Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.

com

Open Access Research

Comparative efficacy and acceptability

of antidepressant treatment in
poststroke depression: a multiple-
treatments meta-analysis
Yefei Sun,1 Yifan Liang,2 Yang Jiao,2 Jueying Lin,3 Huiling Qu,4 Junjie Xu,5
Chuansheng Zhao2

To cite: Sun Y, Liang Y, Abstract

Jiao Y, et al. Comparative Strengths and limitations of this study
Objective  The aim of this study is to create a rank order
efficacy and acceptability of the comparative efficacy and acceptability (risk of
of antidepressant treatment ►► In the absence of head-to-head comparisons, it
all-cause discontinuation) of antidepressant treatment
in poststroke depression: provided us with a means of ranking different
in poststroke depression (PSD) by integrating direct and
a multiple-treatments antidepressants against each other for poststroke
meta-analysis. BMJ Open indirect evidence.
depression (PSD).
2017;7:e016499. doi:10.1136/ Design  Multiple-treatments meta-analysis of randomised
►► In the case of limited sample sizes, it strengthened
bmjopen-2017-016499 controlled trials.
inferences by integrating data from direct and
Participants  Patients with depression following stroke.
►► Prepublication history and indirect comparisons.
Interventions  10 antidepressants and placebo in the
additional material for this ►► The study included most of the antidepressants that
paper are available online. To acute treatment of PSD. are used in the clinical treatment of PSD.
view these files please visit the Outcome measures  The primary outcomes were the ►► The included randomised controlled trials had small
journal online (http://​dx.​doi.​ overall efficacy, defined as the mean change of the sample sizes and the number of studies on the same
org/​10.​1136/​bmjopen-​2017-​ total depression score. The secondary outcome was the kind of antidepressant was too small.
016499). acceptability, defined as risk of all-cause discontinuation. ►► The trials were for the treatment duration of 6–12
These estimates as standardised mean differences or ORs weeks.
Received 18 February 2017 with 95% CIs.
Revised 18 May 2017
Results  We identified 12 suitable trials, with data from
Accepted 5 June 2017
707 participants. All drugs were significantly more
effective than placebo apart from sertraline, nefiracetam Introduction
and fluoxetine. Most of the comparisons for acceptability Poststroke depression (PSD) is common,
revealed no significant differences except that paroxetine affecting approximately one-third of stroke
had significantly lower all-cause discontinuation than survivors.1 There is abundant evidence indi-
doxepin, citalopram and fluoxetine. Standardised mean cating that PSD is associated with increased
differences compared with placebo for efficacy varied
mortality and poor functional outcomes.2–5
from −6.54 for the best drug (reboxetine) to 0.51 for the
worst drug (nefiracetam). ORs compared with placebo
Although evidence has emerged from system-
for acceptability ranged from 0.09 for the best drug atic reviews to indicate that there are both
1
Department of Gastrointestinal (paroxetine) to 3.42 for the worst drug (citalopram). For validated depression screening tools6 and
Surgery, The First Hospital the efficacy rank, reboxetine, paroxetine, doxepin and effective treatment and prevention strategies
of China Medical University, duloxetine were among the most efficacious treatments, for depression after stroke,7–9 there has not
Shenyang, China the cumulative probabilities of which were 100%, been any significant reduction in the pooled
2
Department of Neurology, The
First Hospital of China Medical
85.7%, 83.2%, 62.4%, respectively. With respect to the frequency estimate of patients experiencing
University, Shenyang, China
acceptability rank, paroxetine, placebo, sertraline and PSD (ie, values were 33% in 200510 and 31%
3
Department of Emergency, nortriptyline were among the most acceptable treatments, in 201411). One reason is the high rates of
Zhongshan Hospital Xiamen the cumulative probabilities of which were 92.4%, 63.5%, refusal by stroke clinicians to recommend
University, Xiamen, China 57.3%, 56.3%. antidepressant therapy, because they consider
4
Department of Neurology, The Conclusion  After weighing the efficacy and acceptability,
the therapeutic efficacy of antidepressants
People’s Hospital of Liaoning we conclude that paroxetine might be the best choice
Province, Shenyang, China when starting acute treatment for PSD, and fluoxetine
for PSD treatment to be insignificant and
5
Department of Laboratory might be the worst choice. also being associated with a significant risk of
Medicine, The First Hospital
Trial registration number  This systematic review adverse events.12 Moreover, currently there
of China Medical University, are more than 40 different antidepressants
has been registered in the Prospective Register of
Shenyang, China
Systematic Review Protocols (PROSPERO) public in clinical use, which are divided into nine
Correspondence to database (CRD42017054741; http://www.​crd.​york.​ac.​uk/​ categories. Stroke clinicians seem to have
Yefei Sun; ​yfsun@​cmu.​edu.​cn PROSPERO). difficulties in making a rational choice about

Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499 1

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

Figure 1  PRISMA flow diagram. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; PSD,
poststroke depression. Reproduced with permission from Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009)
Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(7): e1000097. 

which antidepressant to prescribe. Neither the absolute been compared in a head-to-head manner.7 In addition,
nor the relative efficacy of antidepressants has been fully the number of included randomised controlled trials
established. There are even no recommendations in the (RCTs) is limited, which can introduce some bias into
guideline if any of these different drug classes of antide- any conclusions. Thus, it would be beneficial to create a
pressants is superior to the others.13 Therefore, whether rank order taking both efficacy and unwanted effects into
and what antidepressant treatment for PSD should be consideration.
prescribed remains controversial. Previous conventional Multiple-treatments meta-analysis is also known as
pairwise meta-analyses have not been able to generate mixed-treatment comparisons meta-analysis or network
clear rank orders for the efficacy and acceptability of avail- meta-analysis.14 It can provide us with a way to rank
able treatments, because many antidepressants have not different interventions against each other. It also helps

2 Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

inferences concerning the relative efficacy and accept-

ability of antidepressant treatment in PSD.

Methods
Criteria for considering studies
This systematic review has been registered in the Prospec-
tive Register of Systematic Review Protocols (PROSPERO)
public database (CRD42017054741; http://www.​ crd.​
york.​ac.​uk/​PROSPERO).
We included only RCTs that compared antidepressants
as monotherapy in the acute-phase treatment of patients
with PSD.
The patients with stroke  had to be diagnosed clini-
cally and/or by CT scan or MRI. They had also received a
diagnosis of depression following stroke, as confirmed by
Figure 2  Network plot of the included studies for the either Diagnostic and Statistical Manual of Mental Disor-
multiple-treatments meta-analysis for efficacy. The width of ders (DSM) criteria or some other validated rating scale
the lines is proportional to the number of studies comparing for depression. There were no inclusion restrictions on
each pair of treatments, and the size of each node is the basis of patient and study characteristics, such as age,
proportional to the number of randomised participants sex and classes of antidepressants.
(sample size). The network plot of included studies for
We also included placebo in the comparison group,
acceptability analysis is similar.
because there is no consensus about either the efficacy or
acceptability of antidepressant therapy.
to strengthen inferences of how large the differences Search methods and study selection
are between all the available interventions, since this We searched Medline, Embase, PsycINFO, Cochrane
approach integrates data from direct (when treatments Central Register of Controlled Trials, Web of Science
are compared within a randomised trial) and indirect (science and social science citation index) prior to
comparisons (when treatments are compared between December 2016. Further relevant trials were obtained by
trials).15 We aimed to compare the efficacy and accept- manual search of reference lists of all available records
ability of antidepressant treatment in PSD by conducting identified in the initial search. The authors were contacted
a multiple-treatments meta-analysis. It was intended for further information regarding unpublished trials and
to create the rank orders of different drugs to alleviate reports found in published databases. Keywords used
depression while taking into account the risk of all-cause in the searches were ‘depress* AND stroke’ (see online
discontinuation. Simultaneously, it would strengthen supplementary file 1). Various combinations of the search

Figure 3  Risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all
included studies.

Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499 3

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

Outcome measures and data extraction

The primary outcomes were the overall efficacy. We
defined overall efficacy as the mean change of the total
score of the Hamilton Depression Rating Scale (HDRS)
or Montgomery-Asberg Depression Rating Scale or Beck
Depression Inventory or Bech-Rafaelsen Melancholia
Scale or Zung Self-Rating Depression Scale (ZDS) from
baseline to end point. When trials reported results from
all of the above rating scales, we used them in the order
described above. Intention-to-treat datasets were used
whenever available.
The secondary outcome was the acceptability of the
antidepressant treatment. We defined acceptability of
treatment (treatment discontinuation) as the proportion
of patients who left the study early for any reason out of
the total number of patients randomly assigned to each
antidepressant.
Because a multiple-treatments meta-analysis requires
reasonable homogeneity, we focused on acute treatment,
which we defined as 6 weeks’ duration taking the main
guidelines into account.9 16 17 If 6-week data were not avail-
able, we used data from between 4 and 12 weeks (the data
point closest to 6 weeks was given preference).
In the included studies, data were extracted by two
reviewers (YS, HQ) independently. Once completed, any
disagreements on data extraction and study evaluation
were resolved through discussion. Information including
study name, study characteristics, patient characteristics,
depression diagnosis criteria, depression rating scale,
comparators, dose range, treatment duration, sample
size, effect sizes for two outcomes were extracted from
each included study. Corresponding authors would be
contacted for any missing information. If the article
could not provide the mean change of the total score, we
calculated the score based on a difference between the
end point and the initial value or by measuring graphs
presented in article.
We used network plots to explore the geometry of the
treatment network of the included studies. The width
of the lines is proportional to the number of studies
comparing each pair of treatments, and the size of each
node is proportional to the number of randomised partic-
ipants (sample size).
The Cochrane risk of bias tool was used to assess the
risk bias in the included studies.18 The tool contains seven
Figure 4  Risk of bias summary: review of the authors’ domains, which are random sequence generation, alloca-
judgements about each type of risk of bias for each included tion concealment, blinding of participants and personnel,
study.
blinding of outcome assessment, incomplete outcome
data, selective reporting and other bias. The judgement
for each domain includes a low risk of bias, a high risk of
terms were used, depending on the database. Studies were bias or unclear risk of bias. Two authors independently
restricted to those published and unpublished in English, evaluated the risk of studies.
prior to November 2016, and the ‘human’ study. Two
review authors independently decided on the selection Data synthesis and analysis
based on title and abstract. Any disagreement between We conducted two types of meta-analysis using a frequen-
review authors was resolved by discussion. If there was still tist model. First, we conducted a pairwise meta-analysis
some disagreement, a further reviewer and expert (CZ) for all direct comparisons with a random-effects
was consulted. model,19 assessing heterogeneity in these analyses with

4 Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

the I2 metric.20 Second, we fitted multiple-treatments be randomised to any of the interventions represented
meta-analysis models using the multivariate meta-regres- in the network.24 For the assessment of consistency
sion approach proposed by White et al,21–23 assuming a assumption, we initially conducted qualitative assess-
common heterogeneity variable for all comparisons (the ment based on clinical diversity, where we compared
tau (τ) value). τ is the estimated SD of underlying effects the distributions of characteristics that may modify the
of treatment across studies in a meta-analysis. The relative treatment effect. Subsequently, we evaluated the statis-
effect sizes were calculated as standardised mean differ- tical disagreement of direct and indirect evidence (also
ences (Hedges’ g) for continuous data (eg, the overall known as inconsistency) in two ways. One was a loop-spe-
efficacy) or as ORs for binary outcomes (eg, the accept- cific approach,25 26 in which inconsistent loops were
ability). Both types of effect sizes are reported with their identified as those yielding a 95% CI excluding zero. The
95% CI. other was the design-by-treatment interaction model
Consistency assumption is the key to multi- that provides a single inference, using the χ2 test, about
ple-treatments meta-analysis, which implies that the plausibility of assuming consistency throughout the
participants included in the network could hypothetically entire network.25 An inconsistency multiple-treatments

