Personal View

Tapering of SSRI treatment to mitigate withdrawal symptoms

Mark Abie Horowitz, David Taylor

Lancet Psychiatry 2019; All classes of drug that are prescribed to treat depression are associated with withdrawal syndromes. SSRI withdrawal
6: 538–46 syndrome occurs often and can be severe, and might compel patients to recommence their medication. Although the
Published Online withdrawal syndrome can be differentiated from recurrence of the underlying disorder, it might also be mistaken for
March 5, 2019
recurrence, leading to long-term unnecessary medication. Guidelines recommend short tapers, of between 2 weeks
http://dx.doi.org/10.1016/
S2215-0366(19)30032-X and 4 weeks, down to therapeutic minimum doses, or half-minimum doses, before complete cessation. Studies have
Prince of Wales Hospital, shown that these tapers show minimal benefits over abrupt discontinuation, and are often not tolerated by patients.
Sydney, NSW, Australia Tapers over a period of months and down to doses much lower than minimum therapeutic doses have shown greater
(M A Horowitz PhD); Health and success in reducing withdrawal symptoms. Other types of medication associated with withdrawal, such as
Environment Action Lab,
benzodiazepenes, are tapered to reduce their biological effect at receptors by fixed amounts to minimise withdrawal
London, UK (M A Horowitz);
and Institute of Pharmaceutical symptoms. These dose reductions are done with exponential tapering programmes that reach very small doses. This
Science, King’s College London, method could have relevance for tapering of SSRIs. We examined the PET imaging data of serotonin transporter
London, UK (Prof D Taylor PhD) occupancy by SSRIs and found that hyperbolically reducing doses of SSRIs reduces their effect on serotonin
Correspondence to: transporter inhibition in a linear manner. We therefore suggest that SSRIs should be tapered hyperbolically and
Dr Mark Abie Horowitz, Prince of
slowly to doses much lower than those of therapeutic minimums, in line with tapering regimens for other medications
Wales Hospital, Sydney,
NSW 2031, Australia associated with withdrawal symptoms. Withdrawal symptoms will then be minimised.
[email protected]
Introduction withdrawal effects and reported that nearly half of par­
Many medications are associated with withdrawal ticipants who had experienced withdrawal effects chose
syndromes, most commonly those that act on the the most extreme option in the scale offered to them to
cardiovascular system and CNS.1 All major classes of describe the severity of those effects.13 The discontinuation
antidepressants—monoamine oxidase inhibitors, tri­cyclic period (14 days after cessation) is also associated with a
antidepressants, SSRIs, and SNRIs—are associated with 60% increase in suicide attempts compared with previous
withdrawal symptoms on cessation.2,3 The term discon­ users of antidepressants (the increased risk therefore
tinuation syndrome was coined to refer to the withdrawal attributed to the process of withdrawal and not to being
syn­ drome related to antidepressants.4 SSRI discontin­ untreated).20
uation syndrome, as outlined in DSM-5,5 and captured in SSRI withdrawal syndrome can last substantially longer
the Discontinuation-Emergent Signs and Symptoms than the period of 1–2 weeks13 that has been previously
checklist,6 comprises a wide variety of somatic and suggested.4 In one study, withdrawal symptoms generally
psychological symptoms (figure 1). lasted for up to 6 weeks, with a quarter of patients
SSRI withdrawal symptoms can, in part, resemble the reporting symptoms that lasted more than 12 weeks.18
symptoms of anxiety or depression for which the Another study reported that for 86·7% of respondents the
medication was originally given.5 However, the with­ syndrome had lasted at least 2 months, for 58·6% it had
drawal syndrome can be distinguished from a relapse or lasted at least 1 year, and for 16·2% it had lasted for more
recurrence of the underlying disorder by its quickness of than 3 years.21 Case reports identify symptoms lasting for
onset (days rather than weeks),3,7,8 rapid response to a year or longer.22,23
reintroduction of the SSRI (generally within hours, The increasingly long-term use of SSRIs (with nearly
certainly within days),3,7,9 and the presence of somatic and half of the patients in the UK who take antidepressants
psychological symptoms quite distinct from the original [usually SSRIs] doing so for more than 2 years)19,24 has
illness (including dizziness, nausea, and shock-like arisen in part because patients are unwilling to stop due
sensations).7,10 The withdrawal syndrome can be mis­ to the aversive nature of the withdrawal syndrome,19,25
diagnosed as depressive recurrence, leading to prolonged and a scarcity of information on how to mitigate the
treatment for patients who might not require it,11–13 but it syndrome.19,25 Doctors feel that there is not enough
is not clear how often this occurs.14 guidance on how to proceed with discontinuation.19
SSRI withdrawal symptoms occur in many patients,
with reported incidence varying from 42% to 100% for Tapering SSRIs
paroxetine,6,15–18 and from 9% to 77% for fluoxetine,6,15,17,18 Guidelines recommend short tapers of SSRIs, rather than
with a mean rate of 53·6% for SSRIs across 14 studies abrupt discon­tinuation, to avoid withdrawal symptoms.
that examined antidepressant withdrawal.13 The incidence The National Institute for Health and Care Excellence,26
and severity appear to be influenced by half-life and the British Association for Psychopharmacology,12 the
receptor affinities, treatment duration and dose, method Monthly Index of Medical Specialities,27 and UpToDate28
of tapering, and individual patient characteristics, poten­ suggest tapering periods of between 2 weeks and 4 weeks,
tially including anticipation effects.3,9,19 A systematic with linear reductions of dose down to the minimum
review identified five studies that evaluated the severity of therapeutic dose, or half of the minimum therapeutic

