The big news this week was an antidepressant withdrawal
paper that showed most people can stop an antidepressant without experiencing severe withdrawal symptoms. This has
been known for over 20 years. It is only
big news because of the rhetorical approach to paper writing in the psychiatric
literature. Everybody knows what
confirmation bias is these days and that has a lot to do with the
literature. Quite remarkably, the camp
that claims a high prevalence of severe withdrawal also writes from the perspective
that most psychiatrists seem ignorant of withdrawal phenomena and need special
instruction. That despite decades of
practice modification, teaching residents how to taper and discontinue
medications, managing much more complex medication problems in acute care
settings, diagnosing life threatening medication related problems, and having
access to widely published guidelines on how to taper and cross taper and
titrate medications.
With that backdrop here are the highpoints and limitations
of the paper. The first remarkable
finding is the title: “Incidence and
Nature of Antidepressant Discontinuation Symptoms.” Discontinuation versus withdrawal
is a decade old point of contention. The antidepressant detractors use the term
withdrawal and at times have incorrectly suggested that antidepressants are
addictive drugs. The antidepressant defenders have preferred the term discontinuation
symptoms although some have just started using the term withdrawal as well.
There is no consistent standard for medication that can lead to addiction and
those that do not. Part of the reason
may be that physicians are trained to discontinue medications and with
all classes that typically involves a gradual taper or a taper while titrating a
medication that targets the same symptoms.
The overall study looked at 50 studies (N=17,858) in a
meta-analysis and systematic review. All
studies were randomized clinical trials or open label trials with a randomized
double blind discontinuation phase. All
the studies had to use a standardized measurement for discontinuation symptoms
or adverse events. The antidepressant trials covered several diagnoses in
addition to depression. From the paper:
“The following diagnoses were studied: major depressive
disorder (MDD) (k = 28), generalized anxiety disorder (k = 9), panic disorder (k
= 4), fibromyalgia (k = 2), premenstrual dysphoric disorder(k = 2),
posttraumatic stress disorder, generalized social anxiety disorder (k = 1),and compulsive-shopping
disorder (k = 1). Two studies included women with (post)menopause.”
The main methodological points that need further elaboration
are the withdrawal symptom scoring and the duration of treatment. The original DESS (Discontinuation Emergent
Signs and Symptoms) is from reference 2.
It was originally used on the sample in that paper of research 242
research subjects who had been on effective maintenance therapy for greater
than 4 months or less than 24 months with either paroxetine, sertraline, or
fluoxetine. The maintenance therapy was
interrupted for 5-8 days with placebo and their discontinuation symptoms were rated
using the DESS. The DESS items are
listed below along with the original question format.
As I read through the 43-item checklist – I noted that one of the commonest symptoms I have seen in antidepressant withdrawal – brain zaps was not present. Brain zaps are typically described as a sharp electrical sensation in the head or neck during SSRI/SNRI withdrawal. They can be worsened by head, neck, or eye movements. They are typically described as paresthesias and in the case of this checklist may be partially reflected in items 39 and 40.
The best starting point to gain an appreciation of the scope
and complexity of this study is eTable 2a Study Characteristics in the Supplemental
Content. I have copied page 1 (of 5)
below. 51 studies are listed including 44
for the individual symptoms meta-analysis, 16 for the continuous DESS meta-analysis,
and 1 for the qualitative synthesis. The duration of treatment range for all
groups was 13.5-17.8 weeks with significant outliers at each end. Most tapering protocols were abrupt (35/51)
with 14/51 1-week taper, 3/51 2-week taper, and 1/51 5–6-week tapers. Just based on the characteristics in this
table it would appear that most subjects were treated long enough to potentially
develop acute discontinuation symptoms and that in most cases they would have
been precipitated by the tapers used.
The medications studied in this table are also relevant since several are more likely to precipitate a withdrawal syndrome than others. In this case paroxetine (6), venlafaxine (1), and duloxetine (9) are less common in the table. In this case I am not counting the extended-release versions (where noted) because they were designed to reduce the risk of discontinuation/withdrawal. This can be noted in eTable 2B. Summary DESS Scores of Studies Included Meta-Analysis where they produce the highest DESS score with paroxetine producing the highest. Fewer significant withdrawal producing medications in the study is a strength because it reflects current psychiatric practice.
The authors’ analysis shows that antidepressant discontinuation results in expected discontinuations symptoms at one week. Dizziness and nausea were the commonest symptoms. The incidence of withdrawal symptoms was less when active drug was compared to placebo. Their key conclusion that all of the press has been based on:
“In conclusion, data from RCTs suggest that on average,
those who discontinue antidepressants experience 1 more discontinuation symptom
compared to placebo or continuation of antidepressants, which is below the
threshold for clinically important discontinuation syndrome.”
They end by discussing the notoriety issue when limited data
and analysis has suggested that severe, prolonged withdrawal/discontinuation syndromes
are common and need elaborate tapering schedules.
What are the overall lessons from this trial:
1: The research
design was well done to detect withdrawal symptoms measured with standardized methods. The DESS scale is available and can be used
as a reference to these research findings.
