Content uploaded by Pia Brandt Danborg
Author content
All content in this area was uploaded by Pia Brandt Danborg on Feb 12, 2018
Content may be subject to copyright.
Research
Precursors to suicidality and violence on antidepressants:
systematic review of trials in adult healthy volunteers
Andreas Ø Bielefeldt
1
, Pia B Danborg
1,2
and Peter C Gøtzsche
1
1
Nordic Cochrane Centre, Rigshospitalet, 2100 Copenhagen Ø, Denmark
2
Faculty of Health and Medical Sciences, University of Copenhagen, 2200 Copenhagen N, Denmark
Corresponding author: Peter C Gøtzsche. Email: pcg@cochrane.dk
Abstract
Objective: To quantify the risk of suicidality and violence
when selective serotonin and serotonin-norepinephrine
reuptake inhibitors are given to adult healthy volunteers
with no signs of a mental disorder.
Design: Systematic review and meta-analysis.
Main outcome measure: Harms related to suicidality, hostil-
ity, activation events, psychotic events and mood disturbances.
Setting: Published trials identified by searching PubMed and
Embase and clinical study reports obtained from the
European and UK drug regulators.
Participants: Double-blind, placebo-controlled trials in
adult healthy volunteers that reported on suicidality or vio-
lence or precursor events to suicidality or violence.
Results: A total of 5787 publications were screened and 130
trials fulfilled our inclusion criteria. The trials were generally
uninformative; 97 trials did not report the randomisation
method, 75 trials did not report any discontinuations and
63 trials did not report any adverse events or lack thereof.
Eleven of the 130 published trials and two of 29 clinical study
reports we received from the regulator y agencies presented
data for our meta-analysis. Treatment of adult healthy vol-
unteers with antidepressants doubled their risk of harms
related to suicidality and violence, odds ratio 1.85 (95%
confidence interval 1.11 to 3.08, p¼0.02, I
2
¼18%). The
number needed to treat to harm one healthy person was
16 (95% confidence interval 8 to 100; Mantel-Haenszel risk
difference 0.06). There can be little doubt that we under-
estimated the harms of antidepressants, as we only had
access to the published articles for 11 of our 13 trials.
Conclusions: Antidepressants double the occurrence of
events in adult healthy volunteers that can lead to suicide
and violence.
Keywords
Psychiatry, antidepressants, depression, healthy volunteers
Introduction
The reporting of harms in drug trials is generally
poor, with inadequate explanation of how they were
collected, and often harms are missing altogether.
1,2
From 2011 to 2012, the Nordic Cochrane Centre
received unredacted clinical study reports on
antidepressants filed for regulatory approval at the
European Medicines Agency and the UK Medicines
and Healthcare products Regulatory Agency. We
demonstrated selective reporting of major harms in
the published articles of duloxetine and inconsisten-
cies between the protocols and the clinical study
reports.
3
Furthermore, we noticed that the adverse
events tables in clinical study reports had hidden sui-
cidal events due to the medical dictionaries and
coding conventions used.
4
Based on the 70 clinical
study reports we received, we found that antidepres-
sants more than doubled the risk of suicidal and
aggressive behaviour in children and adolescents.
5
A 2012 systematic review of 33 trials in healthy
volunteers documented various effects of selective
serotonin reuptake inhibitors but only mentioned
the adverse events in a few words.
6
The review was
based on published articles and none of these lived
fully up to the CONSORT guideline for good
reporting.
Since their introduction in the late 1980s, the
benefits and harms of selective serotonin reuptake
inhibitors and serotonin-norepinephrine reuptake
inhibitors have been the subject of huge debate.
Because their harms – in particular suicidality –
have often been explained away as if they were dis-
ease symptoms or only a problem in children,
7–10
we
wished to quantify the risk of suicidality and violence
when these drugs are given to adult healthy
volunteers.
Methods
According to our prespecified protocol (available
from the authors), we included double-blind,
randomised placebo-controlled trials of selective
serotonin reuptake inhibitors or serotonin-norepi-
nephrine reuptake inhibitors (called antidepressants
throughout this paper) in adult healthy volunteers
!The Royal Society of Medicine 2016
Reprints and permissions: sagepub.co.uk/journalsPermissions.nav
Journal of the Royal Society of Medicine; 2016, Vol. 109(10) 381–392
DOI: 10.1177/0141076816666805
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
with no signs of a mental disorder. There were no
language restrictions. We excluded trials in abusers
of tobacco, drugs or alcohol and also imaging studies,
as these have another objective and furthermore are
of poor methodological quality.
11
We searched PubMed (1966 to December 2015)
and Embase (1974 to December 2015) using the
search strings specified in Table 1. The searches
were conducted in collaboration with an information
specialist to ensure the inclusion of drugs that were
not indexed in PubMed’s MeSH thesaurus. One
researcher (AØB) screened all search results by read-
ing titles and abstracts, and all excluded studies were
rescreened by the same researcher who also read the
included studies and extracted relevant trial data, e.g.
design, population size, harms, discontinuations and
funding, to a spreadsheet. A second researcher (PBD)
extracted data on a randomly selected sample of 30
articles; as there were no discrepancies, we did not
perform data extraction in duplicate.
