
Age
(years)
Alcohol misuse outcome,
criteria for improvement
(months follow-up)
Retention
at first
follow-up
Location
(Funding)
Smart
et al., 1966
800 mcg
(10)
60 mg
ephedrine
sulfate (10) or
no drug (10)
Double-blind,
independent
assessors
Male and female
alcoholics, ’all
had a long history
of excessive and
uncontrolled drinking’
Median
38.5,
range
26–59
Drinking History
Questionnaire,
% change in time
abstinent,
continuous (6 mo)
100% ARF,
Toronto,
Canada (NR)
Hollister
et al., 1969
600 mcg
(36)
60 mg
d-amphetamine
(36)
Double-blind,
independent
assessors
Male veterans, ’acute
alcoholic episode’
within 2 weeks of
admission, ’all were
problem drinkers’
Median
45, range
31–51
Drinking Behaviour
Interview, score ≤ 10,
’Abstinent’ or ’Social’
drinking (2, 6 mo)b
81% LSD;
64%
control
VA Hospital,
Palo Alto,
CA, USA
(NIMH)
Ludwig
et al., 1969
3 mcg/kg,
~210 mcg
(132)
No drug, sit
alone and write
for 3 hr (44)
Double-blind
until LSD session,
independent
assessors
Male alcoholics, up
to four previous
admissions for
treatment of alcoholism
Range
21–55
Abstinence (1, 3 mo);
Behavior Rating Scale,
change score ≥ 5, ’Much
improved’ (6, 12 mo)b
100% MSH,
Madison, WI,
USA (NIMH)
Bowen
et al., 1970
500 mcg
(22)
25 mcg LSD
(22)
Double-blind, not
stated if assessors
independentc
Male veterans,
voluntarily applied for
treatment of alcoholism
Median
44.5
Adjustment Scale, score
≥ 6, ’Good adjustment’
(12 mo)
100% VA Hospital,
Topeka, KS,
USA (NR)
Pahnke
et al., 1970
450 mcg
(73)
50 mcg LSD
(44)
Double-blind,
independent
assessors
Male alcoholics,
voluntarily applied for
treatment of alcoholism
NR Drinking Behaviour
Scale, score ≥ 8,
’Minimal departure from
total abstinence’ (6, 12
mo)
88% LSD;
91%
control
MPRC,
Baltimore,
MD, USA
(NIMH)
Tomsovic &
Edwards, 1970
500 mcg
(52)
Treatment as
usual (45)
Double-blind until
LSD session, self-
report assessmentc
Male alcoholics,
average 12 years of
problem drinking
Mean 43 Drinking Adjustment
Scale, no more than
1 drinking episode in
follow-up period, ’Much
improved’ (3, 6, 12 mo)b
92% LSD;
73%
control
VA Hospital,
Sheridan,
WY, USA
(VA)
ARF: Alcoholism and Drug Addiction Research Foundation; MPRC: Maryland Psychiatric Research Center; MSH: Mendota State Hospital; NIMH: National Institute of Mental
Health; NR: not reported; VA: Veterans Administration.
aAll participants were recruited after admission to alcoholism treatment programs.
bProvided data on abstinence from alcohol.
cAssessment also included interview of close relative.
by guest on March 19, 2012jop.sagepub.comDownloaded from

4 Journal of Psychopharmacology 0(0)
Table 2. Details of treatment programs in included trials of LSD for alcoholism.