538 www.thelancet.com/psychiatry Vol 6 June 2019

 Personal View

dose, before complete cessation. These guidelines suggest

that fluoxetine does not require tapering,28 or that, at Affective symptoms Sleep disruption
Irritability Insomnia
high doses, it could be reduced over 2 weeks.27 Drug Anxiety or agitation Nightmares
manufacturer advice was found to be similarly “vague Low mood or depression Excessive dreaming
Sensory symptoms Tearfulness
and non-specific” according to a systematic review.29 Dread
Paraesthesia
In randomised studies, tapering for up to 14 days Numbness Gastrointestinal symptoms
showed either no16 or minimal30 reduction in withdrawal Shock-like sensations Nausea
Rushing noises
symptom severity compared with abrupt discontinuation.31 Palinopsia (visual trails)
Vomiting
SSRI withdrawal Diarrhoea
It has generally been concluded from these studies that Anorexia
(discontinuation)
longer tapering regimens are required.3,32 Indeed, studies syndrome
using tapering periods of months in duration33–35 have General somatic symptoms
Flu-like symptoms Sexual symptoms
shown better outcomes (table 1). In one study, reduction Lethargy or fatigue Genital hypersensitivity
of paroxetine by 10 mg every 2 weeks lowered withdrawal Headache Premature ejaculation
Tremor Disequilibrium
incidence from 33·8% to 4·6%.33 When the patients Sweating Dizziness (most common)
who had withdrawal symptoms in this study were Anorexia Light-headedness Cognitive symptoms
Weakness
recommenced on medication and then tapered at 5 mg Tachycardia
Vertigo Confusion
Ataxia Decreased concentration
every 2–4 weeks, withdrawal symptoms were successfully Gait instability Amnesia
avoided.33 In another study, patients who tapered their
SSRI dose over up to 4 months had 5·1 Discontinuation-
Figure 1: Symptoms of SSRI withdrawal (discontinuation) syndrome
Emergent Signs and Symptoms events, compared with
11·7 events for patients who discontinued abruptly.34
In another study with paroxetine, patients who tapered intensity of withdrawal symptoms.3,12,32,42 Receptors that
their dose over an average duration of 38·6 weeks (range, are activated by a medication are often downregulated, or
2–197 weeks), titrated to the individual, had a 6·1% exhibit reduced sensitivity, to maintain homoeostasis.43
incidence of withdrawal syndrome, compared with Abrupt removal of medication disturbs the homoeostatic
78·2% for abrupt discontinuation (table 1).35 Tapering equilibrium, resulting in reduced stimulation, which is
strips for anti­depressants, which reduce the medication experienced as withdrawal symptoms that are often
to small fractions of the minimum therapeutic dose opposite in nature to the original effect of the drug.43
(eg, 0·5 mg for paroxetine and citalopram), have shown For example, the withdrawal syndrome from tricyclic
favourable outcomes; 71% of 895 patients, 97% of whom antidepressants, which have strong anti­ cholinergic
had experienced withdrawal previously, were able to actions, is typified by cholinergic effects.44 Adaptation to
discontinue their medication over a median of medication is more likely in the case of long-term and
56 days (IQR 28–84 days).14 Several case studies also high-dose use.45,46 Medications with shorter half-lives
support the improved efficacy of slower tapering.36–38 In produce withdrawal symptoms with greater incidence,
one instructive case, several months of tapering down to greater severity, and quicker onset than medications with
an average dose of 6·25 mg of sertraline per day was longer half-lives, probably because their withdrawal is
required to avoid withdrawal symptoms in one man, associated with more rapid decreases in the amount of
whose withdrawal-induced orthostatic hypotension available ligand.45,47,48 Withdrawal symptoms can usually
allowed objective measurement.36 be eliminated by reintroduction of the discontinued
Two studies from 2018 confirm that shorter tapering agent, returning the system to homoeostatic equilibrium.43
regimens, as advised by guidelines, are not effective. One The principal approach to mitigate withdrawal symp­
study found that tapering over 4 weeks was not feasible, toms is to reduce the rate at which this equilibrium is
with 60% of patients (51 of 85) tapering their medications disrupted, allowing time for adaptation of the system
over 4 months.39 Another study that used largely linear to lowered levels of ligand, thus limiting withdrawal
reductions, with final doses equal to minimum thera­ symptoms to tolerable severity.45 This process is achieved
peutic doses (or half that value) found that only 37% of either by substitution of a longer-acting medication before
patients (26 of 71) were able to discontinue their tapering, or slow tapering of a drug with a short half-life.45,48
medication.40 A large study involving 400 patients showed Notably, decreasing medication by constant amounts
a significantly lower risk of relapse if antidepressants (linear tapering) tends to cause increasingly severe side-
were tapered gradually (>14 days) rather than rapidly effects over time.45,48,49 This effect is probably a con­
(1–7 days).41 sequence of the hyperbolic dose-response relationship
between a drug and receptor, following the law of mass
Neurobiology of withdrawal and its action,50 as typified by the effect of diazepam on its target
management receptor, γ-aminobutyric-acid A (GABA-A, figure 2A).
The strategy of tapering SSRIs is based on the rationale Consequently, tapering recommendations for benzo­
that biological systems will have more time to adapt diazepines advise increasingly small decreases in dosage
to reductions in available ligand, thus reducing the as it approaches zero,45,48,49 or “stop slow as you go low”.1

www.thelancet.com/psychiatry Vol 6 June 2019 539

 Personal View

Number of patients Medication Tapering period Lowest dose before Outcome (% with Odds ratio AD Comment
zero discontinuation versus taper
syndrome or DESS score)
Groot and van Os 895; antidepressant Paroxetine, Median 56 days 0·5 mg (paroxetine); 71% able to cease N/A Included patients who had
(2018)14 use median 2–5 years venlafaxine (IQR 28–84 days) 1·0 mg (venlafaxine) had severe withdrawal
syndrome previously
Tint and colleagues 28 SSRI, 3 days; 14 days Unknown 46% (3 days); 46% N/A No difference between 3 day
(2008)16 venlafaxine (14 days) and 14 day taper
Baldwin and 249 Paroxetine 7 days; 14 days 10 mg (paroxetine); DESS 5·4 (SD 8·3) for N/A No difference between 7 day
colleagues (2006)30 (N=115) or 5 mg (escitalopram) paroxetine; DESS 3·2 and 14 day taper (but both
escitalopram (SD 4·8) for escitalopram; slightly better than AD when
(N=134) no difference between 7 compared with other studies)
and 14 days
Himei and Okamura 385 Paroxetine AD (N=80); taper, 10 mg 33·8% (AD); 4·6% (taper) 7·4 36 patients with withdrawal
(2006)33 reducing by 10 mg syndrome were recommenced
every 2 weeks on paroxetine and tapered at
(N=305) 5 mg every 2–4 weeks with no
re-emergence of symptoms
van Geffen (2005)34 74 Fluvoxamine, AD (N=14); taper, Unknown 86% (AD); 52% (taper) 1·65 Significant reduction in
fluoxetine, 2 weeks–4 months withdrawal symptoms with
paroxetine, (N=52) tapering compared with AD
citalopram
Murata and 56 Paroxetine AD (N=23); taper, 10 mg 78·2% (AD); 6·1% (taper) 12 Odds ratio of 55·8 by
colleagues (2010)35 average univariate logistic regression
38·6 weeks, range of tapering compared with AD
2–197 weeks
(N=33)
DESS=Discontinuation-Emergent Signs and Symptoms. AD=abrupt discontinuation. N/A=not applicable.

