ArticlePDF Available

Effectiveness of antidepressants: An evidence myth constructed from a thousand randomized trials?


Abstract and Figures

Antidepressants, in particular newer agents, are among the most widely prescribed medications worldwide with annual sales of billions of dollars. The introduction of these agents in the market has passed through seemingly strict regulatory control. Over a thousand randomized trials have been conducted with antidepressants. Statistically significant benefits have been repeatedly demonstrated and the medical literature is flooded with several hundreds of "positive" trials (both pre-approval and post-approval). However, two recent meta-analyses question this picture. The first meta-analysis used data that were submitted to FDA for the approval of 12 antidepressant drugs. While only half of these trials had formally significant effectiveness, published reports almost ubiquitously claimed significant results. "Negative" trials were either left unpublished or were distorted to present "positive" results. The average benefit of these drugs based on the FDA data was of small magnitude, while the published literature suggested larger benefits. A second meta-analysis using also FDA-submitted data examined the relationship between treatment effect and baseline severity of depression. Drug-placebo differences increased with increasing baseline severity and the difference became large enough to be clinically important only in the very small minority of patient populations with severe major depression. In severe major depression, antidepressants did not become more effective, simply placebo lost effectiveness. These data suggest that antidepressants may be less effective than their wide marketing suggests. Short-term benefits are small and long-term balance of benefits and harms is understudied. I discuss how the use of many small randomized trials with clinically non-relevant outcomes, improper interpretation of statistical significance, manipulated study design, biased selection of study populations, short follow-up, and selective and distorted reporting of results has built and nourished a seemingly evidence-based myth on antidepressant effectiveness and how higher evidence standards, with very large long-term trials and careful prospective meta-analyses of individual-level data may reach closer to the truth and clinically useful evidence.
Content may be subject to copyright.
BioMed Central
Page 1 of 9
(page number not for citation purposes)
Philosophy, Ethics, and Humanities
in Medicine
Open Access
Effectiveness of antidepressants: an evidence myth constructed
from a thousand randomized trials?
John PA Ioannidis1,2
Address: 1Clinical Trials and Evidence-Based Medicine Unit, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine
and the Biomedical Research Institute, Foundation for Research and Technology-Hellas, Ioannina, Greece and 2Institute for Clinical Research and
Health Policy Studies, Department of Medicine, Tufts University School of Medicine, Boston, USA
Email: John PA Ioannidis -
Antidepressants, in particular newer agents, are among the most widely prescribed medications
worldwide with annual sales of billions of dollars. The introduction of these agents in the market
has passed through seemingly strict regulatory control. Over a thousand randomized trials have
been conducted with antidepressants. Statistically significant benefits have been repeatedly
demonstrated and the medical literature is flooded with several hundreds of "positive" trials (both
pre-approval and post-approval). However, two recent meta-analyses question this picture. The
first meta-analysis used data that were submitted to FDA for the approval of 12 antidepressant
drugs. While only half of these trials had formally significant effectiveness, published reports almost
ubiquitously claimed significant results. "Negative" trials were either left unpublished or were
distorted to present "positive" results. The average benefit of these drugs based on the FDA data
was of small magnitude, while the published literature suggested larger benefits. A second meta-
analysis using also FDA-submitted data examined the relationship between treatment effect and
baseline severity of depression. Drug-placebo differences increased with increasing baseline
severity and the difference became large enough to be clinically important only in the very small
minority of patient populations with severe major depression. In severe major depression,
antidepressants did not become more effective, simply placebo lost effectiveness. These data
suggest that antidepressants may be less effective than their wide marketing suggests. Short-term
benefits are small and long-term balance of benefits and harms is understudied. I discuss how the
use of many small randomized trials with clinically non-relevant outcomes, improper interpretation
of statistical significance, manipulated study design, biased selection of study populations, short
follow-up, and selective and distorted reporting of results has built and nourished a seemingly
evidence-based myth on antidepressant effectiveness and how higher evidence standards, with very
large long-term trials and careful prospective meta-analyses of individual-level data may reach
closer to the truth and clinically useful evidence.
Few drugs have been as successful blockbusters as the class
of antidepressants. Cumulatively, hundreds of millions of
patients have taken these medications, and the selective
serotonin reuptake inhibitors (SSRIs) and newer genera-
tion drugs in particular have been immensely popular.
Antidepressants reflect one of the major manifestations of
medicalization of modern society [1]. In 2006, 5 of the 35
Published: 27 May 2008
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14 doi:10.1186/1747-5341-3-14
Received: 19 March 2008
Accepted: 27 May 2008
This article is available from:
© 2008 Ioannidis; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 2 of 9
(page number not for citation purposes)
drugs with top sales in the USA were antidepressants, and
each of them had sales of 1.08–2.25 billion dollars in that
year (Table 1) [2]. About 30% of the cost of depression in
the USA (80 billion dollars per year) goes to drug expen-
ditures [3].
This is not an epidemic that lacks evidence-based material
to support it. Few drugs have had such a long chain of
double-blind, placebo-controlled trials performed to
demonstrate their effectiveness and to pass through seem-
ingly strict regulatory approvals. The randomized litera-
ture of antidepressants is apparently one of the richest in
evidence-based credentials. While for a large proportion
of medical interventions, we have no or few clinical trials
ever conducted, for antidepressants there are probably
well over a thousand. PsiTri, an online library of clinical
trials for mental health conditions [4], lists 4058 clinical
trials for depression, and a large share of them (over a
quarter of the total, exact count depends on eligibility cri-
teria) pertain to randomized trials of antidepressants. A
systematic review of SSRI trials for diverse indications
until 2003 [5] found 702 trials (411 comparisons
between SSRIs and placebo, 220 comparisons between
SSRIs and tricyclic antidepressants, and 159 comparisons
between SSRIs and active therapies other than placebos or
tricyclic antidepressants). In another review [6] of 12 anti-
depressants where only double-blind, placebo-controlled
trials for diverse indications in adults were involved,
sponsors furnished data to the FDA on 406 trials with
approximately 100,000 randomized patients.
Formally, statistically significant benefits have been
repeatedly demonstrated and the medical literature is
flooded with several hundreds of "positive" trials, both
pre-approval and post-approval. In theory, this is a proto-
type of evidence-based medicine where treatments pass
through rigorous randomized testing, and are vastly suc-
cessful both in the clinical science arena as well as in the
This picture of bliss was questioned recently by two large,
well-conducted meta-analyses [7,8]. In this review, I
examine first what these meta-analyses have found and
what are some possible limitations of these studies. Then
I try to dissect what are the components that have con-
structed the seemingly evidence-based picture that antide-
pressants are so effective and why this picture may be
problematic on close scrutiny. I try to address eventually
whether antidepressants are widely indicated to treat
depression and whether it is unethical to kill a living
myth. Finally, I make some suggestions about how we can
get appropriate evidence on these drugs.
Upset by meta-analyses
Selective reporting
The first meta-analysis used data that were submitted to
the U.S. Food and Drug Administration (FDA) for 12 anti-
depressant drugs that were approved between 1987 and
2004 [7]. These were bupropion SR (Wellbutrin SR, Glax-
oSmithKline), citalopram (Celexa, Forest), duloxetine
(Cymbalta, Eli Lilly), escitalopram (Lexapro, Forest),
fluoxetine (Prozac, Eli Lilly), mirtazapine (Remeron,
Organon), nefazodone (Serzone, Bristol-Meyers Squibb),
paroxetine (Paxil CR, GlaxoSmithKline), sertraline
(Zoloft, Pfizer), venlafaxine (Effexor, Wyeth), and venla-
faxine XR (Effexor XR, Wyeth). The major advantage of
using data submitted to FDA is that this includes all the
trials that each company had registered as evidence in sup-
port for marketing approval or change in labelling. This
registration allows one to have knowledge of all these tri-
als, regardless of whether they were eventually published
or not. Moreover, the process of regulatory review is such
that there is less room for manipulating analyses and dis-
torting results in data entered in the FDA registry tables.
The meta-analysts found 74 eligible FDA-registered trials
with 12,564 patients. Among them, a third (n = 26 trials
[31%] with 3449 patients) had remained unpublished.
The FDA had determined that half of the registered trials
(38/74) had found statistically significant benefits for the
antidepressant ("positive" trials). All but one of these tri-
als had been published in journals. Conversely, of the
other half trials (36/74) that were deemed to be "nega-
tive" by the FDA, one in three were published as "nega-
tive" results; another 11 trials were published, but the
results were presented in such a way so as to seem "posi-
tive" and 22 "negative" trials were silenced and never
appeared in the literature.