Treatment program
(approximate length
in days)
Preparation for
LSD session
Treatment during
experimental session
Setting of
experimental
session room
Aftercare related to
experimental session
Smart
et al., 1966
Individual and group
therapy within a
therapeutic community
Brief orientation; not
told name of LSD nor
that an active control
drug was used
3 h interview, followed by
occasional observation
No music or visual
stimuli; all patients
strapped to bed by
waist belt
One follow-up
review session with
interviewer
Hollister
et al., 1969
Brief counselling on alcohol
misuse; focus on alcohol
withdrawal (7)
Brief orientation; not
told name of LSD nor
that an active control
drug was used
Brief supportive reassurance;
emphasis on self-examination
Music, comfortable
furniture
None mentioned;
discharged within
48 hours; overall
’little or no specific
psychotherapy’
Ludwig
et al., 1969
Highly structured intensive
milieu therapy, including
group therapy (30)
Brief orientation;
minimal discussion of
LSD effects
3 h (a) psychotherapy, (b)
hypnosis + psychotherapy, or
(c) silent observation, followed
by occasional observation
Not described No follow-up with
experimental
session therapist
Bowen
et al., 1970
Interpersonal skill training
in groups (60)
Several group
orientation lectures on
LSD effects
Supportive reassurance; emphasis
on non-verbal introspection
Music, flowers, pictures,
’tasteful furniture’, two
quiet rooms
None mentioned
Pahnke
et al., 1970
Intensive individual
psychotherapy (49)
Extensive individual
preparation for LSD
Guidance aimed at eliciting
a ’peak or transcendental
experience’
Music, flowers,
pictures, ’comfortable
living room’
Multiple follow-up
review sessions
Tomsovic
& Edwards,
1970
Group psychotherapy (90) Lecture and reading
material; review of
problems and treatment
intentions
Supportive reassurance; not
encouraged to talk extensively
Music, flowers,
pictures, scenic view,
quiet room
One follow-up
review session in
group therapy
Figure 2. Improvement on alcohol misuse at the first available follow-up after LSD versus control treatments.
aContinuous outcome data.
data does not change the calculated pooled benefit difference or
number needed to treat.
There was a significant beneficial effect of LSD on alcohol
misuse in the short-term and medium-term, which was not statisti-
cally significant in the long-term, see Figure 3. At short-term fol-
low-up (2–3 months post-treatment), three trials reported
treatment response, and the pooled odds ratio between the LSD
and control groups was 1.85 (95% CI, 1.14–3.00; p = 0.01). At
medium-term follow-up (6 months post-treatment), five trials
reported treatment response, and the pooled odds ratio between
the LSD and control groups was 1.66 (95% CI, 1.11–2.47; p =
0.01). At long-term follow-up (12 months post-treatment), four
trials reported treatment response, and the pooled odds ratio
between the LSD and control groups was 1.19 (95% CI, 0.74–
1.90; p = 0.47).
Heterogeneity of the between-trial treatment outcome was
negligible in the pooled comparisons for alcohol misuse at the first
reported follow-up, short-term follow-up and medium-term fol-
low-up (I2 = 0%, for all p ≥ 0.60 for the χ2 test), and heterogeneity
was low at long-term follow-up (I2 = 15%, p = 0.32 for the χ2 test).
Effect of LSD on abstinence from alcohol
Among the three trials that reported maintained abstinence from
alcohol use, there was a beneficial effect of LSD at the first reported
by guest on March 19, 2012jop.sagepub.comDownloaded from

Krebs and Johansen 5
follow-up (1–3 months post-treatment) (OR, 2.07; 95% CI, 1.26–
3.42; p = 0.004) and short-term follow-up (2–3 months post-treat-
ment) (OR, 1.80; 95% CI, 1.07–3.04; p = 0.03), which was not
statistically significant at medium-term follow-up (6 months post-
treatment) (OR, 1.42; 95% CI, 0.65–3.10; p = 0.38), see Figure 4.
Heterogeneity of the between-trial treatment outcome was
negligible in the pooled comparisons for abstinence at first
reported follow-up and short-term follow-up (I2 = 0%, for both
p ≥ 0.38 for the χ2 test), while heterogeneity was moderate at
medium-term follow-up (I2 = 44%, p = 0.41 for the χ2 test).
Adverse events
Five trials reported a total of eight acute adverse reactions to LSD,
without any lasting harmful effects. Trial investigators did not
specifically mention whether there were adverse events among
participants in the control conditions. During the LSD experience,
two people ‘acted bizarrely’ (Tomsovic and Edwards, 1970), one
person became agitated (Hollister et al., 1969), another person had
a grand mal seizure during a period of agitation (this patient had a
history of alcohol withdrawal seizures and had been abstinent
from alcohol for only a few days) (Hollister et al., 1969) and two
people had unspecified ‘adverse reactions’ (Ludwig et al., 1969).
In the days after LSD, one person experienced transient ‘moderate
confusion’ (Hollister et al., 1969) and one person had a transient
‘adverse reaction’ (Pahnke et al., 1970). Additionally, investiga-
tors in one trial reported mild adverse reactions to LSD in a small
number of participants, including nausea, vomiting and ‘moderate
agitation’ that was relieved by social support, relaxation, or chang-
ing the lights and music (Hollister et al., 1969). Furthermore, in
one trial, about a third of the participants who received LSD
reported briefly experiencing ‘any perceptual thought or feeling
experience which impressed the patient with its vividness and
which was clearly related to the [LSD] experience’ on one or a
few occasions within a year after LSD, typically after using alco-
hol (Tomsovic and Edwards, 1970), while participants in another
trial specifically did not mention such experiences at follow-up
(Hollister et al., 1969).