We also included relevant clinical study reports on
antidepressant drugs we received from regulatory
agencies for our projects in this area.
3–5
We checked
whether the clinical study reports had been published
by searching for investigator names and drugs on
PubMed and Embase.
We defined harms as adverse events that were
either suicidality or violence or were considered pre-
cursor events to suicidality or violence in the litera-
ture, with a particular focus on the list of criteria used
by the Food and Drug Administration for its 2006
meta-analysis of suicidality in 100,000 patients
(Table 2).
12–18
We categorised the harms as suicidal-
ity, violence, activation events, psychotic events and
mood disturbances.
Uncertainties during data extraction were resolved
by discussion between the authors. If the reporting of
harms was unclear, we contacted the authors of the
trials for clarification. We assessed the risk of bias in
the trials focusing on randomisation and blinding.
We performed a meta-analysis calculating Peto’s
odds ratio for dichotomous outcomes with Review
Manager. Because of the considerable risk of carry-
over effects in crossover trials, we only used trials that
reported on periods separately.
19,20
Results
We identified 5787 publications in PubMed and
Embase, removed 569 duplicate entries, and screened
the titles and abstracts of the remaining 5218 records,
which left 316 articles for full-text reading (Figure 1).
We excluded 174 of these articles because the studies
were crossover trials using antidepressant inhibitor, a
placebo and a third drug, or because the studies were
parallel trials that used a drug other than an antidepres-
sant as an interaction drug. This left 142 articles on a
total of 130 trials that fulfilled our inclusion criteria, 52
of which were two-period crossover trials. Ten different
antidepressants were tested in the trials; citalopram
(n¼33) and paroxetine (n¼27) were most commonly
used.
The trials did not report much about their
methodology. All 130 trials were allegedly randomised,
but 97 (75%) did not describe the method and 75 (58%)
did not report any discontinuations or lack thereof.
Reporting of adverse events was generally inad-
equate; 63 trials (48%) did not report any adverse
events or stated that there were none; 43 trials
(33%) reported at least one adverse event; while 24
trials (18%) reported only the most frequently occur-
ring adverse events or those leading to discontinu-
ation. The source of funding was industry in 29
trials (22%), non-industrial sources in 47 trials
(36%), mixed in 17 trials (13%) and not reported in
37 trials (28%).
Thirteen of the 130 trials reported on at least one
of our predefined harms, 10 of which were parallel
group trials and three crossover trials. We could not
include two of the crossover trials, as there were no
Table 1. Search strings for PubMed and Embase.
Database Search string entered
PubMed (‘‘Healthy Volunteers’’[Mesh] OR Healthy OR Normal) AND (‘‘atomoxetine’’[Supplementary Concept] OR
‘‘reboxetine’’[Supplementary Concept] OR ‘‘O-desmethylvenlafaxine’’[Supplementary Concept] OR
‘‘clovoxamine’’[Supplementary Concept] OR ‘‘Serotonin Uptake Inhibitors’’[Pharmacological Action])
Embase 1 exp serotonin uptake inhibitor/
2 exp serotonin noradrenalin reuptake inhibitor/
3 1 or 2
4 normal human
5 3 and 4
382 Journal of the Royal Society of Medicine 109(10)
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
data for each period separately. We included the first
period of the third crossover trial, which ended pre-
maturely due to carryover effects despite a four-week
washout period.
A1
We received 29 clinical study reports from the
regulatory agencies, two of which fulfilled our inclu-
sion criteria for the meta-analysis. None of them had
been published. Most of the remaining clinical study
reports described non-blinded bioavailability studies
or drug interaction studies.
Thus, we included 11 trials from our literature
searches
A1–A11
and two from the regulatory
agencies
A12,A13
in our meta-analysis. The drugs were
citalopram, escitalopram, fluoxetine, paroxetine, ser-
traline or venlafaxine, in all cases given orally. Four
trials were not industry sponsored and did not have
industry employees among the authors.
A2,A3,A5,A 7
Trial characteristics, length of treatment and our pre-
defined harms with reported severity are shown in
Table 3. The harms we found all occurred within the
randomised phase of the trials, not during the with-
drawal phase after the trials were finished. The median
age was 30 years; 40% of the volunteers were women,
and the median publication year was 2008.
Risk of bias in the included trials
Adequate methods for sequence generation and con-
cealment of treatment allocation were described for
six trials
A3–A6,A8,A11
and for concealment of alloca-
tion for two trials.
A7,A12
The methods were not spe-
cified in five trials.
A1,A2,A9,A10,A13
Adequate blinding
methods were mentioned for seven trials
A5–A8,A10–A12
and not specified in six trials.
A1–A4,A9,A13
We did not
Table 2. Types of harms looked for when reading the reports.