Table 1: Studies that measured withdrawal symptoms in patients tapered off antidepressants

Withdrawal guidelines for benzodiazepines recom­mend Pharmacological characteristics of SSRI

dose reductions that are proportional to the present dose withdrawal
(most commonly 10% reductions), yielding expo­nentially Withdrawal or discontinuation syndrome from SSRIs
decreasing regimens, as opposed to linear reductions.45,49,52 has the same determinants as described previously.
For example, tapering from 20 mg of diazepam at a rate of Withdrawal symptoms are more common when SSRIs
10% a week would entail a 2 mg reduction in the first week. are given in high doses,53,54 or for long periods.53
The second week’s cumulative reduction would be 3·8 mg Medications with shorter half-lives, such as paroxetine,
(a further 1·8 mg reduction), the third week’s cumulative produce withdrawal symptoms with greater incidence,6,15–18
reduction would be 5·42 mg (a further 1·62 mg), and so on quicker onset,6,15–18 and greater severity6,15–18 than medi­
(figure 2B). These expo­ nentially decreasing regimens cations with longer half-lives, such as fluoxetine.6,15,17,18
produce approximately linear reductions of effect at the Paroxetine produces withdrawal symptoms within
target receptor. Reductions are continued to doses well 2 days,55 whereas symptoms of fluoxetine withdrawal can
under the minimum therapeutic dose (that might appear be delayed by 2–6 weeks.9,56
miniscule) before complete cessation. This process is done As with withdrawal from other medications, the
to avoid a step down in action at the target receptor that is appearance of these withdrawal effects correlates with
substantially greater than the size of steps previously percentage reductions in plasma concentration.55 Higher
tolerated. For example, the final dose of diazepam SSRI plasma levels before cessation57 and just after
recommended by tapering guidelines is 1 mg45 (equivalent cessation58 predict increased withdrawal symptoms.
to 4% GABA-A receptor occupancy).52 Reintroduction of the dis­continued SSRI generally resolves
As withdrawal symptoms are thought to abate because of symptoms within 24 h.3 Approaches have been trialled to
homoeostatic adaptations to reduced medication levels, a diminish withdrawal symptoms by tapering of SSRIs,9,12 or
pause is recommended between dose reductions.45,48,49 substitution for the longest acting SSRI, fluoxetine,3,12
As the exact timing of these adaptations is not fully under­ according to the approaches used for with­drawal from
stood, most guidelines for withdrawal have been developed other agents. Individual factors, including genetics,35 are
on the basis of clinical experience; a consensus suggests thought to play a role in determining withdrawal effects.
waiting 1–4 weeks between dose reductions, to allow
withdrawal symptoms to resolve.45,48 Most guide­ lines Neurobiology of SSRI withdrawal
recommend individualisation of this process, given the SSRIs are thought to produce their effect through an
variation in adaptation to changes in drug levels, and con­ initiating step of inhibition of the serotonin transporter,
sequent severity and duration of with­drawal symptoms.45,48 leading to an increase in synaptic levels of serotonin,

540 www.thelancet.com/psychiatry Vol 6 June 2019

 Personal View

thereby transducing increased responses at serotonergic

A
receptors.59,60 Serotonergic neurons also modulate other 100
neurotransmitter systems, including noradrenaline,
dopamine, and GABA.47 Effects on neurogenesis, 80

GABA occupancy (%)

inflammation, and the hypothalamic–pituitary–adrenal
axis, downstream of serotonergic actions, are also 60
hypothesised in the antidepressant effects of SSRIs.59,61,62
40
Although the details remain to be elucidated, SSRI
withdrawal has been attributed to a relative deficiency of 20
serotonin in the context of widespread adaptation of
serotonergic receptors.9,47,63 SSRI treatment has been 0
shown to down-regulate the density of serotonergic recep­ 0 1 2 3 4 5
Diazepam dose (mg/kg)
tors in rats.64 It has also been shown in humans that even
short-term SSRI administration reduces the sensitivity B
of cortical 5-hydroxytryptamine receptor 2A65 and 25
5-hydroxytryptamine receptor 4.66 The reversal of effects on
Cumulative dose reduction (mg)

neurotransmitters that are indirectly affected by SSRIs, 20

including noradrenaline, glutamate, and GABA, among

15
other targets, could also play a role in SSRI withdrawal.9,47,63
The role of serotonin in coordinating sensory and auto­ 10
nomic function with motor activity has been implicated
in SSRI withdrawal syndrome.42 Reduced stimulation of 5
5-hydroxytryptamine receptor 1A in the raphe nucleus,
known to be involved in motion sickness,47 is thought to 0
0 10 20 30 40 50
be related to the dizziness, vertigo, nausea, and lethargy Time (weeks)
of the withdrawal syndrome.47 Dysregulation of somato­
sensory functions could result in paraesthesia, whereas Figure 2: Recommended tapering regimen for diazepam based on its
dose-response relationship
movement disorders (eg, dystonia) could be due to
(A) Relationship between dose of diazepam and action at GABA-A receptors in
altered dopaminergic function.47 non-human primates. Adapted from Brouillet and colleagues.51 (B) Reductions in
Aspects of SSRI withdrawal syndrome might also be diazepam dose recommended by tapering guidelines for 20 mg diazepam
attributed to neuronal changes in tissues outside the (10% dose reduction per week).45,49 GABA=γ-aminobutyric-acid.
brain, given the presence of serotonergic receptors in
sites such as the vasculature and gut.67 of 3% from 20 mg to 15 mg, 6% from 15 mg to 10 mg,
13% from 10 mg to 5 mg, and 58% from 5 mg to 0 mg.
Pharmacological principles of tapering SSRIs Even reductions from 2·5 mg (a quarter of the smallest
As with other withdrawal syndromes, a rational tapering available tablet) to 0·0 mg will produce an absolute
regimen for SSRIs will entail step-wise reductions in their reduction in serotonin transporter inhibition of 42·9%,
action at serotonin transporters, their principal receptor and reduction from 1·25 mg (an eighth of a tablet)
targets.59 PET studies in which a radioligand was bound to to 0·00 mg will produce a 28% reduction (larger than
serotonin transporters have shown that the dose-response the change from 40 mg to 5 mg, which produces a
curve between SSRIs and serotonin transporters conforms 27·3% reduction). These large reductions in inhibition
to the typical hyperbolic relationship, which arises as a could account for the paucity of success of previous
consequence of the law of mass action (figure 3). The line tapering regimens,39,40 and particularly for the difficulties
of best fit for the dose-response curve, which corresponds with withdrawal symptoms that patients have towards
to a Michaelis-Menten equation,68 can be used to derive the end of their taper, at low doses.14,36
values for the percentage inhibition of SERT at different To produce a linear reduction of pharmacological effect,
doses of citalopram (figure 3, table 2).60 Notably, serotonin a hyperbolically decreasing pattern of dose reduc­tion is
transporter inhibition drops off sharply for doses lower required (figure 4B). Rather than decreasing the dose by
than the minimum therapeutic dose for SSRIs. fixed amounts, the dose should be decreased according
It is therefore likely that tapering regimens with linear to fixed intervals of biological effect, for example,
dose reductions will cause increasingly severe withdrawal 10% reductions of serotonin transporter occupancy
reactions, as the reductions in serotonin transporter (20% reductions are shown in figure 4B). A tapering
inhibition become increasingly large (figure 4A). For regimen that would produce approximately 10% reductions
example, reducing the dose of citalopram in 5 mg in serotonin receptor occupancy with each citalopram
increments from 20 mg will produce hyperbolically dose reduction would be: 20 mg, 9·1 mg, 5·4 mg, 3·4 mg,
enlarging decreases in serotonin transporter inhibition: 2·3 mg, 1·5 mg, 0·8 mg, 0·4 mg, and 0·00 mg (table 2).
an absolute decrease in serotonin transporter inhibition Further SSRI examples are shown in the appendix. These See Online for appendix