Table 1: Top-selling antidepressants in the USA, 2006
Drug (brand name) Rank across all drugs Sales (billions $)
Venlafaxine XR (Effexor XR) 6 2.25
Escitalopram (Lexapro) 10 2.10
Sertraline (Zoloft) 15 1.77
Bupropion XL (Wellbutrin XL) 16 1.67
Duloxetine (Cymbalta) 35 1.08
Information is derived from [2]
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 3 of 9
(page number not for citation purposes)
The meta-analysts studied the estimated effectiveness of
these drugs when data were combined from the FDA
records and when data were combined from the pub-
lished literature. For all drugs, the published literature
inflated the effect sizes. The inflation varied from 11% to
69% and it was 32% on average. The FDA data would sug-
gest that these agents had small, modest benefits (stand-
ardized effect size [ES] = 0.31 on average). Conversely, for
4 of the 12 agents, if one were to perform unawares only
a meta-analysis of the published data, the summary result
would suggest clinically important effectiveness (ES>0.5).
This was not true for any agent based on more complete
FDA data.
Treatment and placebo effect as a function of baseline severity
A second meta-analysis in PLoS Medicine used also data
that were submitted to FDA on 6 new generation antide-
pressants and eventually used the information on four of
them (5 trials on fluoxetine, 4 on venlafaxine, 8 on
nefazodone, and 16 on paroxetine) [8]. For 2 other drugs
(sertraline and citalopram), some trials were simply
reported even in the FDA databases as having non-signif-
icant results. In contrast with the other meta-analysis, the
investigators of this meta-analysis did not wish to impute
data when there was such missing information.
The PLoS Medicine meta-analysis asked the question: is
there a relationship between the baseline severity of
depression and the difference in effectiveness between
drug and placebo? Meta-regression analyses identified
such a relationship. Drug-placebo differences were gener-
ally small, but they increased with increasing baseline
severity. The meta-analysts used a previous consensus [9]
to propose that a clinically important difference needs to
be at least 3 points in the Hamilton scale or ES>0.50. The
difference between drug and placebo became large
enough to be clinically important only in the small
minority of patient populations with severe depression
(baseline score exceeding 28 in the study population).
Even in these severely depressed patients, the difference
between drug and placebo was due to the fact that placebo
became less effective; there was no evidence that the anti-
depressants became more effective. The authors con-
cluded that most of the benefit from antidepressants is
duplicated by the placebo effect. This is a conclusion that
had been proposed also based on earlier meta-analysis
[9]. Moreover, the current meta-analysis added the insight
that these agents may be of clinical use only in severely
depressed people, a small minority compared with the
vast populations who take antidepressants currently. Even
in the few extremely depressed patients, the eventual ben-
efit was due to lack of responsiveness of placebo, not due
to increased responsiveness to antidepressants.
Limitations in the meta-analyses
Both meta-analyses have some limitations. Many more
trials are conducted after approval or outside of the FDA
approval process. Moreover, registries of approved agents
do not include antidepressants that were possibly tested
in clinical trials in the USA, but did not make it (presum-
ably because of more "negative" results), although they
made it and were approved in other countries, e.g. fluox-
amine, milnacipran, or mianserin. Among antidepressive
drugs tested in the USA, only the "luckier" ones, the ones
with larger ES, went to the FDA and received approval. The
lack of a comprehensive global database is a major deficit
in that we may be missing trials done in countries where
the overall results for a particular agent were not very
promising or overtly negative. Figure 1 shows a simple
simulation: suppose that a drug is tested in 40 countries
and 5 small trials are preformed for licensing purposes in
each country. Let us suppose that on average the drug has
a true effect that is small (ES = 0.20). Each of the perfectly
unbiased studies is expected to find on average ES = 0.20
and there can be some variability. We can examine situa-
tions with different levels of variability around this aver-
age, corresponding to standard deviations of 0.20, 0.40,
and 0.60. The smaller the trials and the larger the diversity
of the populations and drug response, the more variability
is expected around the mean of ES = 0.20. Suppose the
drug is approved only in countries where the 5 trials show
average ES at least 0.20. This is expected to happen in
about half the countries. Figure 1 shows what the average
ES estimates are in the trials registered in countries where
the drug was approved: ES is markedly inflated. Similar
considerations apply, if we consider not only many coun-
tries, but also many drugs tested in many countries.
Even focusing on FDA-registered trials, their data are not
necessarily totally unbiased. Inherent biases in the study
design and analysis cannot be corrected by simple regis-
tration. Data collection, arbitration of measurements and
outcomes and multiple analysis options leave room for
selectivity and for presentation of more optimal conclu-
sions – even in FDA-registered results. Second, even these
data are eventually incomplete in important details. This
is amply demonstrated by the considerable number of
studies that were simply registered as having "negative"
results without further details on effect sizes, and by the
additional missing information that the meta-analysts
had to impute even for FDA-registered data. Third, these
trials did not have available individual-level information
and the data collection and arbitration of outcomes and
measurements remained out of reach of the meta-ana-
lysts. For the considerable proportion of patients who did
not complete the trials, typically last observation carried
forward (LOCF) methods were applied, but these have
limitations and may lead to overestimation of treatment
effects in some circumstances [10].
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 4 of 9
(page number not for citation purposes)
All these limitations are more likely to have resulted in
inflation of the treatment benefit, although there is con-
siderable uncertainty about the exact bias. Of note, the
PLoS Medicine meta-analysis [8] noticed funnel plot asym-
metry, i.e. smaller trials had larger effects than larger trials.
Funnel plot asymmetry is typically considered a sign of
publication bias (small "negative" trials remaining
unpublished), but this is clearly a misleading simplifica-
tion [11]. Here publication bias in theory is impossible for
FDA-registered data. The authors attributed the asymme-
try to confounding due to higher severity scores in smaller
trials [8]. However, an alternative explanation is that even
for FDA-registered trials, results may still be biased. Exclu-
sion or inclusion of specific patients and data due to ques-
tionable eligibility criteria or grey measurements,
selection of imputation techniques, use or not or adjust-
ments, and selective reporting of outcomes allow for
manipulation in effect size estimation. In small trials, the
same amount of manipulation will inflate the effect size
more than in large trials, in other words the vibration of
the effect size is larger [12,13]. The FDA review process
will of course decrease analytical flexibility, but evalua-
tion of depression involves messy outcomes and analyses
are not cut in stone. In all, if anything, expectation of these
biases further reinforces the message about antidepres-
sants being less effective than thought.
A more serious limitation is inherent in the use of meta-
regression techniques in the PLoS Medicine meta-analysis.
The main analysis used a fixed effects meta-regression,
and only a secondary analysis used a mixed effects
approach. The latter, which may be more appropriate
than fixed effects [14], had less conclusive results. Meta-
regression modelling can be biased [15]. When the trials
have only small differences in the average values of sever-
ity (as in this case), the slope of the regression terms can
be affected by outliers and leverage problems. The most
important limitation stems from the ecological fallacy
[16,17]. The regression used as a moderator variable the
average baseline severity of depression in each group of
participants in each trial arm. However, this is a proxy that
does not represent equally well all participants. For exam-
ple, the average baseline score may be 28, but this may
include patients with scores of 17, 27, 32, and 36. The
relationship may not have been the same, if data could
have been analyzed for individual patients. This is to say,
while net effectiveness (difference of drug from placebo)
seemed to increase with increasing average severity,
within a specific trial it could be that the effectiveness
decreased with increasing severity. Ecological fallacy is the
main reason why meta-regression analyses with group
average are viewed with scepticism [17].
Finally, the PLoS Medicine meta-analysis describes the end
of the severity spectrum in the analyzed trials as contain-
ing patients who are "most extremely depressed" or alter-
natively "very severe" depression. In fact, "very severe"
depression would correspond to patients with even worse
depression status, primarily those hospitalized because of
major depression. In fact, the analyzed regulatory trials
have typically avoided including hospitalized patients
with truly so extreme depression, because these newer
agents had been shown early on to be ineffective – or at
least less effective than older agents – in such patients.
Understanding the construction of the myth of
antidepressant effectiveness
Acknowledging these caveats, the lessons we get from
these meta-analyses and from the previous literature on
antidepressants show us that there are many components
that have helped to create the impression that antidepres-
sants are very effective and worthy of being so popular in
the general population.
Statistical versus clinical significance
The typical trial in the antidepressant field is a small inves-
tigation with anywhere between a few dozen and a few
Anticipated mean and range of average effect size (ES) for tri-als in countries where a drug is approved, when the true average effectiveness is ES = 0.20 and the standard deviation (SD) of the estimated ES across trials is 0.20, 0.40, and 0.60Figure 1
Anticipated mean and range of average effect size (ES) for tri-
als in countries where a drug is approved, when the true
average effectiveness is ES = 0.20 and the standard deviation
(SD) of the estimated ES across trials is 0.20, 0.40, and 0.60.