Other outcomes
Other reported trial outcomes were difficult to assess and sum-
marize in detail, owing to large variation in the approaches
between the trials and lack of data for statistical analysis. However,
Figure 3. Improvement in alcohol misuse at short-, medium- and long-term follow-up after LSD versus control treatments.
aContinuous outcome data.
by guest on March 19, 2012jop.sagepub.comDownloaded from

6 Journal of Psychopharmacology 0(0)
no trials reported any detrimental effects of LSD on psychosocial
functioning or other outcomes. Of note, two of the three trials that
reported data on employment found statistically significant
improvements in employment in participants who received LSD
compared to those assigned to control conditions (Hollister et al.,
1969; Smart et al., 1966) but not Ludwig et al., (1969).
Risk of bias
Based on the definitions from the Cochrane risk of bias assess-
ment tool (Higgins and Altman, 2008), no trials were judged to
have a high risk of bias related to sequence generation or alloca-
tion concealment. All trials used random assignment and
attempted to conceal allocation; however, most trials did not
describe methods in detail. Two trials were judged to have a high
risk of bias due to inadequate blinding of patients or staff because
treatment allocation was concealed only until the time of the
possible LSD session (Ludwig et al., 1969; Tomsovic and
Edwards, 1970); the other four trials used double-blind designs
with active placebos. All trials were judged to have low or an
unclear risk of bias due to blinding of outcome assessment; in
four trials outcome was assessed by treatment-independent, allo-
cation-blind interviewers (Hollister et al., 1969; Ludwig et al.,
1969; Pahnke et al., 1970; Smart et al., 1966), in one trial the
outcome assessor was not explicitly described as allocation-
blind (Bowen et al., 1970) and in the remaining trial outcome
assessment was collected by self-report questionnaire, con-
firmed by telephone interview with a close relation (Tomsovic
and Edwards, 1970). Two trials were judged to have a high risk
of bias due to incomplete outcome data because participants
were excluded if they did not complete the intended treatment
program (Bowen et al., 1970) or if they received additional
doses of LSD (Pahnke et al., 1970). Retention rates were gener-
ally high, see Table 1, but two studies had substantial rates of
missing participants at follow-up (Hollister et al., 1969;
Tomsovic and Edwards, 1970). However, authors of both of
these trials expressed that missing participants had probably
relapsed to problem alcohol use, consistent with the strategy of
considering missing participants as unimproved. Two trials were
judged to have a high risk of bias because of possible selective
outcome reporting (Hollister et al., 1969; Ludwig et al., 1969);
both of these trials de-emphasized evidence for a treatment
effect at short-term follow-up and gave more detailed outcome
data on alcohol misuse at medium-term or late-term follow-up;
note, we were not able to obtain the protocol for any of the trials.
One trial was judged to have a high risk of bias due to baseline
imbalance (Pahnke et al., 1970); in this trial, participants who
received full-dose LSD were less likely than control participants
to be divorced, and more likely to have four or less prior admis-
sions for alcohol treatment, or to have graduated from high-
school. Importantly, however, also in this trial the treatment
groups were matched on baseline ratings of alcohol misuse.
Figure 4. Maintained abstinence from alcohol after LSD versus control treatments.
by guest on March 19, 2012jop.sagepub.comDownloaded from

Krebs and Johansen 7
Sensitivity analysis
For the primary outcome, improvement on alcohol misuse at first
follow-up, the beneficial effect of LSD remained statistically sig-
nificant (p ≤ 0.02) when excluding any two of the four larger tri-
als, with or without excluding either or both of the two smaller
trials. In a series of post hoc sensitivity analyses, excluding all
trials with a high risk of bias on each domain of the Cochrane risk
of bias assessment tool did not substantially change the primary
outcome. In particular, the effect of LSD increased and remained
significant when we excluded the two trials that used non-blinded
control conditions without an active placebo. Furthermore, the
primary outcome did not change when we limited analysis to the
four trials reporting outcome specifically on alcohol use, rather
than composite scores of alcohol use and social functioning, or
when we excluded the two trials with lower retention rates.