www.thelancet.com/psychiatry Vol 6 June 2019 541

 Personal View

successfully in trials involving tapering strips,14 and in

A
100
case studies involving difficult withdrawal syndromes.36

Limitations
Striatal SERT occupancy (%)

80
There are potential limitations to interpreting the dose-
60 response curve of the PET study presented.60 The number
of participants in each group is relatively small, perhaps
40
limiting the ability to capture individual variation.
20
However, the shape of the dose-response curve
(ie, hyperbolic) should be the same for each individual,
0 suggesting hyperbolic dose reduction regimens should
0 10 20 30 40 50 60 be universally applicable.
Dose (mg/day)
SSRIs might also exert neurotrophic, anti-inflammatory,
B and neuroendocrine effects;61,62 however, these are thought
100 to be downstream of effects on serotonin transporters
and consequent to changes to the serotonergic system,59,61,62
Striatal SERT occupancy (%)

80 indicating that serotonin transporter occupancy is likely

to be a key indicator of biological response to SSRIs.
60
It is difficult to determine whether serotonin trans­
40
porter inhibition will linearly correspond to withdrawal
effects. Serotonin transporter binding is related to the
20 anti­depressant effects of SSRIs; the ratio of serotonin
transporter binding between terminal regions and the
0 median raphe nucleus has been shown to predict treat­
0 20 40 60 80 100 120 140 160 180 200 220
Plasma citalopram (µg/L)
ment response to SSRIs.69 It is theoretically possible that
a minimum threshold of serotonin transporter inhibition
Figure 3: Hyperbolic relationship between SERT and dose or plasma is required before a clinical effect is seen, with levels
concentration of citalopram lower than this having minimal effects;60 this could also
(A) Relationship between dose of citalopram and SERT occupancy (%).
(B) Relationship between plasma level of citalopram and SERT occupancy (%).
correspond to withdrawal effects. However, with­drawal
SERT=serotonin transporter. Reproduced from Meyer and colleagues,60 by effects from other medications do not observe threshold
permission of the American Journal of Psychiatry. effects,45,48 and withdrawal effects have been observed at
many doses during tapering of SSRIs,9,35 suggesting that
withdrawal is likely to be a continuous entity. Furthermore,
Citalopram dose (mg) SERT occupancy (%)
a hyperbolic relation­ship exists between dose of SSRI and
60·0 87·8% reduction in depressed mood, as shown in a mega-
40·0 85·9% analysis;70 a hyperbolic relationship has also been shown
20·0 80·5% between dose of SSRI and risk of withdrawal symptoms.55
19·0 80·0% These findings indicate that the hyperbolic relationship
9·1 70·0% between dose and serotonin transporter inhibition
5·4 60·0% could also extend to withdrawal effects, suggesting that
3·4 50·0% serotonin transporter inhibition could be approximately
2·3 40·0% linearly related to withdrawal effects.
1·5 30·0% Another potential limitation to interpreting the dose-
0·8 20·0% response curve from Meyer and colleagues60 is that
0·37 10·0% serotonin transporter occupancy was measured in the
SERT occupancy was calculated using the Michaelis-Menten equation of best fit striatum, which might not have direct relevance to
derived by Meyer and colleagues.60 Common clinical doses and doses antidepressant actions. However, this and a subsequent
corresponding to 10% decrements of SERT inhibition are displayed. These doses PET study71 showed that SSRIs cause similar serotonin
could be produced by a combination of tablets and liquid formulations.
Approximations might be necessary. SERT=serotonin transporter.
transporter inhibition in brain regions relevant to SSRI
action (eg, subgenual cingulate, amygdala, and raphe
Table 2: Derivation of SERT occupancy from citalopram dose using the
nuclei), with a similar hyperbolic relationship between
Michaelis-Menten equation of best fit
dose of SSRI and serotonin transporter occupancy at all
regions examined.71 Therefore, it seems reasonable to
regimens allow pharmacologically informed application conclude that SSRIs, like most pharma­cological agents,
of the withdrawing principles outlined above (“stop slow have a hyperbolic relationship between dose and biological
as you go low”).1,49 The tapering regimen described above effects, and that this could be relevant in generating
uses doses that are close to those that have been used rational discontinuation regimens.