Trials are assumed to be of similar size and similar weight in
the calculations. Forty approval packages with 5 trials each
have been simulated in each of the three settings and the
data show the ES in the successful packages (those where the
average ES is at least 0.20).
Anticipated effect sizes in approved settings
(average and range)
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 5 of 9
(page number not for citation purposes)
hundreds of participants. Nevertheless, these trials have
used outcomes that can pick formally statistically signifi-
cant differences between the compared arms even with
such small sample sizes. The typical choice is depression
scales such as the Hamilton Rating Scale of Depression.
Such continuous outcomes can show formally statistically
significant results (p < 0.05) even for differences that are
small and trivial. Statistical significance is confused with
clinical significance. A consensus by the National Institute
for Clinical Excellence (NICE) has suggested that at least a
3 point difference is needed in the Hamilton scale or
equivalently ES = 0.50 to claim a clinically important
effect [9]. Nevertheless, the results of several single trials
with statistically significant results and the results of prac-
tically all meta-analyses with statistically significant
results exclude that and ES = 0.50 can be conferred by anti-
depressants, when we examine the 95% confidence inter-
vals of the effects. Here one should acknowledge that
there is nothing absolute about the cut-off of ES = 0.50
and some investigators may disagree with this cut-off.
Typically proposed thresholds in the literature for small,
moderate and large effects for continuous outcomes are
0.2, 0.5, and 0.8 standard deviations, but even this is arbi-
trary [18].
Study design and selection of study populations
The industry of randomized trials of antidepressants has
generated over the years a long chain of practices and
design "standards" that aims to maximize the chances of
showing larger benefits from given drugs. These practices
and standards include, but are not limited to, the use of
placebo-controls, placebo lead-in periods, and a set of
exclusion criteria for recruited participants.
Placebo-controls are of course dictated by regulatory agen-
cies, and there is some considerable justification for them.
Empirical data, even before the meta-analyses discussed
above, have repeatedly shown that the placebo effect in
depressive symptoms is large, variable, and seemingly
increasing in magnitude in more recent trials [19]. How-
ever, this also generates the paradox that although we
acknowledge, approve and massively sell drugs that we
consider to be more effective than placebo, we continue to
perform trials that require some of the participants to take
what we consider to be ineffective treatments [20]. As a
compromise, some trials randomize patients 2:1 or even
3:1, 4:1 and 5:1 to active drug versus placebo, a contradic-
tion to the sense of equipoise that should permeate the
ethics of running a clinical trial. Obviously, the placebo-
controlled trials have a better chance of showing larger
benefits in terms of absolute effects; head-to-head com-
parisons typically show no or very small differences in
overall efficacy of one antidepressant over the other,
regardless of whether old or new agents are involved [21-
23]. For example, a meticulous systematic review of fluox-
etine versus other antidepressants concluded than even if
some nominally significant results were seen in some
comparisons, "the clinical meaning of these differences is
uncertain, and no definitive implications for clinical prac-
tice can be drawn" [21].
The use of placebo lead-in excludes patients who show a
good response to placebo over a brief period of time,
before participating in the proper trial. In theory, this may
inflate the difference between drug and placebo in the
patients who are eventually enrolled in the trial, if the
good response to placebo in the lead-in period correlates
with good response to placebo also during the longer fol-
low-up of the trial. The presence and strength of this cor-
relation is debatable [24].
Finally, the corpus of antidepressant trials has silently
adopted a series of exclusion criteria. An empirical evalu-
ation found that common exclusion criteria are short-epi-
sode duration, mild severity of illness, psychiatric
comorbidities, long duration of illness, medical comor-
bidities, and prior non-response to treatment [24]. Actu-
ally, there is little empirical evidence that any of these
criteria, perhaps with the exception of short-episode dura-
tion, does affect the magnitude of the net treatment effect.
Conversely, they all diminish the generalizability of the
trial findings. Based on their extensive sales, we have to
infer that antidepressants are used far more widely in clin-
ical practice compared with the narrow clinical trial set-
ting defined by such restrictive eligibility criteria.
Overall, we know little about how design features can
influence results. For example, even the dosing schedule
(flexible versus fixed), the number of treatment arms, and
the percentage of female patients have been reported to be
associated with the magnitude of the treatment effect in
trials [25].
Short follow-up
Most antidepressant trials have limited duration of fol-
low-up, typically 6 weeks, and rarely exceeding 8 weeks.
Some trials even last only 3–4 weeks. Even with such short
follow-up, attrition (losses to follow-up, and/or discon-
tinuation of study medication) is very common in these
trials [26]. Imputing outcomes when information is miss-
ing due to attrition is not easy and leaves room for bias.
It is mostly differences in long-term, hard clinical out-
comes that would matter (suicide, loss of job, other major
personal or social events), because these drugs are some-
times used by patients in the community on a far more
long-term basis. The evidence on long-term maintenance
has been reviewed by a meta-analysis of trials where
patients responding at the acute phase have been rand-
omized to maintenance versus placebo [27]. While large
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 6 of 9
(page number not for citation purposes)
reduction in the odds of relapse has been demonstrated,
these 31 trials are still small (total n = 4410), most do not
exceed 12 months of follow-up, they focus on those who
responded acutely, and have not been able to address
other hard outcomes of depression.
Selective and distorted reporting
As demonstrated by the New England Journal of Medicine
meta-analysis [7], selective and distorted reporting of
results is a major problem in the antidepressant literature.
Even with all the manipulations listed above and with
potentially selective analyses, only half of the trials of
antidepressants reach conventional statistical signifi-
cance. The other half either disappear or are further dis-
torted so as to be published with the impression that they
also have found "positive" results. Meta-analyses of the
published literature are thus likely to give misleading
impressions about the effectiveness of these drugs.
Unknown harms
Antidepressant trials are not geared towards demonstrat-
ing the possible harms of these medications. The imbal-
ance of emphasis between effectiveness and harms in the
design and reporting of randomized trials has been
repeatedly demonstrated in various medical specialties
[28], including mental health interventions [29]. Small
trials are unlikely to pick any major harms, even relatively
common ones, let alone uncommon harms that are life-
threatening and may lead to death. Antidepressants are
thus licensed in almost perfect vacuum on harms infor-
mation. Harms may be detected subsequently from meta-
analyses, large registries, or other means of post-market-
ing surveillance, but all of these methods have limitations
and imperfect sensitivity and specificity. The large debate
about the unrecognized suicide risk for children taking
antidepressants is one example about how very important
harms can go unrecognized [30-32]. The late addition of
black box warnings increases the sense of uncertainty
about these drugs [33]. Uncertainty about harms does not
help anyone, even the industry. The reputation of the
industry can be ruined even by postulated harms that may
not exist, e.g. as in the debate about increase in suicides in
Extension of the market
As described above, clinical trials of antidepressants enrol
highly restricted types of patient populations, and typi-
cally employ short-term administration. However, once
licensed, the antidepressant drugs are used widely in the
general population and are often prescribed for very long-
term use [1-3]. Antidepressants are a prime example of the
over-medicalization of our society [1]. Direct to consumer
advertisement probably contributes extensively to this dif-
fusion of use. A pharmacoeconomic evaluation found
that direct to consumer advertisement of antidepressants
may result in treating people so widely that 94% of anti-
depressant use due to direct-to-consumer advertising is
from non-depressed individuals [34]. The same evalua-
tion concluded that this is more than counterbalanced by
the accumulated societal benefits conferred by the treat-
ment of truly depressed patients who would not have
been treated otherwise [34]. However, this inference is
based on an assumption of large effectiveness of antide-
pressants that is probably not commensurate with the cur-
rent evidence.
Building the supporting scientific myth
Besides randomized trials, the industry has used biologi-
cal arguments to promote the idea in the wider commu-
nity that antidepressants have mechanisms of action that
correct major chemical imbalances in the brain [1]. The
typical example is SSRI marketing based on the serotonin
hypothesis. It is unfair or even wrong to summarize dec-
ades of neuroscience research as showing that depression
is caused by the imbalance of one or another neurotrans-
mitter (serotonin, norepinephrine, or other) in the brain.
If anything, evolving knowledge of biology shows that
depression is an extremely complex behavioural pheno-
type regulated by a large number of biological pathways,
external exposures, and genetic factors, each one contrib-
uting a small effect [1,35]. Drugs that supposedly impact
on a single pathway are expected to have small impact on
the overall biology of depression and would be equili-
brated by balancing changes. Clinical effects would thus
also be expected to be small on average.
Answering the main practical questions
Are antidepressants indicated in depression?