The findings on secondary outcomes of alcohol misuse at
short-term and medium-term follow-up and abstinence at first and
short-term follow-up are more sensitive to removing trials. In par-
ticular, none of the secondary outcomes remain statistically sig-
nificant (p ≥ 0.06) after removing the trial with the most favourable
effect of LSD in each respective analysis. Note that the analyses
of secondary outcomes are based on only three to five trials each.
Discussion
In a pooled analysis of six randomized controlled clinical trials, a
single dose of LSD had a significant beneficial effect on alcohol
misuse at the first reported follow-up assessment, which ranged
from 1 to 12 months after discharge from each treatment program.
This treatment effect from LSD on alcohol misuse was also seen
at 2 to 3 months and at 6 months, but was not statistically signifi-
cant at 12 months post-treatment. Among the three trials that
reported total abstinence from alcohol use, there was also a sig-
nificant beneficial effect of LSD at the first reported follow-up,
which ranged from 1 to 3 months after discharge from each treat-
ment program.
The findings from randomized controlled trials of a sustained
treatment effect of a single dose of LSD on alcohol misuse, which
may fade within 12 months, are consistent with many reports of
clinical experience and with data from most non-randomized con-
trolled and open-label studies of LSD for alcoholism (reviewed in
Mangini (1998)). In particular, a quasi-randomized trial reported
beneficial effects of LSD on alcohol misuse at 3 months post-
treatment (Osmond et al., 1967). Additionally, four non-
randomized controlled studies reported beneficial effects of LSD
on alcohol misuse at follow-up periods ranging from 6 to 18
months. However, these studies were poorly described (Ables and
Eng, 1967; Ables et al., 1970; Jensen, 1962; Jensen, 1963). Also
consistent with our findings, three controlled studies, excluded
from this meta-analysis because the control groups were non-ran-
domized (Bowen et al., 1970; Van Dusen et al., 1967) or because
of lack of extractable data (Johnson, 1969), reported no significant
treatment effect of a single dose of LSD on alcohol misuse at 12
to 18 months follow-up. Importantly, in the Bowen et al. (1970)
and Van Dusen et al. (1967) studies, the comparison group did not
volunteer to possibly receive LSD, probably creating selection
bias (see, for example, Ditman et al. (1970) on differences between
alcoholics who volunteer and those who decline to participate in
an LSD study), and in the Johnson (1969) study all patients were
administered the tranquilizer chlorpromazine during the acute
LSD effects, probably attenuating the LSD effects. Additionally,
in a randomized controlled trial of a single dose of LSD for heroin
addiction, daily urine test data covering the entire follow-up
period showed a significantly lower rate of relapse in the LSD
group compared to no drug group at 3, 6, 9 and 12 months post-
treatment (Savage and McCabe, 1973).
Given the evidence for a beneficial effect of LSD on alcohol-
ism, it is puzzling why this treatment approach has been largely
overlooked. Based on reviewing the literature, we have four sug-
gestions for why this happened. First, the randomized controlled
trials were underpowered and most did not reach statistical sig-
nificance when considered individually. Second, trial authors
expected unrealistic results and tended to discount moderate or
short-term effects. Third, early non-randomized clinical trials
were poorly described and had methodological problems, creating
the mistaken impression that well-designed studies did not exist.
Finally, the complicated social and political history of LSD led to
increasing difficulties in obtaining regulatory approval for clinical
trials (reviewed in Mangini (1998)).
The effectiveness of a single dose of LSD compares well with
the effectiveness of daily naltrexone, acamprosate, or disulfiram
(Krampe and Ehrenreich, 2010; Rösner et al., 2010a, 2010b), see
Table 3 for data from recent meta-analyses of these three com-
monly prescribed, approved medications for reducing relapse in
alcohol dependence.
Table 3. Data from recent meta-analyses of randomized controlled clinical trials on the effectiveness of LSD, naltrexone, acamprosate and disulfiram
for alcoholism or alcohol dependence.