Suicidality Core event Accident, asphyxia, attempt, burn, cut, drown, firearm, gas, gun, hanged, immol-
ation, injury, jump, monoxide, mutilate, overdose, poison, self-damage, self-
harm, self-inflicted, self-injury, shoot, suicide, suffocate
Potential event –
Violence Core event Antisocial behaviour, assaultive behaviour, criminal behaviour, homicidal idea-
tion, homicide, physical abuse, physical assault, physically threatening
behaviour, violence-related symptom
Potential event –
Activation events Core event Aggression, agitation, akathisia, amphetamine speed-like response, anxiety,
anxiety attack, argumentative, caffeine feeling, CNS stimulation, euphoria,
excessive energy, feeling euphoric, fidgetiness, frustration, hostility, hyper-
activity, hyperkinesia, hypomania (also listed as psychotic event), increased
agitation, increased anxiety, increased energy, increased irritability, inner
shakiness, irritability, jitteriness, jittery, mania (also listed as psychotic event),
motor restlessness, nervousness, overanxious, overstimulation, racing
thoughts, restless, restlessness, shakiness, skin crawling, unusual energy
Potential event Easily startled, panic, panic attack, panic symptoms, tenseness, tension, tension
increased, trembling, tremor, unable to relax
Psychotic events Core event Abnormal feelings, abnormal thinking, behavioural dyscontrol, confusion, delir-
ium, delusions, disorientation, feelings of doom, hallucinations, hypomania
(also listed as activation event), hysteria, incoherent thoughts, intrusive
thoughts, mania (also listed as activation event), paranoia, psychosis, unusual
thoughts
Potential event Abnormal dreams, bad dreams, increased dreams, intense dreams, nightmares,
strange dreams, vivid dreams
Emotional
disturbances
Core event Anhedonic, apathy, depersonalisation, derealisation, disinhibition, emotionally
detached, emotional lability, feeling blank, flat effect, flattened, impulsivity,
indifference, lack of empathy, reduced ability to feel positive emotions,
spaciness
Potential event –
Bielefeldt et al. 383
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
look at attrition because the subject of our research
was side effects, not beneficial effects.
Meta-analysis
Treatment of adult healthy volunteers with anti-
depressants doubled their risk of harms related to
suicidality and violence, odds ratio 1.85 (95% confi-
dence interval 1.11 to 3.08, p¼0.02, I
2
¼18%)
(Figure 2). The number needed to treat to harm one
healthy person was 16 (95% confidence interval 8 to
100; Mantel-Haenszel risk difference 0.06). Two clin-
ical study reports and one published trial reported the
severity of the harms.
Discussion
The century-old belief that patients with depression
are at heightened risk of suicide as they begin to
recover and their energy and motivation return
21
Figure 1. PRISMA flow diagram of included studies based on the literature searches.
5218 records screened
4902 obviously ineligible
records excluded
142 arcles (130 trials) included in
qualitave synthesis
5787 records idenfied in databases
13 published trials
117 trials did not report on any of
our predefined harms
569 duplicates excluded
316 full-text arcles
assessed
174 arcles excluded
because the trials used
drugs other than
andepressants and placebo
2 crossover trials did not
show results for first
period separately
11 published trials plus
2 unpublished clinical study reports
included in meta-analysis
29 clinical study reports from
drug agencies
27 clinical study reports
were either about open
label studies or used an
interacn
g
dru
g
384 Journal of the Royal Society of Medicine 109(10)
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
is being propagated everywhere, e.g. in the 2003 prac-
tice guideline from the American Psychiatric
Association, which states that ‘clinical observations
suggest that there may be an early increase in suicide
risk as depressive symptoms begin to lift but before
they are fully resolved’.
22
Because of this deeply ingrained idea, many
psychiatrists believe that when patients become sui-
cidal on an antidepressant drug, it is not an adverse
effect of the drug but a positive sign that the drug
starts working.
7,10
However, a systematic review
from 2009 showed that the research that has been
carried out contradicts this belief,
21
and our review
also suggests that it is wrong. We found that anti-
depressants double the risk of suicidality and vio-
lence, and it is particularly interesting that the
volunteers in the studies we reviewed were healthy
adults with no signs of a mental disorder. Our
results agree closely with a review of paroxetine
trials in both adults and children with mental dis-
orders using regulatory data released after a court
case. It included events both during treatment and
in the subsequent withdrawal phase and found a
doubling in hostility events (odds ratio 2.10, 95%
confidence interval 1.27 to 3.48).
23
While it is now generally accepted that antidepres-
sants increase the risk of suicide and violence in chil-
dren and adolescents
5,12
(although many psychiatrists
still deny this
10
), most people believe that these drugs
are not dangerous for adults. This is a potentially
lethal misconception.
7,10,15,24
As far as we know, our review is the first of the risk
of suicide and violence in healthy volunteers. It was
inspired by David Healy’s work.
7
In 2000, Healy pub-
lished a study he had carried out with 20 healthy
volunteers – all with no history of depression or
other mental illness – and to his big surprise, two of
them became suicidal when they received sertraline.