Based on the above considerations, antidepressants are
probably indicated only in select patients with major
depression, probably preferentially in those who have
severe symptoms and have not responded to anything
else. For most patients with some depressive symptoms
who are currently taking antidepressants, using these
drugs would not have been the preferred option, placebo
would be practically as good, if not better, and would save
toxicities and cost.
Current approval of specific antidepressants by regulatory
agencies means that some specific criteria have been met
that demonstrate some nominally statistically significant
results in some trials and for some scales, but this is not
equivalent to proof of major clinical benefit, effectiveness
in the wider population level. The trials that are under-
taken are not necessarily scientific exercises but rather
exercises that are designed to produce a demonstration of
a particular effect for regulatory/legal purposes.
Some other meta-analysts have found even more gloomy
results than the two meta-analyses reviewed above. For
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 7 of 9
(page number not for citation purposes)
example, after including both published and unpublished
data (29 and 11 trials, respectively), one meta-analysis
recently found absolutely no benefit from paroxetine over
placebo when focusing on the "hard" clinical outcome of
treatment discontinuation [36]. Some treatment effect
probably exists in "softer" scale-based outcomes, but it is
likely to be small, and possibly the effect estimates are
inflated compared to the truth. Cumulative meta-analyses
suggest that for drug treatments in mental health indica-
tions in general, treatment effects tend to decrease over
time, as more studies are conducted [37]. Empirical evi-
dence suggests also that much of the placebo effect in anti-
depressant trials is related simply to the number of patient
visits during a trial [38]. Trials where patients are sched-
uled to visit physicians more often for evaluation during
the study improve more, regardless of whether they are
given drug or placebo. There is a 0.6–0.9 point improve-
ment in the Hamilton scale for each additional visit per-
formed during the trial.
Perhaps most people given antidepressants for depressive
symptoms would just need some attention from their
physician and people to talk to and take some care of
them. Antidepressants may be covering largely the lost
placebo of human interaction and patient-physician inter-
action that has become so sparse in modern society. How-
ever, I will not comment here whether formal
psychotherapies are as or more effective than antidepres-
sants. The evidence base of psychotherapies is not any bet-
ter, it pertains to similarly small trials, affected by most of
the same problems (and more), and it would need a
whole separate review on its own [39].
Is it unethical to kill a living myth?
If most of the antidepressant efficacy reflects simply the
placebo effect, and if most people just benefit as much as
the placebo effect allows, is it unethical to kill a living
myth? One might argue that if the general population is
informed that antidepressants are not really effective, this
might demolish the benefits that we get from the placebo
effect when we administer these drugs. However, is it not
unethical to lie to patients that an intervention is effective
when it is not? Moreover, if we want to utilize this placebo
effect, why is it justified that this should cost cumulatively
more billion dollars to society than almost any other
(truly effective) pharmaceutical intervention for any other
condition? It would be weird for our society to compro-
mise with a view that someone should make fortunes by
selling official placebos.
The pharmaceutical industry is going through rough times
[40]. As the exclusive licenses of most major blockbusters
are expiring currently, the fear is that unless the big
pharma manages to keep up in its profits, then its R&D
effort will be stalled. An even more cynical view is that
since our society aims at making profit, if we don't allow
the pharmaceutical industry to make large profits, even by
selling largely ineffective drugs to the population at-large,
then some other industry will attract capital investments
and will flourish instead. This may be in an entrepreneur-
ial domain that is less useful or even very harmful for soci-
ety, e.g. the weapons industry. My reply is that the
pharmaceutical industry should be encouraged and sup-
ported to conduct high-quality research in antidepressant
drugs and beyond. Serious investigation in the life sci-
ences by corporate, government and academic investiga-
tors should be given priority as societal goals, but
expectations should be realistic: most major promises do
not materialize [41], major discoveries are rare, drugs are
not miraculous, and we should be honest about how
much has been achieved and how much we can achieve.
Telling lies to compete against more unscrupulous pro-
ducers of lies in our society is not the way to make scien-
tific progress.
After the publication of the PLoS Medicine meta-analysis,
Eli Lilly, the producers of fluoxetine made a statement
pointing attention to the fact that "more than 40 million
people suffering from depression have been treated with
fluoxetine in over 100 countries" [42]. Fluoxetine received
approval from FDA based on 5 registered trials where a
total of 817 patients with depression took fluoxetine [4].
A simple calculation shows that for each randomized
patient receiving fluoxetine in the FDA application pack-
age, approximately 50,000 patients have been treated sub-
sequently in everyday practice. Even if we consider all the
other trials of fluoxetine (pre- and post-approval and with
various indications), the ratio is likely to remain way
above 1,000 to 1. Similar considerations apply practically
to all antidepressants. While we have a large number of
trials, they are all small. Typically these trials do not pro-
vide sufficient evidence to determine the exact benefit
with any accuracy in specific types of patients, e.g. as a
function of the baseline risk of depression [43]. Moreover,
they say little about hard clinical outcomes and the long-
term benefit-harm ratio.
No matter how careful the efforts to collate small trials,
the interpretation of the available fragmented evidence
will remain largely subjective. For example, the authors of
the New England Journal of Medicine meta-analysis wrote
an editorial in the BMJ where they said that compared to
the more negative stance of the PLoS Medicine meta-anal-
ysis, their stance was more critical rather than dismissive
[44]. To settle this important matter with robust evidence,
we need large trials with 100-fold more patients than
what has been the norm to-date. These trials should be
linked to long-term follow-up registries with thorough
recording of long-term outcomes for both efficacy and
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 8 of 9
(page number not for citation purposes)
harms. Efficacy outcomes should include long-term man-
agement and evolution of depression, suicide and major
life events. These trials should be designed in a way that
they do not simply have power to show a nominally sta-
tistically significant benefit (of debatable clinical impor-
tance), but for demonstrating whether the benefit is
conclusively within the range of clinical importance and
has a long-term portend. These trials should also have suf-
ficient power to answer this question not only in the over-
all study population, but also in specific risk strata and as
a function of the baseline risk of severity. When hundreds
of millions have been treated with these drugs with such
a suboptimal evidence base, asking to conduct mega-trials
of 50,000 participants for each antidepressant agent is not
an exaggerated wish. If there are concerns that different
agents or different classes thereof may have different
effects, then large-scale evidence should be collected for
each of them and for head-to-head comparisons thereof.
The general public should be sensitized not to compro-
mise with suboptimal evidence for drugs that are used by
hundreds of millions of people.
Antidepressant trials should be pre-registered before their
conduct [45], but this alone does not suffice. Their proto-
cols need to be entirely transparent and their analysis
plans explicit in detail upfront. There should be no room
for flexibility in the collected data and performed analy-
ses. The full data should be available in public, for scru-
tiny and use by all interested scientists. Finally, the
complete agenda of these trials should be co-ordinated in
a way that they will eventually be part of an overarching
global prospective meta-analysis of individual-level data.
Simply waiting to perform meta-analyses after the fact
with data that are selectively analyzed, presented, submit-
ted, and published is an invitation to confusion. Under
such circumstances, meta-analyses end up being mostly
detective tools that struggle to unearth problems under
dark circumstances. Conversely, full registration of large-
scale transparent trials with full detailed availability of
data and prospective meta-analyses would show that we
are serious about finding the true merits and risk-benefit
of antidepressants or any other type of medical interven-
Clinical trial registration has made major progress [45],
but details of protocols are still missing and flexibility and
selective analyses and reporting are probably still preva-
lent. This does not mean that this vision is utopian. In
fact, it may be the only way to avoid running into vicious
circles of pseudo-evidence-based medicine where we are
blinded by a blizzard of small, selectively designed, ana-
lyzed and reported trials. This may sound depressing for
the current state of clinical trials research on antidepres-
sants and beyond. Nevertheless, even if one feels a bit
depressed by this state of affairs, there is no reason to take
antidepressants, they probably won't work.
Competing interests
The author declares that he has no competing interests.
Authors' contributions
JI conceived this article, carried out the included analyses,
and wrote the paper.
About the author: John PA Ioannidis, MD, PhD, born in
New York, NY in 1965, is Professor and Chairman of the
Department of Hygiene and Epidemiology, University of
Ioannina School of Medicine and has an adjunct appoint-
ment as Professor of Medicine at Tufts University School
of Medicine. He is interested in research methodology,
evidence-based medicine, clinical and molecular epide-
miology, and mathematical modelling. He has published
approximately 400 peer-reviewed publications and is a
member of the editorial board of 18 peer-reviewed jour-
I am grateful to Professor Toshi Furukawa for his very interesting com-
1. Lacasse JR, Leo J: Serotonin and depression: a disconnect
between the advertisements and the scientific literature.
PLoS Med 2005, 2(12):e392.