542 www.thelancet.com/psychiatry Vol 6 June 2019

 Personal View

Practical application of hyperbolic dose A

reduction 100
There are likely to be individual differences in experiences
of SSRI withdrawal effects.3 We suggest that a personalised

Striatal SERT occupancy (%)

80
rate for withdrawal could be established by an initial trial
reduction of SSRI dose, equivalent to a reduction of 60
10% serotonin transporter occupancy (or 5% if being
40
cautious), with subsequent monitoring of the severity and
duration of withdrawal symptoms. An initial 10% reduc­ 20
tion in serotonin transporter occupancy is suggested
because this would result in approximately halving the 0
dose from the therapeutic minimum dose (eg, from 20 mg
B
to 10 mg of citalopram), which is tolerated well by most 100
patients. If the patient’s Discontinuation-Emergent Signs
and Symptom score were to have returned to baseline
Striatal SERT occupancy (%)

80
1 month after the initial reduction, then a rate equivalent to
10% reduction of serotonin transporter occupancy per 60
month could be prescribed. This process should be subject
40
to ongoing monitoring, with the rate titrated to patient
tolerance. 20
The SSRI should be tapered such that the final
reduction to zero is equal to (or less than) the size of 0
reduction previously tolerated by the patient. This should 0 10 20 30 40 50 60 70
Dose (mg/day)
be when the dose is equivalent to approximately
10% serotonin transporter occupancy. Notably, this will Figure 4: Effect of linear and hyperbolic citalopram dose reductions on SERT
be a very small dose—eg, 0·4 mg for citalopram. occupancy
However, studies have reported tapering regimens that (A) Linear dose decrements of citalopram (eg, intervals of 5 mg) produce
exponentially increasing changes in SERT occupancy. (B) Hyperbolic dose
have been successful only when they taper to similar decreases of citalopram produce linear changes in SERT occupancy (eg, intervals
doses of SSRIs.14,36 The use of liquid formulations of of 20% SERT occupancy). SERT=serotonin transporter. Reproduced from Meyer
SSRIs might be necessary to achieve these small doses. and colleagues,60 by permission of the American Journal of Psychiatry.
It is difficult to establish the optimal time interval
between dose reductions. In the absence of studies Other determinants of withdrawal symptoms
evaluating the rate at which neuroadaptation can occur, from SSRIs
several aspects can provide guidance. For all SSRIs Other drug and patient characteristics are likely to affect
except fluoxetine, pharmacokinetic properties predict the severity of withdrawal syndromes from SSRIs.
that they will achieve steady state between 5 days and Paroxetine and fluoxetine are both metabolised by
14 days after dose reduction (table 3).72 As outlined above, cytochrome P450 2D6 and inhibit their own metabolism,
discon­tinuation symptoms have been detected in patients resulting in non-linear kinetics.76 This predicts dis­
for varying periods of time, from a number of days,9 to proportionate declines in plasma concentrations during
weeks and months,18,21–23,73,74 and, in some cases, for more drug withdrawal. Although this effect might not be
than a year.21,22,73,74 These records have generally been clinically significant for fluoxetine because of its long half-
derived from patients who abruptly cease their life, it is likely to be significant for paroxetine.47 Paroxetine
medication; it is possible that with more cautious might produce a more severe withdrawal syndrome than
reductions, the discontinuation symptoms will last for other SSRIs because it has pronounced muscarinic
shorter periods. The clinical effects of SSRIs can be antagonist effects and moderate nor­ epinephrine trans­
delayed by weeks following their commencement,59,70 porter inhibiting effects.47,63 It is also likely that patient
whereas side-effects arise in days.75 It is unclear whether factors, such as presence of different cytochrome
withdrawal symptoms are likely to follow the temporal enzymes, serotonin transporter sensitivity to inhibition,
pattern of antidepressant effects, or side-effects. It might and psychological factors, could contribute to the risk of
be prudent to allow 4 weeks after a reduction in SSRI to withdrawal symptoms. Further understanding of these
observe for delayed withdrawal effects. This would also factors, and testing of plasma levels of SSRIs, might be
allow for observation of recurrence of underlying instructive in designing personalised tapering regimens.
symptoms as a result of the decrease in SSRI dose.
However, the best guide might be the interval required Practical consequences of hyperbolic dose
for the patient’s Discontinuation-Emergent Signs and reduction
Symptoms score to return to baseline after a test The model above resolves an uncertainty often raised by
reduction. patients and treating physicians, which is whether to use a

www.thelancet.com/psychiatry Vol 6 June 2019 543

 Personal View

symptoms.3,77 Fluoxetine has been routinely identified to

Daily dose Half-life Time to reach Linear kinetics Cytochrome
(mg) steady state P450 inhibition cause less severe withdrawal effects than other SSRIs,
which has been attributed to its longer half-life.3,29,77
Fluoxetine 20–80 1–4 days 5–11 weeks No 2D6
Fluoxetine takes 35–75 days to reach steady state,47 which
Norfluoxetine ·· 7–15 days ·· ·· 2D6, 3A4
is likely to be responsible for observations that its
Fluvoxamine 50–300 15 h 10 days No 1A2, 2C19
withdrawal symptoms can arise weeks after cessation.9,56
Paroxetine 20–50 20 h 7–14 days No 2D6
Therefore, it would be prudent to wait 3 months
Sertraline 50–150 26 h 5–7 days Yes Minimal
(35–75 days plus 4 weeks) to observe for late-arising
Citalopram 10–60 36 h 6–10 days Yes Not relevant
withdrawal symptoms. Given this property of fluoxetine,
Escitalopram 5–30 27–33 h 7–10 days Yes 2C19, 2D6, 3A4
which is similar to an in-built tapering system, it could
Adapted from Hiemke and colleagues. 72
be reasonable to reduce doses by the equivalent of
Table 3: Pharmacokinetic characteristics of SSRIs and their clinically active metabolites approximately 30% serotonin transporter occupancy in
each iteration, titrated to patient tolerance.
Nevertheless, we should be wary of the idea that
40·00 fluoxetine is self-tapering, and can therefore be abruptly
or rapidly ceased, as guidelines suggest.29,78 Although its
20·00
pharmacokinetic profile predicts a gradual decline in
10·00
plasma level, a short reduction (eg, 2 weeks)27 could still
represent a rapid withdrawal schedule that exceeds a
5·00 10% decrease in biological effect per month.
Dose (mg)