25
One was on her way out the door to kill herself in
front of a train or a car when a phone call saved her.
Both volunteers remained disturbed several months
later and seriously questioned the stability of their
personalities.
In one of the two crossover trials we excluded
because we did not have data on the first period sepa-
rately, a healthy volunteer committed suicide, which
was mentioned in both published articles.
A14,A15
She
had received duloxetine in increasing doses for 16
days, tapered off the maximum dose of 400 mg daily
very quickly (in just four days according to the design
of the study) and killed herself four days later while
on placebo. The authors, several of whom were
employees of Eli Lilly or owned stock in the com-
pany, judged her suicide ‘to be unrelated to study
drug treatment’,
A15
although it is well known that
the suicide risk is high when an antidepressant is
stopped abruptly.
10,23
There was no more informa-
tion about the suicide in the articles, and it was not
included in the listing of adverse events we acquired
from Eli Lilly, which only mentioned a woman who
reported suicidal ideation twice while on placebo. As
we do not know if this was the same patient, we asked
Eli Lilly for access to anonymised data for the volun-
teer who committed suicide and the detailed person
narrative, as we also wanted to know how it could be
possible to state that the suicide was not related to
Figure 2. Meta-analysis of suicidal or violent events or precursors to such events.
Bielefeldt et al. 385
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
Table 3. Harms in the meta-analysed trials. Some trials included arms with drugs that were not antidepressants; these data are not shown.
Study Study design Sex and age
Drug, dosage,
length of treatment
Number of
volunteers
Harms of interest
in antidepressant
group
Harms of interest
in placebo group
Almeida (2010)
A2
Parallel study Based on 18 completers:
Male: 18
Mean age, citalopram: 27.8
years (no range). Mean
age, placebo: 26.7 years
(no range).
20 mg citalopram,
28 days
Citalopram: 10
Placebo: 10
1 (agitation)
Briscoe (2008)
A3
Parallel study Male: 10. Female 10.
Mean age: 29 years (20–44).
20–80 mg fluoxe-
tine, 56 days
Fluoxetine: 14
Placebo: 6
1 (vivid dreams)
Carpenter (2011)
A4
Parallel study Male: 8.
Female: 13.
Mean age, sertraline: 30.2
years (20–46).
Mean age, placebo: 28.7 years
(20–54).
50–100 mg sertra-
line, 42 days
Sertraline: 11
Placebo: 11
1 (severe
nightmares)
Chial (2003)
A5
Parallel study Based on 39 completers:
Male: 23. Female: 16.
Mean age, venlafaxine: 28.8
years (no range).
Mean age, placebo: 31.1 years
(no range).
150 mg venlafaxine,
1day
Venlafaxine: 20
Placebo: 20
1 (jittery)
Furlan (2001)
A6
Parallel study Based on 49 completers:
Male: 27. Female: 22.
Mean age, paroxetine: 75.1
years (no range).
Mean age, sertraline: 70.7
years (no range).
Mean age, placebo: 70.3 years
(no range).
10–40 mg paroxe-
tine, 21 days;
50–150 mg sertra-
line, 21 days
Paroxetine: 18
Sertraline: 16
Placebo: 20
Sertraline:
1 (nervousness)
Garcia-Leal (2010)
A7
Parallel study Male: 43.
Mean age: 23.6 years (19–31).
10 mg or 20 mg
escitalopram,
1day
Escitalopram: 31
Placebo: 12
1 (anxious, on
10 mg)
(continued)
386 Journal of the Royal Society of Medicine 109(10)
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
Table 3. Continued.
Study Study design Sex and age
Drug, dosage,
length of treatment
Number of
volunteers
Harms of interest
in antidepressant
group
Harms of interest
in placebo group
Knorr (2011)
A8
Parallel study Male: 50.
Female: 28.
Mean age: 32 years (no
range).
10 mg escitalopram,
28 days
Escitalopram: 39
Placebo: 39
6 (restlessness)
1 (tremor)
9 (restlessness)
1 (tremor)
Levine (1987)
A9
Parallel study Male: 14.
Female: 106.
Mean age, fluoxetine: 43
years (no range).
Mean age, placebo: 46 years
(no range).
20–80 mg fluoxe-
tine, 56 days
Fluoxetine: 60
Placebo: 60
4 (anxiety)
4 (depression)
7 (nervousness)
3 (tremor)
5 (anxiety)
1 (depression)
2 (nervousness)
Madeo (2008)
A10
Parallel study Male: 48.
Mean age, citalopram: 31.1
years (no range).
Mean age, fluoxetine: 29.2
years (no range).
Mean age, placebo: 29.2 years
(no range).
20–40 mg citalo-
pram, 31 days;
20 mg fluoxetine,
31 days
Citalopram: 16
Fluoxetine: 16
Placebo: 16
1 (anxiety, on
fluoxetine)
Montejo (2015)
A11
Parallel study No gender data.
Mean age, escitalopram: 24.1
years (no range).
Mean age, placebo: 23.0 years
(no range).