2. []. last accessed
March 8, 2007
3. Greenberg PE, Kessler RC, Birnbaum HG, Leong SA, Lowe SW, Ber-
glund PA, Corey-Lisle PK: The economic burden of depression
in the United States: how did it change between 1990 and
2000? J Clin Psychiatry 2003, 64(12):1465-75.
4. Psi-Tri database []. last
accessed March 8, 2007
5. Fergusson D, Doucette S, Glass KC, Shapiro S, Healy D, Hebert P,
Hutton B: Association between suicide attempts and selective
serotonin reuptake inhibitors: systematic review of ran-
domised controlled trials. BMJ 2005, 330(7488):396.
6. FDA briefing document for Dec. 13, 2006, meeting of Psychopharmacologic
Drugs Advisory Committee 2006 [
ac/06/briefing/2006-4272b1-index.htm]. Rockville (MD): US FDA,
Center for Drug Evaluation and Research last accessed March 8, 2007
7. Turner EH, Matthews AM, Linardatos E, Tell RA, Rosenthal R: Selec-
tive publication of antidepressant trials and its influence on
apparent efficacy. N Engl J Med 2008, 358(3):252-60.
8. Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, John-
son BT: Initial severity and antidepressant benefits: a meta-
analysis of data submitted to the Food and Drug Administra-
tion. PLoS Med 2008, 5(2):e45.
9. National Institute for Clinical Excellence: Depression: manage-
ment of depression in primary and secondary care. Clinical
practice guideline No 23. London: NICE; 2004.
10. Beunckens C, Molenberghs G, Kenward MG: Direct likelihood
analysis versus simple forms of imputation for missing data
in randomized clinical trials. Clin Trials 2005, 2(5):379-86.
11. Lau J, Ioannidis JP, Terrin N, Schmid CH, Olkin I: The case of the
misleading funnel plot. BMJ 2006, 333(7568):597-600.
12. Ioannidis JP: Why most published research findings are false.
PLoS Med 2005, 2(8):e124.
13. Ioannidis JP: Why most discovered true associations are
inflated. Epidemiology 2008 in press.
14. Higgins JP, Thompson SG: Controlling the risk of spurious find-
ings from meta-regression. Stat Med 2004, 23(11):1663-82.
Publish with BioMed Central and every
scientist can read your work free of charge
"BioMed Central will be the most significant development for
disseminating the results of biomedical research in our lifetime."
Sir Paul Nurse, Cancer Research UK
Your research papers will be:
available free of charge to the entire biomedical community
peer reviewed and published immediately upon acceptance
cited in PubMed and archived on PubMed Central
yours — you keep the copyright
Submit your manuscript here:
Philosophy, Ethics, and Humanities in Medicine 2008, 3:14
Page 9 of 9
(page number not for citation purposes)
15. Thompson SG, Higgins JP: How should meta-regression analyses
be undertaken and interpreted? Stat Med 2002, 21(11):1559-73.
16. Schmid CH, Stark PC, Berlin JA, Landais P, Lau J: Meta-regression
detected associations between heterogeneous treatment
effects and study-level, but not patient-level, factors. J Clin
Epidemiol 2004, 57(7):683-97.
17. Berlin JA, Santanna J, Schmid CH, Szczech LA, Feldman HI, Anti-Lym-
phocyte Antibody Induction Therapy Study Group: Individual
patient- versus group-level data meta-regressions for the
investigation of treatment effect modifiers: ecological bias
rears its ugly head. Stat Med 2002, 21(3):371-87.
18. Cohen J: Statistical power analysis for the behavioral sciences.
2nd edition. New York: Lawrence Erlbaum Associates; 1988.
19. Walsh BT, Seidman SN, Sysko R, Gould M: Placebo response in
studies of major depression: variable, substantial, and grow-
ing. JAMA 2002, 287(14):1840-7.
20. Lilford RJ: Ethics of clinical trials from a bayesian and decision
analytic perspective: whose equipoise is it anyway? BMJ 2003,
21. Cipriani A, Brambilla P, Furukawa T, Geddes J, Gregis M, Hotopf M,
Malvini L, Barbui C: Fluoxetine versus other types of pharma-
cotherapy for depression. Cochrane Database Syst Rev
22. MacGillivray S, Arroll B, Hatcher S, Ogston S, Reid I, Sullivan F, Wil-
liams B, Crombie I: Efficacy and tolerability of selective serot-
onin reuptake inhibitors compared with tricyclic
antidepressants in depression treated in primary care: sys-
tematic review and meta-analysis. BMJ 2003, 326(7397):1014.
23. Gartlehner G, Hansen RA, Thieda P, DeVeaugh-Geiss AM, Gaynes
BN, Krebs EE, Lux LJ, Morgan LC, Shumate JA, Monroe LG, Lohr KN:
Comparative effectiveness of second-generation antidepres-
sants in the pharmacologic treatment of adult depression.
Rockville, MD: AHRQ Publication No. 07-EHC007-EF; 2007.
24. Posternak MA, Zimmerman M, Keitner GI, Miller IW: A reevalua-
tion of the exclusion criteria used in antidepressant efficacy
trials. Am J Psychiatry 2002, 159(2):191-200.
25. Khan A, Kolts RL, Thase ME, Krishnan KR, Brown W: Research
design features and patient characteristics associated with
the outcome of antidepressant clinical trials. Am J Psychiatry
2004, 161(11):2045-9.
26. Barbui C, Hotopf M, Freemantle N, Boynton J, Churchill R, Eccles MP,
Geddes JR, Hardy R, Lewis G, Mason JM: Treatment discontinua-
tion with selective serotonin reuptake inhibitors (SSRIs) ver-
sus tricyclic antidepressants (TCAs). Cochrane Database Syst Rev
27. Geddes JR, Carney SM, Davies C, Furukawa TA, Kupfer DJ, Frank E,
Goodwin GM: Relapse prevention with antidepressant drug
treatment in depressive disorders: a systematic review. Lan-
cet 2003, 361(9358):653-61.
28. Ioannidis JP, Lau J: Completeness of safety reporting in rand-
omized trials: an evaluation of 7 medical areas. JAMA 2001,
29. Papanikolaou PN, Churchill R, Wahlbeck K, Ioannidis JP: Safety
reporting in randomized trials of mental health interven-
tions. Am J Psychiatry 2004, 161(9):1692-7.
30. Hetrick S, Merry S, McKenzie J, Sindahl P, Proctor M: Selective
serotonin reuptake inhibitors (SSRIs) for depressive disor-
ders in children and adolescents. Cochrane Database Syst Rev
31. Bridge JA, Iyengar S, Salary CB, Barbe RP, Birmaher B, Pincus HA, Ren
L, Brent DA: Clinical response and risk for reported suicidal
ideation and suicide attempts in pediatric antidepressant
treatment: a meta-analysis of randomized controlled trials.
JAMA 2007, 297:1683-96.
32. Dubicka B, Hadley S, Roberts C: Suicidal behaviour in youths
with depression treated with new-generation antidepres-
sants: meta-analysis. Br J Psychiatry 2006, 189:393-8.
33. Friedman RA, Leon AC: Expanding the black box – depression,
antidepressants, and the risk of suicide. N Engl J Med 2007,
34. Block AE: Costs and benefits of direct-to-consumer advertis-
ing: the case of depression. Pharmacoeconomics 2007,
35. López-León S, Janssens AC, González-Zuloeta Ladd AM, Del-Favero
J, Claes SJ, Oostra BA, van Duijn CM: Meta-analyses of genetic
studies on major depressive disorder. Mol Psychiatry in press.
2007 Oct 16
36. Barbui C, Furukawa TA, Cipriani A: Effectiveness of paroxetine in
the treatment of acute major depression in adults: a system-
atic re-examination of published and unpublished data from
randomized trials. CMAJ 2008, 178:296-305.
37. Trikalinos TA, Churchill R, Ferri M, Leucht S, Tuunainen A, Wahlbeck
K, Ioannidis JP, EU-PSI project: Effect sizes in cumulative meta-
analyses of mental health randomized trials evolved over
time. J Clin Epidemiol 2004, 57(11):1124-30.
38. Posternak MA, Zimmerman M: Therapeutic effect of follow-up
assessments on antidepressant and placebo response rates
in antidepressant efficacy trials: meta-analysis. Br J Psychiatry
2007, 190:287-92.
39. Watanabe N, Hunot V, Omori IM, Churchill R, Furukawa TA: Psy-
chotherapy for depression among children and adolescents:
a systematic review. Acta Psychiatr Scand 2007, 116(2):84-95.
40. Cuatrecasas P: Drug discovery in jeopardy. J Clin Invest 2006,
41. Contopoulos-Ioannidis DG, Ntzani E, Ioannidis JP: Translation of
highly promising basic science research into clinical applica-
tions. Am J Med 2003, 114(6):477-84.