Outcome LSD, single dose Naltrexone, daily Acamprosate, daily Disulfiram, daily
Benefit difference
(95% CI) NNT
Benefit difference
(95% CI) NNT
Benefit difference
(95% CI) NNT
Benefit difference
(95% CI) NNT
Improvement on alcohol misuse,
or return to heavy drinking
16% (8%, 25%) 6 11% (7%, 15%) 9 1% (-2%, 5%) 100 Not reported
Maintained abstinence, or
return to any drinking
15% (4%, 25%) 7 3% (1%, 6%) 33 11% (7%, 15%) 9 11% (-1%, 22%) 9
LSD outcomes are at first follow-up after single dose and are compared to no drug or active placebo. Naltrexone and acamprosate outcomes are during daily drug treat-
ment and are compared to placebo. Disulfiram outcomes are during daily unsupervised drug treatment and are compared to other or no treatment. Data on naltrexone,
acamprosate and disulfiram extracted from published meta-analyses (Rösner et al., 2010a, 2010b; Krampe and Ehrenreich, 2010). Pooled benefit differences calculated
using a random-effects, inverse variance method. Benefit difference = % patients with beneficial outcome in experimental – % patients with beneficial outcome in con-
trol. Number needed to treat (NNT) = 1/(benefit difference).
by guest on March 19, 2012jop.sagepub.comDownloaded from

8 Journal of Psychopharmacology 0(0)
Regarding the effects of the LSD experience, investigators of
one trial noted, ‘It was rather common for patients to claim sig-
nificant insights into their problems, to feel that they had been
given a new lease on life, and to make a strong resolution to dis-
continue their drinking’ (Ludwig et al., 1969). Investigators of
another trial noted, ‘It was not unusual for patients following their
LSD experience to become much more self-accepting, to show
greater openness and accessibility, and to adopt a more positive,
optimistic view of their capacities to face future problems’ (Bowen
et al., 1970). The subjective effects and neurobiological mecha-
nisms of LSD are similar to other psychedelic substances such as
mescaline (contained in peyote and other psychedelic cactus),
psilocybin (magic mushrooms) and dimethyltryptamine (aya-
huasca) that have been used by humans for thousands of years
(Bruhn et al., 2002; McGlothlin, 1964), and in clinical studies the
effects of psychedelics are often regarded as highly valued and
meaningful (Griffiths et al., 2006; Grob et al., 2011; Studerus
et al., 2011). Regular consumption of peyote and ayahuasca have
been claimed by indigenous groups to be helpful in maintaining
sobriety from alcohol and other addictive drugs (Albaugh and
Anderson, 1974; Fábregas et al., 2010).
Estimates of the rate of adverse events of LSD in alcoholics
and others should include data from non-randomized as well as
randomized trials. Based on extensive animal research and human
experience, there is now widespread recognition that LSD and
similar psychedelic substances are physically safe, but acute psy-
chiatric adverse events such as anxiety and confusion should be
anticipated, and LSD administration should occur in a comforta-
ble environment with informed participants (Johnson et al., 2008;
Passie et al., 2008).
Several matters in this meta-analysis deserve discussion. First,
trials typically lacked detailed descriptions of the populations
studied, including diagnosis methods. However, all participants
were recruited into the trials after admission to alcohol treatment
programs with a primary diagnosis of alcoholism, making it likely
that the patients are representative of typical clinical practice.
Second, there were not enough trials to examine the effect of LSD
dose or other treatment variables; all of the trials used a high or
very high dose of LSD and employed different treatment frame-
works. Third, it is possible that additional randomized controlled
trials were never published or were missed by our literature
search. Fourth, three trials either concealed that LSD might be
used (Hollister et al., 1969; Smart et al., 1966) or gave very little
information about its likely effects (Ludwig et al., 1969), and in
two of these trials participants were left alone in a room during
much of the LSD effects (Ludwig et al., 1969; Smart et al., 1966);
including people who might be reluctant to participate in a trial of
LSD or who were unprepared for the LSD effects may have atten-
uated the treatment effect and increased the risk of adverse events.
Fifth, blinding is a common problem to clinical trials of active
interventions, including most pharmacological and behavioural
treatments; most trials included in this meta-analysis attempted to
minimize risks of bias related to blinding by using active place-
bos and/or using explicitly treatment-independent, allocation-
blind interviewers for outcome assessment. However, the use of
low-dose LSD as an active placebo in two of the trials may have
attenuated the between-group treatment effect. Finally, primary
outcome measures on improvement in alcohol misuse varied
between trials; however, all of the clinical trials used standard-
ized questionnaires. Additionally, three trials also reported data
on the same clearly-defined outcome: maintained abstinence
from alcohol use.