The main finding that there was an excess of one symptom in the
withdrawal group but that was not enough to diagnose a withdrawal syndrome was
remarkable and consistent with press reporting that the average withdrawal from
antidepressants is mild and should not be a deterrent to their use. The limitation of that conclusion is that the
DESS is not a quantitative measure as indicated by the authors. Even though there is precedent in the
literature for four symptoms or more being necessary for a significant
withdrawal syndrome – clinically it can be much less. As an example, dizziness (the commonest
symptom) alone can lead to distress and disability. Any person with vertigo can attest to that
fact. I am not in agreement that it takes 4 symptoms for significant withdrawal.
2: While the results
of this trial are being celebrated at the political level it will not have much
of an impact on clinical practice. Psychiatrists do not treat averages – they treat
individual patients with highly individual responses to medications. Despite these results there will be patients
who get severe withdrawal symptoms from antidepressants. There is more than enough information about
the relevant medications and pharmacokinetics to minimize or prevent withdrawal
symptoms from happening.
3: In discussing the
relevance of this widely publicized paper with patients - good clinical
practice still necessitates that the following topics are covered in detail
during informed consent discussions:
a) The indications,
risk, and benefits of antidepressant treatment for the specific patient and the
options of other treatments (psychotherapy, lifestyle changes).
b) The adverse
effects of antidepressants including withdrawal as well as serious life-threatening
adverse effects like serotonin syndrome.
c) Detailed
information on the suggested medication as an option at patient request –
such as the FDA approved package inserts.
d) Explicit call information
for discussing both a lack of efficacy and any potential adverse effects including
any change from baseline that was not explicitly discussed.
4: In terms of
settling any withdrawal or antidepressant controversy once and
for all that is a doubtful outcome. Research
can be designed to support the extreme positions on either end. It is just a
matter of time before the opposition comes out with a “new” analysis to support
their contention that antidepressants are over prescribed dangerous drugs with
very little therapeutic efficacy. There
used to be a research term called face validity or a subjective
assessment of research hypotheses that applies in this situation. It should be
apparent that antidepressants are useful medications and they can be safely
prescribed to many people. It should
also be apparent that clinical care is self-optimized to improve outcomes
rather than looking for possible positive and negative signals in an objective
way. These are all valuable lessons from
another paper in this matter.
5: There is also a thought
experiment that can be considered in the case of known risk for antidepressant
withdrawal. Suppose that a person has tried everything to treat their
depression or anxiety disorder and found that the only thing that worked was an
antidepressant. Let us further suppose
that missing even a single dose of that medication results in severe dizziness,
brain zaps, and nausea. After a detailed conversation about tapering and discontinuing
the medication or tapering the medication while starting a new one, or referral
for ECT or TMS neurostimulation – the patient elects to continue the
antidepressant with the risk of withdrawal.
I have had that conversation and similar conversations about side effects
with many patients. It is a common
conversation and one that most people do not understand until they are in that
position. This is a position that tens
of thousands of people are in – typically with medications that are far more
toxic than antidepressants.
6: At the political and public health level - treating complicated disorders with complicated medications and other therapies do not lend themselves to easy polling solutions. I notice that there are disclaimers on many commentaries these days that criticize one modality or another but in the end state "we do recommend that you seek treatment for this disorder." It reminded me of the decades long practice in psychiatry of dismissing research based on conflicts of interest that were typically research support or compensation for working on drug trials. I would give the same advice that I gave then to any psychiatrist who thinks that antidepressants are too dangerous or ineffective to use - don't use them.
This was an excellent study that needed to be done. It
reflects the reality of clinical psychiatry where practice has been modified
over that past 20-30 years to use medications with better tolerability, safety,
and efficacy. This is done with every
successive wave of newer medications in every medical specialty.
Finally, not enough people see research papers as arguments
for analysis. When you do that – they can
be analyzed from a scientific, rational, and moral dimension and just how rhetorical
those arguments are. In many cases in
psychiatry – the papers are purely rhetorical.
I encourage more editors to do that analysis in what I expect will be
rejoinders to this paper. One of the
best ways to do that is by comparing the paper to current clinical reality.
Yes – real life counts for something.
George Dawson, MD, DFAPA
1: Kalfas M, Tsapekos
D, Butler M, McCutcheon RA, Pillinger T, Strawbridge R, Bhat BB, Haddad PM,
Cowen PJ, Howes OD, Joyce DW, Nutt DJ, Baldwin DS, Pariante CM, Lewis G, Young
AH, Lewis G, Hayes JF, Jauhar S. Incidence and Nature of Antidepressant
Discontinuation Symptoms: A Systematic Review and Meta-Analysis. JAMA
Psychiatry. 2025 Jul 9:e251362. doi:
10.1001/jamapsychiatry.2025.1362. Epub ahead of print. PMID: 40632531;
PMCID: PMC12242823.
2: Rosenbaum JF, Fava
M, Hoog SL, Ascroft RC, Krebs WB. Selective serotonin reuptake inhibitor
discontinuation syndrome: a randomized clinical trial. Biol Psychiatry. 1998
Jul 15;44(2):77-87. doi: 10.1016/s0006-3223(98)00126-7. PMID: 9646889.