2·50 Future directions for research

We have proposed a pharmacologically informed
1·25
method for tapering SSRI treatment, the validity of
0·60 which should be evaluated by randomised controlled
90th trials. Withdrawal nomograms aggregating variation
0·30 percentile in responses to withdrawal could guide taper rates
10th 30th 50th 70th
percentile percentile percentile percentile (figure 5). Risk determinants, such as plasma SSRI
0
0 4 8 12 16 20 24 28 32 level, cytochrome enzyme status, PET measurement of
Time (weeks) serotonin transporter occupancy, and other genetic,
metabolic, and psychological factors, could be in­
Figure 5: Hypothetical nomogram for determining withdrawal rates for citalopram
An example tapering regimen from 20 mg citalopram is indicated (semi-log scale for dose). The patient’s results corporated into the nomogram as their effects are
initially follows the trajectory for the 50th percentile, with dose reductions equivalent to 10% serotonin transporter clarified. Pharmacological and non-pharmacological
occupancy every 4 weeks (20·0 mg, 9·1 mg, and 5·4 mg). She then experiences unpleasant side-effects (eg, means to improve the tolerability of SSRI withdrawal
Discontinuation-Emergent Signs and Symptoms score of >3). Her tapering shifts to the slower trajectory for the also warrant research. Psychological interventions,
70th percentile and she experiences no periods of intolerable withdrawal symptoms as she completes the taper.
such as preventive cognitive therapy, and other CBT
approaches, have been found to reduce the risk of
micro-taper or mini-taper strategy. Micro-tapering involves relapse in patients with recurrent depression and those
miniscule decrements in SSRI medication every day or tapering their antidepressants.39,40
week. Mini-tapering involves larger, step-wise decrements, We suggest that, in the absence of more robust
with longer intervals between decrements (generally evidence to guide tapering (especially where guidelines
intervals of weeks). Mini-tapering appears more sensible advise to taper gradually without specific instructions),
than micro-tapering. Withdrawal symp­toms are reported the tapering regimen described here should be
to last for several weeks (or longer) after medication considered for adop­tion into clinical practice. There are
discontinuation in a large proportion of patients.9,13 few disadvantages of recommending slower tapers.29 It
Consequently, micro-tapering presents the possibility of should at least be recognised that tapering periods of
cumulative withdrawal effects being super­imposed on one 2–4 weeks are likely to be inadequate for reducing
another. This process would make it difficult to establish withdrawal symptoms for many patients, with longer
which reduction (or set of reductions) was responsible for periods of tapering, and regimens that include lower
the symptoms experienced. It there­fore seems prudent to doses of medication, more likely to be effective. Further
decrease the dose of medication, then allow a substantial empirical study of tapering regimens, including the one
period of time to elapse while withdrawal effects resolve, proposed here, is urgently required, with a consequent
before making the next decrement. update of formal guidelines.
Contributors
Fluoxetine MAH conceived the manuscript idea, wrote the manuscript, and drew
Substitution of short-acting SSRIs with fluoxetine has the figures. DT helped to develop the idea, and revised and edited the
manuscript.
been suggested as a way to avoid intolerable withdrawal

544 www.thelancet.com/psychiatry Vol 6 June 2019

 Personal View

Declaration of interests 22 Fava GA, Bernardi M, Tomba E, Rafanelli C. Effects of gradual