20 mg escitalopram,
63 days
Escitalopram: 36
Placebo: 32
4 (abnormal
dreams)
1 (agitation and
tremor)
1 (anxiety)
2 (abnormal
dreams)
Pijl (1991)
A1
Crossover study,
converted to
parallel study by
trialists due to
carryover effects
Female: 23.
Mean age, fluoxetine: 38.1
years (25–53).
Mean age, placebo: 37.3 years
(27–54).
60 mg fluoxetine,
60 days
Fluoxetine: 11
Placebo: 12
3 (tremor) 1 (tremor)
CSR 050-001
A12
Parallel study Male: 52.
Mean age, sertraline: 32.1
years (19–60).
Mean age, placebo: 35.8 years
(19–60).
10 mg, 25 mg,
50 mg, 75 mg,
100 mg, 125 mg,
150 mg, 175 mg,
200 mg, 250 mg,
300 mg, 350 mg or
400 mg sertraline,
1day
Sertraline: 40 (39
were random-
ised)
Placebo: 12 (13
were
randomised)
1 (mild nervous-
ness)
1 (mild tremor)
1 (mild euphoria)
1 (mild tremor)
(continued)
Bielefeldt et al. 387
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
Table 3. Continued.
Study Study design Sex and age
Drug, dosage,
length of treatment
Number of
volunteers
Harms of interest
in antidepressant
group
Harms of interest
in placebo group
CSR 050-201
A13
Parallel study Male: 24.
Mean age, 200 mg sertraline:
25.8 (20–32).
Mean age, 400 mg sertraline:
27.1 (22–34).
Mean age, placebo: 25.9
(21–31).
200 mg or 400 mg
sertraline, 14
days
Sertraline 200 mg: 8
Sertraline 400 mg: 8
Placebo: 8
200 mg:
2 (nervousness,
one moderate,
one severe)
2 (tremor, one
moderate, one
unspecified
severity)
2 (thinking
abnormal, one
mild, one
moderate)
400 mg:
2 (nervousness,
one mild, one
severe)
3 (tremor, two
moderate, one
unspecified
severity)
1 (thinking
abnormal,
unspecified
severity)
1 (moderate
nervousness)
2 (thinking
abnormal, one
mild, one
moderate)
CSR: clinical study report.
388 Journal of the Royal Society of Medicine 109(10)
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
duloxetine, but the company refused to give us the
data.
In another of Eli Lilly’s studies, a healthy 19-year-
old student who had taken duloxetine in order to help
pay her college tuition hanged herself in a laboratory
run by Lilly.
26
It turned out that missing in the
FDA’s files was any record of the college student
and at least four other volunteers known to have
committed suicide, and Lilly admitted that it had
never made public at least two of those deaths.
26
In the other crossover trial we had to exclude, an
unknown number of volunteers discontinued parox-
etine due to restlessness, tremor and other adverse
events.
A16
We contacted the corresponding author
of the study who referred us to the first author, but
despite several attempts of making contact via two
different email addresses and phone (to the doctor’s
assistant), this author never responded.
Exploratory analyses of the clinical study reports
Although only two of the 29 clinical study reports
were eligible for our meta-analysis, e.g. as the studies
needed to be double-blind, two researchers (AØB and
PBD) read them all (2224 pages) and extracted data
independently, as we wanted to explore possible
selective reporting of harms in the published articles.
Nineteen clinical study reports reported on the harms
we investigated and nine of these were published, but
less than half of the harms were reported in the art-
icles (21 of 50 events on antidepressants and two of
four events on placebo).
One of these studies
A17
was mentioned by David
Healy
7
who had spoken with the study investigator,
Ian Hindmarch. It was a crossover trial of the
interaction between sertraline and diazepam that
was terminated due to unexpected adverse events
after only four days, before the first phase had been
completed and before any of the volunteers had
received diazepam. All five volunteers in the sertraline
group became agitated and four of them anxious,
while one of seven volunteers in the placebo group
became aggressive, agitated and anxious. The study
was never published.
Limitations
There can be little doubt that we underestimated the
harms of antidepressants. For 11 of our 13 trials, we
only had access to the published article, and it well
documented that the drug companies underreport
seriously the harms of antidepressants related to sui-
cide and violence, either by simply omitting them
from the reports, by calling them something else or
by committing scientific misconduct.
2–5,7,10,27
In trials
of duloxetine and sertraline, for example, only 33 of
45 cases of suicidal ideation, attempt or injury listed
in a trial register were also mentioned in the pub-
lished reports.
2
Psychiatrists believe that the suicide risk with anti-
depressants is only increased till age 24, but this mis-
conception builds on seriously flawed trial data that
the FDA has published.
12
Several meta-analysts have
pointed out just how unreliable the trials are.
5,10,28,29
A 2005 meta-analysis conducted by independent
researchers of the published trials included 87,650
patients of all ages and found twice as many suicide
attempts on drug than on placebo (odds ratio 2.28,
95% CI 1.14 to 4.55).