Figure 5  Efficacy and acceptability of the 11 antidepressants for PSD. Drugs are reported according to efficacy ranking.
Comparisons between treatments should be read from left to right and the estimate is in the cell in common between the
column-defining treatment and the row-defining treatment. For efficacy, SMDs lower than 0 favour the column-defining
treatment. For acceptability, ORs higher than 1 favour the column-defining treatment. To obtain SMDs for comparisons in the
opposite direction, negative values should be converted into positive values and vice versa. To obtain ORs for comparisons
in the opposite direction, reciprocals should be taken. Significant results are in bold and underlined. CIT, citalopram; DOX,
doxepin; DUL, duloxetine; FLU, fluoxetine; NEF, nefiracetam;  NOR, nortriptyline; PAR, paroxetine; PBA, placebo; PSD,
poststroke depression; REB, reboxetine; SER, sertraline; SMD, standard mean differences; TRA, trazodone.

Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499 5

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

Figure 6  Ranking for efficacy. Ranking indicates the

probability to be the best treatment, the second best, the
third best and so on, among the 11 antidepressants: (1) Figure 8  Clustered ranking plot for efficacy and
placebo; (2) citalopram; (3) doxepin; (4) duloxetine; (5) acceptability. CIT, citalopram; DOX, doxepin; DUL,
fluoxetine; (6) nefiracetam; (7) nortriptyline; (8) paroxetine; (9) duloxetine; FLU, fluoxetine; NEF, nefiracetam;  NOR,
reboxetine; (10) sertraline; (11) trazodone. nortriptyline; PAR, paroxetine; PBA, placebo; REB,
reboxetine; SER, sertraline; TRA, trazodone.

meta-analysis model would be fitted when statistical

inconsistency was present.
To rank the treatments for an outcome, we used surface as well as worsened quality of life of the patient and his/
under the cumulative ranking (SUCRA) probabilities, her caregiver,30 and therefore we defined the start of
which are expressed as a percentage the total efficacy or antidepressant treatment within 3 months after a stroke
acceptability of every intervention relative to an imaginary as early treatment. Meta-analyses were performed with
intervention that is always the best without uncertainty.23 Stata V.13.1 (StataCorp), using a suite of ‘network’23 and
Thus, large SUCRA scores should indicate a more effec- mvmeta command.22
tive or acceptable intervention.
The funnel plot was used to identify the possible publi-
cation bias if the number of studies was larger than 10. Results
We conducted several sensitivity analyses on the primary Study identification and characteristics
outcome to explore potential reasons for heterogeneity The Preferred Reporting Items for Systematic Reviews
or inconsistency, such as whether it was early treatment. It and Meta-Analyses flow diagram of studies selection was
has to be recalled that in the early stage (during the first depicted in figure 1. A total of 9651 references in the
3 to 4 months after a stroke), PSD poses serious problems, primary search were identified. After removal of the
such as worsened functional27 28 and vital prognoses27 29 duplicates, 4026 records were screened. Of these, 3989
were excluded based on screening of titles and abstracts

Figure 7  Ranking for acceptability. Ranking indicates

the probability to be the best treatment, the second best,
the third best and so on, among the 11 antidepressants: Figure 9  Comparison-adjusted funnel plot for efficacy. (01)
(1) placebo; (2) citalopram; (3) doxepin; (4) duloxetine; (5) citalopram; (02) doxepin; (03) duloxetine; (04) fluoxetine; (05)
fluoxetine; (6) nefiracetam; (7) nortriptyline; (8) paroxetine; (9) nefiracetam; (06) nortriptyline; (07) paroxetine; (08) placebo;
reboxetine; (10) sertraline; (11) trazodone. (09) reboxetine; (10) sertraline; (11) trazodone.