MAH declares no competing interests. DT reports personal fees from discontinuation of selective serotonin reuptake inhibitors in panic
Lundbeck, and grants and personal fees from Janssen, outside the disorder with agoraphobia. Int J Neuropsychopharmacol 2007;
submitted work. 10: 835–38.
23 Bhanji NH, Chouinard G, Kolivakis T, Margolese HC.
References Persistent tardive rebound panic disorder, rebound anxiety and
1 Bain KT, Holmes HM, Beers MH, Maio V, Handler SM, Pauker SG. insomnia following paroxetine withdrawal: a review of rebound-
Discontinuing medications: a novel approach for revising the withdrawal phenomena. Can J Clin Pharmacol 2006; 13: e69–74.
prescribing stage of the medication-use process. J Am Geriatr Soc 24 Johnson CF, Macdonald HJ, Atkinson P, Buchanan AI, Downes N,
2008; 56: 1946–52. Dougall N. Reviewing long-term antidepressants can reduce drug
2 Howland RH. Potential adverse effects of discontinuing psychotropic burden: a prospective observational cohort study. Br J Gen Pract 2012;
drugs: part 2: antidepressant drugs. J Psychosoc Nurs Ment Health Serv 62: 773–79.
2010; 48: 9–12. 25 Eveleigh R, Muskens E, Lucassen P, et al. Withdrawal of unnecessary
3 Haddad PM, Anderson IM. Recognising and managing antidepressant medication: a randomised controlled trial in primary
antidepressant discontinuation symptoms. Adv Psychiatr Treat 2007; care. BJGP Open 2017; 34: bjgpopen17X101265.
13: 447–57. 26 NICE. Depression in adults: recognition and management (CG90).
4 Schatzberg AF, Haddad P, Kaplan EM, et al. Serotonin reuptake London: National Institute for Health and Care Excellence, 2009.
inhibitor discontinuation syndrome: a hypothetical definition. 27 MIMS. Antidepressants, a guide to switching and withdrawing.
Discontinuation Consensus panel. J Clin Psychiatry 1997; http://www.mims.co.uk/Tables/882430/Switching-Antidepressants/
58 (suppl 7): 5–10. (accessed Nov 1, 2018).
5 American Psychiatric Association. The diagnostic and statistical 28 Hirsch M, Birnbaum RJ. Antidepressant medication in adults:
manual of mental disorders, 5th edn. Arlington, VA: American switching and discontinuing medication. https://www.uptodate.
Psychiatric Publishing, 2013. com/contents/discontinuing-antidepressant-medications-in-adults
6 Rosenbaum JF, Fava M, Hoog SL, Ascroft RC, Krebs WB. (accessed Feb 11, 2019).
Selective serotonin reuptake inhibitor discontinuation syndrome: 29 Ogle NR, Akkerman SR. Guidance for the discontinuation or
a randomized clinical trial. Biol Psychiatry 1998; 44: 77–87. switching of antidepressant therapies in adults. J Pharm Pract 2013;
7 Warner CH, Bobo W, Warner C, Reid S, Rachal J. 26: 389–96.
Antidepressant discontinuation syndrome. Am Fam Physician 2006; 30 Baldwin DS, Cooper JA, Huusom AKT, Hindmarch I. A double-blind,
74: 449–56. randomized, parallel-group, flexible-dose study to evaluate the
8 Shelton RC. Steps following attainment of remission: tolerability, efficacy and effects of treatment discontinuation with
discontinuation of antidepressant therapy. escitalopram and paroxetine in patients with major depressive
Prim Care Companion J Clin Psychiatry 2001; 3: 168–74. disorder. Int Clin Psychopharmacol 2006; 21: 159–69.
9 Fava GA, Gatti A, Belaise C, Guidi J, Offidani E. 31 Montgomery SA, Kennedy SH, Burrows GD, Lejoyeux M,
Withdrawal symptoms after selective serotonin reuptake inhibitor Hindmarch I. Absence of discontinuation symptoms with
discontinuation: a systematic review. Psychother Psychosom 2015; agomelatine and occurrence of discontinuation symptoms with
84: 72–81. paroxetine: a randomized, double-blind, placebo-controlled
10 Black K, Shea C, Dursun S, Kutcher S. Selective serotonin reuptake discontinuation study. Int Clin Psychopharmacol 2004; 19: 271–80.
inhibitor discontinuation syndrome: proposed diagnostic criteria. 32 Phelps J. Tapering antidepressants: is 3 months slow enough?
J Psychiatry Neurosci 2000; 25: 255–61. Med Hypotheses 2011; 77: 1006–08.
11 Young A, Haddad P. Discontinuation symptoms and psychotropic 33 Himei A, Okamura T. Discontinuation syndrome associated with
drugs. Lancet 2000; 355: 1184. paroxetine in depressed patients: a retrospective analysis of factors
12 Cleare A, Pariante CM, Young AH, et al. Evidence-based guidelines involved in the occurrence of the syndrome. CNS Drugs 2006;
for treating depressive disorders with antidepressants: a revision of 20: 665–72.
the 2008 British Association for Psychopharmacology guidelines. 34 van Geffen EC, Hugtenburg JG, Heerdink ER, van Hulten RP,
J Psychopharmacol 2015; 29: 459–525. Egberts AC. Discontinuation symptoms in users of selective
13 Davies J, Read J. A systematic review into the incidence, severity serotonin reuptake inhibitors in clinical practice: tapering versus
and duration of antidepressant withdrawal effects: are guidelines abrupt discontinuation. Eur J Clin Pharmacol 2005; 61: 303–07.
evidence-based? Addict Behav 2018; published online Sept 4. 35 Murata Y, Kobayashi D, Imuta N, et al. Effects of the serotonin 1A,
DOI:10.1016/j.addbeh.2018.08.027 (preprint). 2A, 2C, 3A, and 3B and serotonin transporter gene polymorphisms
14 Groot PC, van Os J. Antidepressant tapering strips to help people on the occurrence of paroxetine discontinuation syndrome.
come off medication more safely. Psychosis 2018; 10: 142–45. J Clin Psychopharmacol 2010; 30: 11–17.
15 Bogetto F, Bellino S, Revello RB, Patria L. Discontinuation syndrome 36 Amsden GW, Georgian F. Orthostatic hypotension induced by
in dysthymic patients treated with selective serotonin reuptake sertraline withdrawal. Pharmacotherapy 1996; 16: 684–86.
inhibitors: a clinical investigation. CNS Drugs 2002; 16: 273–83. 37 Dominguez RA, Goodnick PJ. Adverse events after the abrupt
16 Tint A, Haddad PM, Anderson IM. The effect of rate of discontinuation of paroxetine. Pharmacotherapy 1995; 15: 778–80.
antidepressant tapering on the incidence of discontinuation 38 Benazzi F. Venlafaxine withdrawal symptoms. Can J Psychiatry 1996;
symptoms: a randomised study. J Psychopharmacol 2008; 22: 330–32. 41: 487.
17 Hindmarch I, Kimber S, Cockle SM. Abrupt and brief 39 Bockting CLH, Klein NS, Elgersma HJ, et al. Effectiveness of
discontinuation of antidepressant treatment: effects on cognitive preventive cognitive therapy while tapering antidepressants versus
function and psychomotor performance. Int Clin Psychopharmacol maintenance antidepressant treatment versus their combination in
2000; 15: 305–18. prevention of depressive relapse or recurrence (DRD study):
18 Royal College of Psychiatrists. Coming off antidepressants. 2012. a three-group, multicentre, randomised controlled trial.
www.rcpsych.ac.uk/healthadvice/treatmentswellbeing/anti Lancet Psychiatry 2018; 5: 401–10.
depressants/comingoffantidepressants.aspx (accessed March 1, 2018). 40 Scholten WD, Batelaan NM, van Oppen P, et al. The efficacy of a
19 Bosman RC, Huijbregts KM, Verhaak PF, et al. Long-term group CBT relapse prevention program for remitted anxiety disorder
antidepressant use: a qualitative study on perspectives of patients patients who discontinue antidepressant medication: a randomized
and GPs in primary care. Br J Gen Pract 2016; 66: e708–19. controlled trial. Psychother Psychosom 2018; 87: 240–42.
20 Valuck RJ, Orton HD, Libby AM. Antidepressant discontinuation 41 Baldessarini RJ, Tondo L, Ghiani C, Lepri B. Illness risk following
and risk of suicide attempt. J Clin Psychiatry 2009; 70: 1069–77. rapid versus gradual discontinuation of antidepressants.
21 Davies J, Regina P, Montagu L. Antidepressant withdrawal: a survey Am J Psychiatry 2010; 167: 934–41.
of patients’ experience by the All-Party Parliamentary Group for 42 Maixner SM, Greden JF. Extended antidepressant maintenance and
Prescribed Drug Dependence. 2018. http://prescribeddrug.org/wp- discontinuation syndromes. Depress Anxiety 1998; 8 (suppl 1): 43–53.
content/uploads/2018/10/APPG-PDD-Survey-of-antidepressant- 43 Turton S, Lingford-Hughes A. Neurobiology and principles of
withdrawal-experiences.pdf (accessed Nov 1, 2018). addiction and tolerance. Medicine (Baltimore) 2016; 44: 693–96.