28
They also found out that
many suicide attempts must have been missing;
some of the investigators responded that there were
suicide attempts they had not reported in their trials,
while others replied that they did not even look for
them. Further, events occurring shortly after active
treatment was stopped were not counted. Another
2005 meta-analysis conducted by independent
researchers used UK drug regulator data and included
40,826 patients; they found a non-significant doubling
in suicides or self-harm events when events occurring
later than 24 hours after the randomised phase was
over were included (relative risk 2.14, 95% confidence
interval 0.96 to 4.75, our calculation).
29
These
researchers also noted that the companies had under-
reported the suicide risk in their trials, and they found
that non-fatal self-harm and suicidality were seriously
underreported compared to the reported suicides.
Even the FDA’s 2006 meta-analysis of 100,000
patients in 372 placebo-controlled trials
12,30
is ser-
iously flawed. Based on trials that were included in
FDA’s analysis, one of us has estimated that there are
likely to have been 15 times more suicides on anti-
depressant drugs than reported by the FDA.
10
Two
important reasons for the underreporting of suicides
are that the FDA trusted the data the companies sent
to them and that they only included events up to 24
hours after the randomised phase was over.
10
We did not plan for any sensitivity analyses related
to whether the randomisation and blinding methods
were adequately described, as we were very well aware
before we started our review that in the sort of trials
we would find, there would likely be very little or no
information about this. We reviewed trials in healthy
volunteers, which have a completely different purpose
than standard treatment trials, and the drug compa-
nies do not have any particular incentive to provide
details about how they blinded the drug and the pla-
cebo and how they randomised the volunteers in such
trials. Furthermore, we studied harms, not clinical
beneficial effects, as we included healthy people. It
would therefore be inappropriate to assume that
Bielefeldt et al. 389
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
trials that did not describe blinding and randomisa-
tion methods are less reliable than other trials.
Conclusions
Antidepressants double the occurrence of events in
adult healthy volunteers that can lead to suicide
and violence. We consider it likely that antidepres-
sants increase suicides at all ages.
Declarations
Competing Interests: All authors have completed the ICMJE
uniform disclosure form at http://www.icmje.org/coi_disclosure.
pdf (available on request from the corresponding author) and
declare no financial relationships with any organisation that
might have an interest in the submitted work in the previous
three years; no other relationships or activities that could appear
to have influenced the submitted work.
Funding: This study was funded by the Nordic Cochrane Centre,
Rigshospitalet. PBD is funded by a scholarship from the
University of Copenhagen, Faculty of Health and Medical
Sciences. The funding sources had no influence on any part of
the study.
Ethical Approval: No patient consent has been obtained for this
study, as it is a systematic review.
Guarantor: PCG
Contributorship: All authors had full access to all the data in the
study and take responsibility for the integrity of the data and the
accuracy of the analysis; AØB and PCG contributed to the study
concept and design and to the acquisition of data; AØB and PBD
contributed to searching, screening and extraction of data. All
authors contributed to the analysis and interpretation of data.
AØB and PCG contributed to the drafts of the manuscript, and
all authors critically reviewed the manuscript for publication. PCG
provided administrative and material support and was the study
supervisor and guarantor.
Acknowledgements: The authors thank information specialist
Henrik Hornemann from the Copenhagen University Library for
assistance in creating search strings in medical databases; Cochrane
collaborators from Japan, Maiko Suto, Erika Ota, and Sadequa
Shahrook, and China, Zhang Xin, for helping accessing, assessing,
translatingand interpretingforeign language articles; Justin Northrup
and Malcolm Mitchell from Eli Lilly, Joan Korth-Bradley and Chris
Gutteridge from Pfizer, and Steven Troy, formerly from Wyeth, for
supplying supplemental data. The authors thank our colleagues
Tarang Sharma, Emma Maund and Asbjørn Hro
´bjartsson for valu-
able comments and input throughout the project.
Provenance: Not commissioned; previously peer-reviewed at
another journal and peer-reviewed for JRSM by Julie Morris.
References
1. Loke YK and Derry S. Reporting of adverse drug reac-
tions in randomised controlled trials – a systematic
survey. BMC Clin Pharmacol 2001; 1: 3.
2. Hughes S, Cohen D and Jaggi R. Differences in report-
ing serious adverse events in industry sponsored clinical
trial registries and journal articles on antidepressant and
antipsychotic drugs: a cross-sectional study. BMJ Open
2014; 4: e005535.
3. Maund E, Tendal B, Hro
´bjartsson A, Jørgensen KJ,
Lundh A, Schroll J, et al. Benefits and harms in clinical
trials of duloxetine for treatment of major depressive
disorder: comparison of clinical study reports, trial
registries, and publications. BMJ 2014; 348: g3510.
4. Maund E, Tendal B, Hro
´bjartsson A, Lundh A and
Gøtzsche PC. Coding of adverse events of suicidality
in clinical study reports of duloxetine for the treatment
of major depressive disorder: descriptive study. BMJ
2014; 348: g3555.