It is uncommon for a psychiatric drug to have a positive treat-
ment effect for months after a single dose. Indeed, investigators of
one LSD trial noted, ‘most alcoholics report a waning of the initial
inspiration, euphoria, and good intentions gleaned from the LSD
experience when they are again confronted with the former
stresses and difficulties in their lives’ (Bowen et al., 1970). As
suggested by many investigators, repeated doses of LSD – for
example weekly or monthly – might elicit more sustained effects
on alcohol misuse than a single dose of LSD (Bowen et al., 1970;
Osmond et al., 1967; Savage and McCabe, 1973; Smart et al.,
1966). We need further data on whether subgroups of individuals
exist for whom LSD present an increased beneficial effect or risk
for adverse events. Future clinical trials could combine a range of
doses of LSD with current evidence-based alcohol relapse preven-
tion treatments. As an alternative to LSD, it may be worthwhile to
evaluate shorter-acting psychedelics, such as mescaline, psilocy-
bin, or dimethyltryptamine.
Funding
This work was supported by the Research Council of Norway (grant num-
ber 185924).
References
Ables MF and Eng EW (1967) Group treatment of chronic alcoholism
with LSD-25. In: Highlights of the twelfth annual conference in coop-
erative studies in psychiatry. Perry Point, MD: Veterans Administra-
tion Central Neuropsychiatric Research Laboratory.
Ables MF, Eng EW and Curtin ME (1970) Group treatment of chronic
alcoholism with LSD-25: study II. Newsletter for research in psychol-
ogy, United States Veterans Administration 12: 17–21.
Albaugh BJ and, Anderson PO (1974) Peyote in the treatment of alcohol-
ism among American Indians. Am J Psychiatry 131: 1247–1250.
American Psychiatric Association (1952) Diagnostic and Statistical Man-
ual for Mental Disorders. First Edition. Washington, DC: American
Psychiatric Association, 39.
Bowen WT, Soskin RA and Chotlos JW (1970) Lysergic acid diethylam-
ide as a variable in the hospital treatment of alcoholism: a follow-up
study. J Nerv Ment Dis 150: 111–118.
Bruhn JG, De Smet PA, El-Seedi HR and Beck O (2002) Mescaline use
for 5700 years. Lancet 359: 1866.
Denson R and Sydiaha D (1970) A controlled study of LSD treatment in
alcoholism and neurosis. Br J Psychiatry 116: 443–445.
Ditman KS, Moss T, Forgy E, et al. (1970) Characteristics of alcoholics
volunteering for lysergide treatment. Q J Stud Alcohol 31: 414–422.
Dyck E (2008) Psychedelic Psychiatry: LSD from Clinic to Campus. Bal-
timore, MD: Johns Hopkins University Press.
Fábregas JM, González D, Fondevila S, et al. (2010) Assessment of addic-
tion severity among ritual users of ayahuasca. Drug Alcohol Depend
111: 257–261.
Griffiths RR, Richards WA, McCann U and Jesse R (2006) Psilocybin can
occasion mystical-type experiences having substantial and sustained
personal meaning and spiritual significance. Psychopharmacology
(Berl) 187: 268–283.
Grob CS, Danforth AL, Chopra GS, et al. (2011) Pilot study of psilocybin
treatment for anxiety in patients with advanced-stage cancer. Arch
Gen Psychiatry 68: 71–78.
Henderson LH and Glass WJ (1994) LSD: Still With Us After All These
Years. New York: Lexington Books.
Higgins JPT and Altman DG (2008) Chapter 8: Assessing risk of bias
in included studies. In: Higgins JPT and Green S (eds) Cochrane
by guest on March 19, 2012jop.sagepub.comDownloaded from

Krebs and Johansen 9
Handbook for Systematic Reviews of Interventions. Version 5.0.1
(updated September 2008). Cochrane Collaboration. http://www.
cochrane-handbook.org
Hollister LE, Shelton J and Krieger G (1969) A controlled comparison of
lysergic acid diethylamide (LSD) and dextroamphetmine in alcohol-
ics. Am J Psychiatry 125: 1352–1357.
Jensen SE (1962) A treatment program for alcoholics in a mental hospital.
Q J Stud Alcohol 23: 315–320.
Jensen SE (1963) Treatment of chronic alcoholism with lysergic acid
diethylamide. Can Psychiatr Assoc J 8: 182–188.