www.thelancet.com/psychiatry Vol 6 June 2019 545

 Personal View

44 Dilsaver SC. Withdrawal phenomena associated with antidepressant 63 Renoir T. Selective serotonin reuptake inhibitor antidepressant
and antipsychotic agents. Drug Saf 1994; 10: 103–14. treatment discontinuation syndrome: a review of the clinical
45 Ashton H. The diagnosis and management of benzodiazepine evidence and the possible mechanisms involved. Front Pharmacol
dependence. Curr Opin Psychiatry 2005; 18: 249–55. 2013; 4: 45.
46 Hodding GC, Jann M, Ackerman IP. Drug withdrawal syndromes: 64 Wamsley JK, Byerley WF, McCabe RT, McConnell EJ, Dawson TM,
a literature review. West J Med 1980; 133: 383–91. Grosser BI. Receptor alterations associated with serotonergic agents:
47 Olver JS, Burrows GD, Norman TR. Discontinuation syndromes an autoradiographic analysis. J Clin Psychiatry 1987; 48 (suppl): 19–25.
with selective serotonin reuptake inhibitors: are there clinically 65 Meyer J, Kapur S, Eisfeld B, et al. The effect of paroxetine on
relevant differences? CNS Drugs 1999; 12: 171–77. 5-HT(2A) receptors in depression: an [(18)F]setoperone PET
48 Lader M, Tylee A, Donoghue J. Withdrawing benzodiazepines in imaging study. Am J Psychiatry 2001; 158: 78–85.
primary care. CNS Drugs 2009; 23: 19–34. 66 Haahr ME, Fisher PM, Jensen CG, et al. Central 5-HT4 receptor
49 Brett J, Murnion B. Management of benzodiazepine misuse and binding as biomarker of serotonergic tonus in humans:
dependence. Aust Prescr 2015; 38: 152–55. a [11C]SB207145 PET study. Mol Psychiatry 2014; 19: 427–32.
50 Chou TC, Talalay P. Quantitative analysis of dose-effect relationships: 67 Mawe GM, Hoffman JM. Serotonin signalling in the gut: functions,
the combined effects of multiple drugs or enzyme inhibitors. dysfunctions and therapeutic targets. Nat Rev Gastroenterol Hepatol
Adv Enzyme Regul 1984; 22: 27–55. 2013; 10: 473–86.
51 Brouillet E, Chavoix C, Bottlaender M, et al. In vivo bidirectional 68 Atkins GL, Nimmo IA. A comparison of seven methods for fitting
modulatory effect of benzodiazepine receptor ligands on the Michaelis-Menten equation. Biochem J 1975; 149: 775–77.
GABAergic transmission evaluated by positron emission 69 Lanzenberger R, Kranz GS, Haeusler D, et al. Prediction of SSRI
tomography in non-human primates. Brain Res 1991; 557: 167–76. treatment response in major depression based on serotonin
52 Ashton H. The treatment of benzodiazepine dependence. Addiction transporter interplay between median raphe nucleus and projection
1994; 89: 1535–41. areas. Neuroimage 2012; 63: 874–81.
53 Weller IVD, Ashby D, Brook R, et al. Report of the CSM Expert 70 Hieronymus F, Nilsson S, Eriksson E. A mega-analysis of fixed-dose
Working Group on the safety of selective serotonin reuptake trials reveals dose-dependency and a rapid onset of action for the
inhibitors. London: Medicines and Healthcare Products Regulatory antidepressant effect of three selective serotonin reuptake inhibitors.
Agency, 2005. Transl Psychiatry 2016; 6: e834.
54 Perahia DG, Kajdasz DK, Desaiah D, Haddad PM. 71 Baldinger P, Kranz GS, Haeusler D, et al. Regional differences in
Symptoms following abrupt discontinuation of duloxetine SERT occupancy after acute and prolonged SSRI intake investigated
treatment in patients with major depressive disorder. J Affect Disord by brain PET. Neuroimage 2014; 88: 252–62.
2005; 89: 207–12. 72 Hiemke C, Härtter S. Pharmacokinetics of selective serotonin
55 Michelson D, Fava M, Amsterdam J, et al. Interruption of selective reuptake inhibitors. Pharmacol Ther 2000; 85: 11–28.
serotonin reuptake inhibitor treatment. Double-blind, 73 Stockmann T, Odegbaro D, Timimi S, Moncrieff J. SSRI and SNRI
placebo-controlled trial. Br J Psychiatry 2000; 176: 363–68. withdrawal symptoms reported on an internet forum.
56 Zajecka J, Fawcett J, Amsterdam J, et al. Safety of abrupt Int J Risk Saf Med 2018; 29: 175–80.
discontinuation of fluoxetine: a randomized, placebo-controlled 74 Belaise C, Gatti A, Chouinard VA, Chouinard G. Patient online
study. J Clin Psychopharmacol 1998; 18: 193–97. report of selective serotonin reuptake inhibitor-induced persistent
57 Henry ME, Moore CM, Kaufman MJ, et al. Brain kinetics of postwithdrawal anxiety and mood disorders. Psychother Psychosom
paroxetine and fluoxetine on the third day of placebo substitution: 2012; 81: 386–88.
a fluorine MRS study. Am J Psychiatry 2000; 157: 1506–08. 75 Ferguson JM. SSRI antidepressant medications: adverse effects and
58 Yasui-Furukori N, Hashimoto K, Tsuchimine S, Tomita T. tolerability. Prim Care Companion J Clin Psychiatry 2001; 3: 22–27.
Characteristics of escitalopram discontinuation syndrome: 76 Preskorn SH. Clinically relevant pharmacology of selective
a preliminary study. Clin Neuropharmacol 2016; 39: 125–27. serotonin reuptake inhibitors. An overview with emphasis on
59 Stahl SM. Mechanism of action of serotonin selective reuptake pharmacokinetics and effects on oxidative drug metabolism.
inhibitors. Serotonin receptors and pathways mediate therapeutic Clin Pharmacokinet 1997; 32 (suppl 1): 1–21.
effects and side effects. J Affect Disord 1998; 51: 215–35. 77 Schatzberg AF, Blier P, Delgado PL, Fava M, Haddad PM,
60 Meyer JH, Wilson AA, Sagrati S, et al. Serotonin transporter Shelton RC. Antidepressant discontinuation syndrome:
occupancy of five selective serotonin reuptake inhibitors at different consensus panel recommendations for clinical management and
doses: an [11C]DASB positron emission tomography study. additional research. J Clin Psychiatry 2006; 67 (suppl 4): 27–30.
Am J Psychiatry 2004; 161: 826–35. 78 Perahia DG, Quail D, Desaiah D, Corruble E, Fava M. Switching to
61 Malberg JE, Schechter LE. Increasing hippocampal neurogenesis: duloxetine from selective serotonin reuptake inhibitor
a novel mechanism for antidepressant drugs. Curr Pharm Des 2005; antidepressants: a multicenter trial comparing 2 switching
11: 145–55. techniques. J Clin Psychiatry 2008; 69: 95–105.
62 Janssen DG, Caniato RN, Verster JC, Baune BT. © 2019 Elsevier Ltd. All rights reserved.
A psychoneuroimmunological review on cytokines involved in
antidepressant treatment response. Hum Psychopharmacol 2010;
25: 201–15.

546 www.thelancet.com/psychiatry Vol 6 June 2019