5. Sharma T, Guski LS, Freund N and Gøtzsche PC.
Suicidality and aggression during antidepressant treat-
ment: systematic review and meta-analyses based on
clinical study reports. BMJ 2016; 352: i65.
6. Knorr U and Kessing LV. The effect of selective sero-
tonin reuptake inhibitors in healthy subjects. A system-
atic review. Nord J Psychiatry 2010; 64: 153–163.
7. Healy D. Let Them Eat Prozac: The Unhealthy
Relationship Between the Pharmaceutical Industry and
Depression. New York: New York University Press,
2004.
8. Medawar C and Hardon A. Medicines Out of Control?
Antidepressants and the Conspiracy of Goodwill.
Netherlands: Aksant Academic Publishers, 2004.
9. Whitaker R. Anatomy of an Epidemic. New York:
Broadway Paperbacks, 2010.
10. Gøtzsche PC. Deadly Psychiatry and Organised Denial.
Copenhagen: People’s Press, 2015.
11. Carp J. The secret lives of experiments: methods
reporting in the fMRI literature. Neuroimage 2012;
63: 289–300.
12. Stone M, Laughren T, Jones ML, Levenson M,
Holland PC, Hughes A, et al. Risk of suicidality in
clinical trials of antidepressants in adults: analysis of
proprietary data submitted to US Food and Drug
Administration. BMJ 2009; 339: b2880.
13. FDA. Revisions to Product Labeling. See http://www.
fda.gov/downloads/Drugs/DrugSafety/Informationby
DrugClass/UCM173233.pdf (last checked 2 September
2016).
14. Sinclair LI, Christmas DM, Hood SD, Potokar JP,
Issac A, Srivastava S, et al. Antidepressant-induced jit-
teriness/anxiety syndrome: systematic review. Br J
Psychiatry 2009; 194: 483–490.
15. Moore TJ, Glenmullen J and Furberg CD. Prescription
drugs associated with reports of violence towards
others. PLoS One 2010; 5: e15337.
16. Price J, Cole V and Goodwin GM. Emotional side-
effects of selective serotonin reuptake inhibitors: quali-
tative study. Br J Psychiatry 2009; 195: 211–217.
17. Breggin PR. Suicidality, violence and mania caused by
selective serotonin reuptake inhibitors (SSRIs): a review
and analysis. Int J Risk Saf Med 2003; 16: 31–49.
18. Levin R and Daly RS. Nightmares and psychotic
decompensation: a case study. Psychiatry 1998; 61:
217–222.
19. Higgins JP and Green S, eds. Methods for incorporat-
ing cross-over trials into a meta-analysis. In: Cochrane
Handbook for Systematic Reviews of Interventions. The
Cochrane Collaboration 2011.
390 Journal of the Royal Society of Medicine 109(10)
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
20. Elbourne DR, Altman DG, Higgins JPT, Curtin F,
Worthington HV and Vail A. Meta-analyses involving
cross-over trials: methodological issues. Int J Epidemiol
2002; 31: 140–149.
21. Mittal V, Brown WA and Shorter E. Are patients with
depression at heightened risk of suicide as they begin to
recover? Psychiatr Serv 2009; 60: 384–386.
22. Jacobs DG, Baldessarini RJ, Conwell Y, Fawcett JA,
Horton L and Meltzer H et al. Practice guideline for
the assessment and treatment of patients with suicidal
behaviors. American Psychiatric Association, 2003.
23. Healy D, Herxheimer A and Menkes DB.
Antidepressants and violence: problems at the interface
of medicine and law. PLoS Med 2006; 3: e372.
24. Lucire Y and Crotty C. Antidepressant-induced akathi-
sia-related homicides associated with diminishing
mutations in metabolizing genes of the CYP450
family. Pharmgenomics Pers Med 2011; 4: 65–81.
25. Healy D. Emergence of antidepressant induced suicid-
ality. Prim Care Psychiatry 2000; 6: 23–28.
26. Lenzer J. Drug secrets: What the FDA isn’t telling.
Slate 2005 September 27, http://www.slate.com/
articles/health_and_science/medical_examiner/2005/
09/drug_secrets.html.
27. Medawar C and Herxheimer A. A comparison of
adverse drug reaction reports from professionals and
users, relating to risk of dependence and suicidal
behaviour with paroxetine. Int J Risk Saf Med 2003;
16: 5–19.
28. Fergusson D, Doucette S, Glass KC, Shapiro S, Healy P,
Hebert P, et al. Association between suicide attempts and
selective serotonin reuptake inhibitors: systematic review
of randomised controlled trials. BMJ 2005; 330: 396.
29. Gunnell D, Saperia J and Ashby D. Selective serotonin
reuptake inhibitors (SSRIs) and suicide in adults: meta-
analysis of drug company data from placebo con-
trolled, randomised controlled trials submitted to the
MHRA’s safety review. BMJ 2005; 330: 385.