42. Mayor S: Study shows difference between antidepressants
and placebo is significant only in severe depression. BMJ 2008,
43. Kent DM, Hayward RA: Limitations of applying summary
results of clinical trials to individual patients: the need for
risk stratification. JAMA 2007, 298(10):1209-12.
44. Turner EH, Rosenthal R: Efficacy of antidepressants. BMJ 2008,
45. Rennie D: Trial registration: a great idea switches from
ignored to irresistible. JAMA 2004, 292(11):1359-62.
... Naturally, the pressure of producing positive results this way may be most pronounced among trialists with financial interests in the treatment. Antidepressants are a commonly named example here [102], but similar conflicts of interest may also pertain to, e.g., the blooming "digital therapeutics" industry [103], who also need to show that their treatment is effective to sell it. The best way to circumvent these issues is to register a detailed protocol before the beginning of the trial in one of the primary registries listed by the WHO International Clinical Trials Registry Platform (ICTRP) and to analyze and report results in accordance with the original registration. ...
Full-text available
Background Considered one of the highest levels of evidence, results of randomized controlled trials (RCTs) remain an essential building block in mental health research. They are frequently used to confirm that an intervention “works” and to guide treatment decisions. Given their importance in the field, it is concerning that the quality of many RCT evaluations in mental health research remains poor. Common errors range from inadequate missing data handling and inappropriate analyses (e.g., baseline randomization tests or analyses of within-group changes) to unduly interpretations of trial results and insufficient reporting. These deficiencies pose a threat to the robustness of mental health research and its impact on patient care. Many of these issues may be avoided in the future if mental health researchers are provided with a better understanding of what constitutes a high-quality RCT evaluation. Methods In this primer article, we give an introduction to core concepts and caveats of clinical trial evaluations in mental health research. We also show how to implement current best practices using open-source statistical software. Results Drawing on Rubin’s potential outcome framework, we describe that RCTs put us in a privileged position to study causality by ensuring that the potential outcomes of the randomized groups become exchangeable. We discuss how missing data can threaten the validity of our results if dropouts systematically differ from non-dropouts, introduce trial estimands as a way to co-align analyses with the goals of the evaluation, and explain how to set up an appropriate analysis model to test the treatment effect at one or several assessment points. A novice-friendly tutorial is provided alongside this primer. It lays out concepts in greater detail and showcases how to implement techniques using the statistical software R, based on a real-world RCT dataset. Discussion Many problems of RCTs already arise at the design stage, and we examine some avoidable and unavoidable “weak spots” of this design in mental health research. For instance, we discuss how lack of prospective registration can give way to issues like outcome switching and selective reporting, how allegiance biases can inflate effect estimates, review recommendations and challenges in blinding patients in mental health RCTs, and describe problems arising from underpowered trials. Lastly, we discuss why not all randomized trials necessarily have a limited external validity and examine how RCTs relate to ongoing efforts to personalize mental health care.
... Different classes of antidepressants have been used to treat MDD typically, such as selective serotonin reuptake inhibitors (SSRIs), serotonin noradrenaline reuptake inhibitors (SNRIs), tricyclic antidepressants (TCAs) and others (Cipriani et al., 2018). Despite this, there is still considerable debate surrounding their efficacy and effectiveness (Ioannidis, 2008;Cipriani et al., 2018), particularly given that approximately ~50% of patients do not achieve remission with first-line therapy (Caldiroli et al., 2021). Consequently, the need to discover novel antidepressants with superior effectiveness and tolerability in treating MDD is crucial. ...
Full-text available
Introduction Major depressive disorder (MDD) is a highly prevalent and burdensome condition. This study aims to evaluate the effectiveness, tolerability, and safety of vortioxetine in treating MDD based on real-world data. Methods A systematic search of eight electronic databases was performed from inception until October 2022 to identify real-world studies, excluding randomized controlled trials. We conducted subgroup, meta-regression, sensitivity analyses, publication bias and quality assessments using the random-effects model. The effects were summarized by rates or standardized mean difference (SMD) with 95% confidence intervals (CI). Results Of the 870 records identified, 11 studies (3139 participants) and 10 case reports or series were eligible for inclusion. Vortioxetine significantly relieved depression symptoms as assessed by both patients (SMD 2.25, 95% CI: 1.60−2.89) and physicians (SMD 3.73, 95% CI: 2.78−4.69). Cognitive function (1.86, 1.11−2.62) and functional disability (1.71, 1.14−2.29) were similarly markedly improved. Subgroup and meta-regression analyses showed that geographic location and medication regimen (whether combined with other antidepressants) were crucial factors influencing the effectiveness, potentially contributing to significant heterogeneity. The estimated response and remission rates were 66.4% (95% CI: 51.2%−81.5%) and 58.0% (48.9%−67.1%), respectively. Vortioxetine was well-tolerated with a pooled dropout rate of 3.5% (1.8%−5.8%), and the most common adverse event was nausea, with an estimated rate of 8.9% (3.8%−15.8%). Limitations The study has some limitations, including significant heterogeneity and limited evidence for some outcomes. Conclusions Vortioxetine is effective, well-tolerated, and safe for treating MDD in clinical practice, with significant improvements observed in depressive severity, cognitive function, and functioning. Future studies should directly compare vortioxetine with other antidepressants in real-world settings to further evaluate its clinical utility.
... antidepressants with different mechanisms are a milestone in treatment progress (4). Fully functional recovery, which is the ultimate treatment goal for patients with MDD, may be unsuccessful in some patients. ...
Full-text available
Objective This study evaluated the treatment outcomes of agomelatine on anhedonic state, anxiety/somatic symptoms, and sexual function in Chinese patients with major depressive disorder (MDD). Method In total, 93 adult patients with MDD were enrolled, and 68 of them were included in a prospective, open-label, multicenter clinical study. All patients received agomelatine monotherapy during a 9-week treatment phase. The effectiveness of the treatment was reflected by the improvement of anhedonia and somatic symptoms based on the 17-item Hamilton Depression Rating Scale (HAMD-17). In addition, the Arizona Sexual Dysfunction Scale (ASEX), Sheehan Disability Scale (SDS), and Short Form of Quality-of-Life Enjoyment and Satisfaction Questionnaire (Q-LES-Q-SF) were administered to all participants at baseline and at the 3-, 6-, and 9-week follow-ups. Results After 9 weeks of treatment with agomelatine, the response and remission rates were 73.5% and 39.7%, respectively. Somatic symptoms significantly improved at week 9 ( p < 0.001), and significant effects were also observed on the HAMD anhedonia items ( p < 0.001). The patients exhibited lower levels of disease severity (the SDS score dropped from 15.52 ± 4.7 to 7.09 ± 5.62 at week 9; the ASEX score dropped from 21.89 ± 4.06 to 16.19 ± 4.79, p < 0.001) and higher levels of QOL (the Q-LES-Q-SF score dropped from 41.02 ± 5.99 to 50.49 ± 8.57, p < 0.001) during the follow-up. Furthermore, treatment with agomelatine improved depressive symptoms without causing serious adverse events. Conclusion These analyses indicate that agomelatine is a treatment option for improving anhedonic status, anxiety/somatic symptoms, and sexual dysfunction in MDD patients.
... Anti-depressants have been questioned regarding their long-term efficacy and safety for thousands of patients [21,22]. However, they can be helpful for some patients by reducing depressive symptoms and lowering the likelihood of relapse. ...
Full-text available
The present meta-analysis has been conducted to review currently available literature to examine the factors associated with adherence to anti-depressant medications in adults. This meta-analysis and systematic review followed the MOOSE (Meta-analysis of Observational Studies in Epidemiology) guidelines. According to this analysis, the three most important electronic resources for research were CINAHL, EMBASE, and Medline. Google Scholar was used to supplementing the articles already available for review. Keywords used to find relevant articles included "predictors," "non-adherence," "anti-depressants," and "adults." Medical subject headings (MeSH) terms and Boolean operators ("AND" and "OR") were used in the search strategy to refine the search further. Studies included in this meta-analysis had information on factors associated with non-adherence to anti-depressant medication. The study evaluated samples of adult participants over 18 years with a diagnosis of depression and who had been prescribed anti-depressants. In conclusion, this meta-analysis examined the relationship between demographic factors and non-adherence to anti-depressant medications. The findings revealed that gender, educational status, income level, marital status, and area of residence did not significantly predict non-adherence to anti-depressants. However, older age and polypharmacy were significant predictors of adherence to anti-depressants. The study also found that individuals living in urban areas were more likely to adhere to anti-depressants, but the difference was not statistically significant.