Johnson FG (1969) LSD in the treatment of alcoholism. Am J Psychiatry
126: 481–487.
Johnson M, Richards W and Griffiths R (2008) Human hallucinogen
research: guidelines for safety. J Psychopharmacol 22: 603–620.
Krampe H and Ehrenreich H (2010) Supervised disulfiram as adjunct
to psychotherapy in alcoholism treatment. Curr Pharm Des 16:
2076–2090.
Kurland AA, Savage C, Pahnke WN, Grof S and Olsson JE (1971) LSD in
the treatment of alcoholism. In: Vinar O, Votava Z, Bradley PB (eds)
Advances in neuropsychopharmacology: proceedings of the 7th con-
gress of the collegium internationale neuro-psychopharmacologicum.
Amsterdam: North-Holland, 361–372.
Ludwig AM, Levine J, Stark L and Lazar R (1969) A clinical study of
LSD treatment in alcoholism. Am J Psychiatry 126: 59–69.
Ludwig AM, Levine J and Stark LH (1970) LSD and alcoholism: a clini-
cal study of treatment efficacy. Springfield, IL: Charles C Thomas.
Mangini M (1998) Treatment of alcoholism using psychedelic drugs: a
review of the program of research. J Psychoactive Drugs 30: 381–418.
McGlothlin WH (1964) Hallucinogenic Drugs: A Perspective with Spe-
cial Reference to Peyote and Cannabis. Santa Monica, CA: RAND
Corporation. http://www.rand.org/pubs/papers/P2937.html (accessed
25 February 2012).
Nutt DJ, King LA and Phillips LD (2010) Drug harms in the UK: a multi-
criteria decision analysis. Lancet 376: 1558–1565.
Osmond H, Albahary R, Cheek F and Sarett M (1967) Some problems in
the use of LSD 25 in the treatment of alcoholism. In: Abramson HA
(ed) The use of LSD in psychotherapy and alcoholism. Indianapolis,
IN: Bobbs-Merrill, 434–457.
Pahnke WN, Kurland AA, Unger S, Savage C and Grof S (1970) The
experimental use of psychedelic (LSD) psychotherapy. JAMA 212:
1856–1863.
Passie T, Halpern JH, Stichtenoth DO, Emrich HM and Hintzen A (2008)
The pharmacology of lysergic acid diethylamide: a review. CNS Neu-
rosci Ther 14: 295–314.
Rehm J, Mathers C, Popova S, et al. (2009) Global burden of disease and
injury and economic cost attributable to alcohol use and alcohol-use
disorders. Lancet 373: 2223–2233.
Rösner S, Hackl-Herrwerth A, Leucht S, et al. (2010a) Acamprosate for
alcohol dependence. Cochrane Database Syst Rev 9: CD004332.
Rösner S, Hackl-Herrwerth A, Leucht S, et al. (2010b) Opioid antagonists
for alcohol dependence. Cochrane Database Syst Rev 12: CD001867.
Savage C and McCabe OL (1973) Residential psychedelic (LSD) therapy
for the narcotic addict. Arch Gen Psychiatry 28: 808–814.
Schuckit MA (2009) Alcohol-use disorders. Lancet 373: 492–501.
Smart RG, Storm T, Baker EF and Solursh L (1966) A controlled study
of lysergide in the treatment of alcoholism. 1. The effects on drinking
behavior. Q J Stud Alcohol 27: 469–482.
Smart RG, Storm T, Baker EF and Solursh L (1967) Lysergic acid
diethylamide (LSD) in the treatment of alcoholism; an investiga-
tion of its effects on drinking behavior, personality structure, and
social functioning. Toronto: Published for the Alcoholism and
Drug Addiction Research Foundation of Ontario: University of
Toronto Press.
Studerus E, Kometer M, Hasler F and Vollenweider FX (2011) Acute,
subacute and long-term subjective effects of psilocybin in healthy
humans: a pooled analysis of experimental studies. J Psychopharma-
col. 25: 4135–4152.
Tomsovic M and Edwards RV (1970) Lysergide treatment of schizo-
phrenic and nonschizophrenic alcoholics: a controlled evaluation. Q J
Stud Alcohol 31: 932–949.
Van Dusen W, Wilson W, Miners W and Hook H (1967) Treatment of
alcoholism with lysergide. Q J Stud Alcohol 28: 295–304.
by guest on March 19, 2012jop.sagepub.comDownloaded from