30. Laughren TP. Overview for December 13 Meeting of
Psychopharmacologic Drugs Advisory Committee
(PDAC). See www.fda.gov/ohrms/dockets/ac/06/
briefing/2006-4272b1-01-FDA.pdf (last checked 2
September 2016).
Appendix
A1. Pijl H, Koppeschaar HP, Willekens FL, Op de
Kamp L, Veldhuis HD and Meinders AE. Effect
of serotonin re-uptake inhibition by fluoxetine
on body weight and spontaneous food choice
in obesity. Int J Obes 1991; 15: 237–242.
A2. Almeida S, Glahn DC, Argyropoulos SV and
Frangou S.Acute citalopram administration
may disrupt contextual information processing
in healthy males. Eur Psychiatry 2010; 25: 87–91.
A3. Briscoe VJ, Ertl AC, Tate DB, Dawling S and
Davis SN.Effects of a selective serotonin reup-
take inhibitor, fluoxetine, on counterregulatory
responses to hypoglycemia in healthy indivi-
duals. Diabetes 2008; 57: 2453–2460.
A4. Carpenter LL, Tyrka AR, Lee JK, Tracy AP,
Wilkinson CW and Price LH.A placebo-con-
trolled study of sertraline’s effect on cortisol
response to the dexamethasone/corticotropin-
releasing hormone test in healthy adults.
Psychopharmacology (Berl) 2011; 218: 371–
379.
A5. Chial HJ, Camilleri M, Ferber I, Delgado-Aros
S, Burton D, McKinzie S, et al. Effects of ven-
lafaxine, buspirone, and placebo on colonic
sensorimotor functions in healthy humans.
Clin Gastroenterol Hepatol 2003; 1: 211–218.
A6. Furlan PM, Kallan MJ, Ten Have T, Pollock
BG, Katz I and Lucki I.Cognitive and psycho-
motor effects of paroxetine and sertraline on
healthy elderly volunteers. Am J Geriatr
Psychiatry 2001; 9: 429–438.
A7. Garcia-Leal C, Del-Ben CM, Leal FM, Graeff
FG and Guimaraes FS.Escitalopram pro-
longed fear induced by simulated public speak-
ing and released hypothalamic-pituitary-
adrenal axis activation. J Psychopharmacol
2010; 24: 683–694.
A8. Knorr U, Vinberg M, Hansen A, Klose M,
Feldt-Rasmussen U, Hilsted L, et al.
Escitalopram and neuroendocrine response in
healthy first-degree relatives to depressed
patients – a randomized placebo-controlled
trial. PLoS One 2011; 6: e21224.
A9. Levine LR, Rosenblatt S and Bosomworth J.
Use of a serotonin re-uptake inhibitor, fluox-
etine, in the treatment of obesity. Int J Obes
1987; 11suppl 3: 185–190.
A10. Madeo B, Bettica P, Milleri S, Balestrieri A,
Granata AR, Carani C, et al. The effects of
citalopram and fluoxetine on sexual behavior
in healthy men: evidence of delayed ejaculation
and unaffected sexual desire. A randomized,
placebo-controlled, double-blind, double-
dummy, parallel group study. J Sex Med
2008; 5: 2431–2441.
A11. Montejo AL, Deakin JFW, Gaillard R, Harmer
C, Meyniel F, Jabourian A, et al. Better sexual
acceptability of agomelatine (25 and 50 mg) com-
pared to escitalopram (20 mg) in healthy volun-
teers. A 9-week, placebo-controlled study using
the PRSexDQ scale. J Psychopharmacol 2015;
29: 1119–1128.
A12. Pfizer. Phase 1 single dose titration study to
assess the safety and pharmacokinetics of
sertraline. Clinical study report: 050-001. 1988.
A13. Pfizer. A double-blind, placebo-controlled,
parallel-group, multiple dose study of
Bielefeldt et al. 391
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
sertraline in healthy male volunteers. Clinical
study report: 050-201. 1988.
A14. Derby MA, Zhang L, Chappell JC, Gonzales
CR, Callaghan JT, Leibowitz M, et al. The
effects of supratherapeutic doses of duloxetine
on blood pressure and pulse rate. J Cardiovasc
Pharmacol 2007; 49: 384–393.
A15. Zhang L, Chappell J, Gonzales CR, Small D,
Knadler MP, Callaghan JT, et al. QT effects of
duloxetine at supratherapeutic doses: a pla-
cebo and positive controlled study. J
Cardiovasc Pharmacol 2007; 49: 146–153.
A16. Hergovich N, Aigner M, Eichler HG, Entlicher
J, Drucker C, Jilma B.Paroxetine decreases
platelet serotonin storage and platelet function
in human beings. Clin Pharmacol Ther 2000;
68: 435–442.
A17. Pfizer. A double-blind, placebo-controlled,
cross-over study to assess the effects of
sertraline, alone and with diazepam, on psy-
chomotor performance. Clinical study report:
050-206. 1983.
392 Journal of the Royal Society of Medicine 109(10)
at Copenhagen University Library on October 12, 2016jrs.sagepub.comDownloaded from