... However, increasing evidence has suggested that pharmacotherapy may have minimal or non-existence effect on mild or moderate depression (Ferrando, 2009;Fournier et al., 2010) and may even lead to potential somatic distress, drug abuse, and lethality in overdose (European AIDS Clinical Society, 2022;Ferrando, 2009). Pharmacotherapy may also be complicated by HIV illness and drug interactions with antiretroviral therapy (Arseniou et al., 2014;Ferrando, 2009), while the long-term balance of benefits and harms of pharmacotherapy is generally understudied (Ioannidis, 2008). These considerations have led to a growing interest in non-pharmacological treatments for depression in people living with HIV. ...
Full-text available
Background: Treatment for depression in people living with HIV has increasingly turned to non-pharmacological treatments due to the adverse reactions of pharmacotherapy. However, it remains unclear which non-pharma- cological treatment is the most effective and acceptable for depression in people living with HIV. Objective: To compare and rank the efficacy and acceptability of different non-pharmacological treatments for de- pression in people living with HIV. Design: A systematic review and Bayesian network meta-analysis. Methods: We systematically searched PubMed, EMBASE, the Cochrane Central Register of Controlled Trials, PsycArticles, CINAHL, ProQuest, OpenGrey, and international trial registers for published and unpublished studies from their inception to September 1, 2022, and searched key conference proceedings from January 1, 2020, to September 25, 2022. We searched for randomized controlled trials of any non-pharmacological treatments for depression in adults living with HIV (≥ 18 years old). Primary outcomes were efficacy (mean change scores in de- pression) and acceptability (all-cause discontinuation). We used a random-effects network meta-analysis model to synthesize all available evidence. The methodological quality of the included studies was assessed using the Cochrane Collaboration Risk of Bias Tool. We registered this study in PROSPERO, number CRD42021244230. Results: A total of 53 randomized controlled trials were included in this network meta-analysis involving seven non-pharmacological treatments for depression in people living with HIV. For efficacy, mind–body therapy, in- terpersonal psychotherapy, cognitive-behavioral therapy, supportive therapy, and education were significantly more effective than most control conditions (standardized mean differences ranged from −0.96 to −0.36). Rankings probabilities indicated that mind–body therapy (79%), interpersonal psychotherapy (71%), cognitive- behavioral therapy (62%), supportive therapy (57%), and education (57%) might be the top five most significantly effective treatments for depression in people living with HIV, in that order. For acceptability, only supportive therapy and interpersonal psychotherapy were significantly less acceptable than most control conditions (odds ratios ranged from 1.92 to 3.43). Rankings probabilities indicated that education might be the most acceptable treatment for people living with HIV (66%), while supportive therapy (26%) and interpersonal psy- chotherapy (10%) might rank the worst. The GRADE assessment results suggested that most results were rated as “moderate” to “very low” for the confidence of evidence. Conclusions: Our study confirmed the efficacy and acceptability of several non-pharmacological treatments for depression in people living with HIV. These results should inform future guidelines and clinical decisions for depression treatment in people living with HIV.
Target-based drug discovery is the dominant paradigm of drug discovery; however, a comprehensive evaluation of its real-world efficiency is lacking. Here, a manual systematic review of about 32000 articles and patents dating back to 150 years ago demonstrates its apparent inefficiency. Analyzing the origins of all approved drugs reveals that, despite several decades of dominance, only 9.4% of small-molecule drugs have been discovered through "target-based" assays. Moreover, the therapeutic effects of even this minimal share cannot be solely attributed and reduced to their purported targets, as they depend on numerous off-target mechanisms unconsciously incorporated by phenotypic observations. The data suggest that reductionist target-based drug discovery may be a cause of the productivity crisis in drug discovery. An evidence-based approach to enhance efficiency seems to be prioritizing, in selecting and optimizing molecules, higher-level phenotypic observations that are closer to the sought-after therapeutic effects using tools like artificial intelligence and machine learning.
Although the etiology of major depressive disorder remains poorly understood, reduced gamma oscillations is an emerging biomarker. Olfactory bulbectomy, an established model of depression that reduces limbic gamma oscillations, suffers from non-specific effects of structural damage. Here, we show that transient functional suppression of olfactory bulb neurons or their piriform cortex efferents decreased gamma oscillation power in limbic areas and induced depression-like behaviors in rodents. Enhancing transmission of gamma oscillations from olfactory bulb to limbic structures by closed-loop electrical neuromodulation alleviated these behaviors. By contrast, silencing gamma transmission by anti-phase closed-loop stimulation strengthened depression-like behaviors in naive animals. These induced behaviors were neutralized by ketamine treatment that restored limbic gamma power. Taken together, our results reveal a causal link between limbic gamma oscillations and depression-like behaviors in rodents. Interfering with these endogenous rhythms can affect behaviors in rodent models of depression, suggesting that restoring gamma oscillations may alleviate depressive symptoms.
Full-text available
Purpose At present, clinicians typically prescribe antidepressants based on the widely accepted “serotonin hypothesis.” This study explores an alternative mechanism, the stress mechanism, for selecting antidepressants based on patients’ medical history. Methods This study investigated clinicians’ prescribing patterns for the 15 most common antidepressants, including amitriptyline, bupropion, citalopram, desvenlafaxine, doxepin, duloxetine, escitalopram, fluoxetine, mirtazapine, nortriptyline, paroxetine, ropinirole, sertraline, trazodone, and Venlafaxine. The least absolute shrinkage and selection operator (LASSO) logistic regression was used to identify factors that affect the remission of depression symptoms after receiving an antidepressant. Results The study found that a wide range of factors influenced the propensity of clinicians to prescribe antidepressants, with the number of predictors ranging from 51 to 206 variables. The prevalence of prescribing an antidepressant ranged from 0.5% for doxepin to 24% for the combination of more than one antidepressant. The area under the receiver operating curves (AROC) ranged from 77.2% for venlafaxine to 90.5% for ropinirole, with an average AROC of 82% for predicting the propensity of medications. A variety of diagnoses and prior medications affected remission, in agreement that the central mechanism for the impact of medications on the brain is through stress reduction. For example, psychotherapy, whether done individually or in a group, whether done for a short or long time, and whether done with evaluation/assessment or not, had an impact on remission. Specifically, teenagers and octogenarians were less likely to benefit from bupropion, citalopram, escitalopram, fluoxetine, and sertraline compared to patients between 40 and 65 years old. The findings of this study suggest that considering a patient’s medical history and individual characteristics is crucial for selecting the most effective antidepressant treatment. Conclusions Many studies have raised doubt about the serotonin hypothesis as the central mechanism for depression treatment. The identification of a wide range of predictors for prescribing antidepressants highlights the complexity of depression treatment and the need for individualized approaches that consider patients’ comorbidities and previous treatments. The significant impact of comorbidities on the response to treatment makes it improbable that the mechanism of action of antidepressants is solely based on the serotonin hypothesis. It is hard to explain how comorbidities lead to the depletion of serotonin. These findings open up a variety of courses of action for the clinical treatment of depression, each addressing a different source of chronic stress in the brain. Overall, this study contributes to a better understanding of depression treatment and provides valuable insights for clinicians in selecting antidepressants based on patients’ medical history.
Full-text available
The authors examined which, if any, research design features and patient characteristics would significantly differ between successful and unsuccessful antidepressant trials. Clinical trial data were reviewed for nine antidepressants approved by the Food and Drug Administration between 1985 and 2000. From the antidepressant research programs on these medications, 52 clinical trials were included in the study. The authors evaluated trial design features, patient characteristics, and difference in response between placebo and antidepressant. Nine trial design features and patient characteristics were present in the research programs for all nine of the antidepressants. The severity of depressive symptoms before patient randomization, the dosing schedule (flexible versus fixed), the number of treatment arms, and the percentage of female patients were significantly associated with the difference in response to antidepressant and placebo. The duration of the antidepressant trial, number of patients per treatment arm, number of sites, and mean age of the patients were similar in successful trials (with a greater antidepressant-placebo difference) and less successful trials (with a smaller antidepressant-placebo difference). These findings may help in the design of future antidepressant trials.
Depressive disorders are common in young people and are associated with significant negative impacts. Selective serotonin reuptake inhibitors (SSRIs) are often used, however, evidence of their effectiveness in children and adolescents is not clear. Furthermore, there have been warnings against their use in this population due to concerns about increased risk of suicidal ideation and behaviour. To determine the efficacy and adverse outcomes, including definitive suicidal behaviour and suicidal ideation, of SSRIs compared to placebo in the treatment of depressive disorders in children and adolescents. We searched the CCDAN Trials Register, MEDLINE, PSYCHINFO and CENTRAL. Reference lists were checked, letters were sent to key researchers and internet databases searched. We included published and unpublished randomised controlled trials. Two or three review authors selected the trials, assessed the quality and extracted trial and outcome data. We used a fixed-effect meta-analysis. The relative risk was used to summarise dichotomous outcomes and the mean difference to summarise continuous measures. Twelve trials were eligible for inclusion, with ten providing usable data. At 8-12 weeks, there was evidence that children and adolescents 'responded' to treatment with SSRIs (RR 1.28, 95% CI 1.17 to 1.41). There was also evidence of an increased risk of suicidal ideation and behaviour for those prescribed SSRIs (RR 1.80, 95% CI 1.19 to 2.72). Fluoxetine was the only SSRI where there was consistent evidence from three trials that it was effective in reducing depression symptoms in both children and adolescents (CDRS-R treatment effect -5.63, 95% CI -7.38 to -3.88), and 'response' to treatment (RR 1.86, 95% CI 1.49 to 2.32). Where rates of adverse events were reported, this was higher for those prescribed SSRIs. Caution is required to interpret the results. First, there were methodological issues, including high attrition, issues regarding measurement instruments and clinical usefulness of outcomes, often variously defined across trials. Second, patients seen in clinical practice are likely to be more unwell, and at greater risk of suicide, compared to those in the trials, and it is unclear how this group would respond to SSRIs. This needs to be considered, along with the evidence of an increased risk of suicide related outcomes in those treated with SSRIs. It is unclear what the effect of SSRIs is on suicide completion. While untreated depression is associated with the risk of completed suicide and impacts on functioning, it is unclear whether SSRIs would modify this risk in a clinically meaningful way. Selective serotonin reuptake inhibitors (new generation antidepressants) for depressive disorders in children and adolescents Depressive disorders are common in young people and have significant negative impacts. Selective serotonin reuptake inhibitors (SSRIs) are commonly prescribed for the treatment of depressive disorder in children and adolescents. The review of 12 trials highlighted limitations with the data, making it difficult to answer questions about the effectiveness and safety of SSRIs in clinical practice. Overall, there was evidence of greater reduction in depressive symptoms to a predetermined level deemed a "response" on SSRI compared to placebo. However, response was variously defined across trials making interpretation of this outcome difficult. Fluoxetine was the only SSRI where there was consistent evidence from three trials showing that it was effective in reducing symptoms of depressive disorder in both children and adolescents. Those receiving fluoxetine had a greater improvement, scoring on average 5.63 lower on the Children's Depression Rating Scale-Revised (CDRS-R) scale (range 17-113) than those on placebo. It is unclear whether this small difference is a meaningful outcome for children and adolescents with depressive disorders. Nor is it apparent how children and adolescents with co-morbid conditions and at risk of suicide would respond to SSRIs, given this group were largely excluded from the trials. There is evidence that those prescribed SSRIs are at an increased risk of suicidal ideation and attempts (RR 1.80, 95% CI 1.19 to 2.72) consistent with a number of similar reviews in the area. Additionally, there was an increased risk of other adverse events. It is unclear how this relates to the risk of suicide completion. The trials were not designed to measure any of the suicide related outcomes adequately. At the same time, untreated depression is associated with the risk of completed suicide and impacts on academic and social functioning, however, it is not clear whether treatment with an SSRI will modify this risk in a clinically meaningful way for children and young people. Clinicians need to provide accurate information to children and adolescents and their families about the uncertainties regarding the benefits and risks of SSRI medication for depressive disorders.
Newly discovered true (non-null) associations often have inflated effects compared with the true effect sizes. I discuss here the main reasons for this inflation. First, theoretical considerations prove that when true discovery is claimed based on crossing a threshold of statistical significance and the discovery study is underpowered, the observed effects are expected to be inflated. This has been demonstrated in various fields ranging from early stopped clinical trials to genome-wide associations. Second, flexible analyses coupled with selective reporting may inflate the published discovered effects. The vibration ratio (the ratio of the largest vs. smallest effect on the same association approached with different analytic choices) can be very large. Third, effects may be inflated at the stage of interpretation due to diverse conflicts of interest. Discovered effects are not always inflated, and under some circumstances may be deflated-for example, in the setting of late discovery of associations in sequentially accumulated overpowered evidence, in some types of misclassification from measurement error, and in conflicts causing reverse biases. Finally, I discuss potential approaches to this problem. These include being cautious about newly discovered effect sizes, considering some rational down-adjustment, using analytical methods that correct for the anticipated inflation, ignoring the magnitude of the effect (if not necessary), conducting large studies in the discovery phase, using strict protocols for analyses, pursuing complete and transparent reporting of all results, placing emphasis on replication, and being fair with interpretation of results.
It remains unclear how much various factors contribute to the placebo response. To estimate the therapeutic impact of follow-up assessments on placebo response in antidepressant trials. Double-blind, placebo-controlled antidepressant trials that reported weekly changes in Hamilton Rating Scale for Depression (HRSD) scores over 6 weeks were selected. Included studies (n=41) were divided into those that conducted four, five or six follow-up assessments. Reductions in HRSD scores as a function of the different follow-up schedules were compared. An extra follow-up visit at week 3 was associated with a 0.86 further reduction in HRSD score; an extra visit at week 5 was associated with a 0.67 further reduction. These effects represented approximately 34-44% of the placebo response that occurred over these time frames. Two additional visits were associated with twice the reduction in HRSD score than one, suggesting that the therapeutic impact of assessment visits is cumulative and proportional. A comparable therapeutic effect was also found in participants receiving active medication. Follow-up assessments in antidepressant treatment trials incur a significant therapeutic effect for participants on placebo, and this represents about 40% of the placebo response.
Selective serotonin reuptake inhibitors are thought to have better discontinuation rates (i.e. less people dropping out) than tricyclic and heterocyclic antidepressant drugs. It is important to quantify the drop-out rates of different antidepressant drugs in order to have a better understanding of the relative tolerability of these drugs. To assess the comparative tolerability of selective serotonin reuptake inhibitors and tricyclic/heterocyclic antidepressant drugs. We searched the Cochrane Collaboration Depression, Anxiety and Neurosis Controlled Trials Registers (1997 to 1999), MEDLINE (1966 to 1999), EMBASE (1974 to 1999) We also searched specialist journals, the reference lists of relevant papers and previous systematic reviews, conference abstracts and government documents. Representatives of the pharmaceutical industry were contacted. Parallel group randomised controlled trials comparing selective serotonin reuptake inhibitors with tricyclic or heterocyclic antidepressants in people with depression. Two reviewers independently extracted data and a third reviewer checked any cases of disagreement. We included 136 trials. The selective serotonin reuptake inhibitors showed less participants dropping out compared to the tricyclic/heterocyclic group (odds ratio 1.21, 95% confidence interval 1.12 to 1.30). A statistically significant difference was found in total drop-outs between the selective serotonin reuptake inhibitors and the old tricyclics as well as the newer tricyclics. When the selective serotonin reuptake inhibitors were compared to the heterocyclic antidepressants, there was a non significant difference favouring the selective serotonin reuptake inhibitors. The poor tolerability profile of the old tricyclics was explained by differences in drop-outs for side-effects, but not for inefficacy. Whilst selective serotonin reuptake inhibitors do appear to show an advantage over tricyclic drugs in terms of total drop-outs, this advantage is relatively modest. This has implications for pharmaco-economic models, some of which may have overestimated the difference of drop-out rates between selective serotonin reuptake inhibitors and tricyclic antdepressants. These results are based on short-term randomised controlled trials, and may not generalise into clinical practice.
Randomized trials with adequate sample size offer an opportunity to assess the safety of new medications in a controlled setting; however, generalizable data on drug safety reporting are sparse. To scrutinize the completeness of safety reporting in randomized trials. Survey of safety reporting in 192 randomized drug trials 7 diverse topics with sample sizes of at least 100 patients and at least 50 patients in a study arm (N = 130074 patients). Trial reports were identified from comprehensive meta-analyses in 7 medical areas. Adequate reporting of specific adverse effects and frequency and reasons for withdrawals due to toxic effects; article space allocated to safety reporting and predictors of such reporting. Severity of clinical adverse effects and laboratory-determined toxicity was adequately defined in only 39% and 29% of trial reports, respectively. Only 46% of trials stated the frequency of specific reasons for discontinuation of study treatment due to toxicity. For these 3 parameters, there was significant heterogeneity in rates of adequate reporting across topics (P =.003, P<.001, and P =.02, respectively). Overall, the median space allocated to safety results was 0.3 page. A similar amount of space was devoted to contributor names and affiliations (P =.16). On average, the percentage of space devoted to safety in the results section was 9.3% larger in trials involving dose comparisons than in those that did not (P<.001) and 3.8% smaller in trials reporting statistically significant results for efficacy outcomes (P =.047). The quality and quantity of safety reporting vary across medical areas, study designs, and settings but they are largely inadequate. Current standards for safety reporting in randomized trials should be revised to address this inadequacy.