6 Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

blinding of outcome assessment,34 while eight studies

exhibited a low risk.33 35–37 39 41 42 All of studies had a
low risk of incomplete outcome data31–35 37–42 except for
one.36 Seven studies had a low risk of selectively reporting
results.32 34 36 38 39 41 42
Efficacy analysis and acceptability analysis
The results of the direct comparisons for efficacy and
acceptability analysis are shown in the online supplemen-
tary file 3.
Figure 5 summarises the results of the multiple-treat-
ments meta-analysis. With respect to overall efficacy, all
drugs were significantly superior to placebo (range of
mean effect sizes –0.59 to –6.54) apart from sertraline,
nefiracetam and fluoxetine. Reboxetine was signifi-
Figure 10  Comparison-adjusted funnel plot for cantly more effective than all of the other drugs. With
acceptability. (01) citalopram; (02) doxepin; (03) duloxetine; respect to acceptability (all-cause discontinuation),
(04) fluoxetine; (05) nefiracetam; (06) nortriptyline; (07) most of the comparisons revealed no significant differ-
paroxetine; (08) placebo; (09) reboxetine; (10) sertraline; (11) ences except that paroxetine had significantly lower
trazodone.
all-cause discontinuation than doxepin, citalopram and
fluoxetine.
and thus 37 were included in the narrative review, and
Figure 6 and 7 show the distribution of probabilities of
data from 12 of these studies31–42 were included in the
11 treatments being ranked for efficacy and acceptability,
meta-analysis. In the study conducted by Robinson et
respectively. For the efficacy rank (figure 6), reboxetine,
al,41 we selected the comparison between nefiracetam
paroxetine, doxepin, duloxetine were among the most
(600 mg) and placebo instead of the comparison between
nefiracetam (900 mg) and placebo, because there was no efficacious treatments, the cumulative probabilities of
statistically significant difference between the two doses. which were 100%, 85.7%, 83.2%, 62.4%, respectively.
Whereas, the group of patients receiving nefiracetam The following drugs were less effective; trazodone
(600 mg) seemed to enjoy greater efficacy when the treat- (59.8%), nortriptyline (54.5%), citalopram (35.7%),
ment duration was 9 weeks. sertraline (36.1%), placebo (16.3%), nefiracetam
The characteristics of 12 studies published from 1984 (14.1%), fluoxetine (2.3%). With respect to the accept-
to 2012 are shown in online supplementary file 2. A total ability rank (figure 7), paroxetine, placebo, sertraline
of 707 participants were included in the review and the and nortriptyline were among the most acceptable treat-
study sample size ranged from 22 to 123. The mean age ments, the cumulative probabilities of which were 92.4%,
of all patients was above 50. About half of the participants 63.5%, 57.3% and 56.3%. The following drugs were less
(51%) were female and 42% had a right-sided stroke loca- acceptable; trazodone (55.1%), reboxetine (54.6%),
tion. With respect to the nine studies31 33 35–40 42 in which nefiracetam (47.4%), duloxetine (41.8%), doxepin
a specific time of assessment after stroke was available, (31.6%), fluoxetine (27.5%) and citalopram (22.5%).
three studies32 34 41 were early treatment. Treatment dura- Figure 8 shows the clustered ranking based on the prob-
tion ranged from 4 weeks to 12 weeks. abilities of efficacy and acceptability. The exploration of
The network plot of the included studies for the inconsistency is described in detail in the online supple-
multiple-treatments meta-analysis for efficacy is shown mentary file 4.
in figure 2. A total of 11 antidepressants were eligible,
including placebo. Eighteen possible comparisons Reporting biases and sensitivity analyses
could be made, 12 of which were examined directly in Comparison-adjusted funnel plots for efficacy (figure 9)
1 study,and two of which (nortriptyline vs placebo, fluox- and acceptability (figure 10) reveal the reporting bias in
etine vs placebo) were examined directly in three studies. all 12 studies. Both of the plots were basically symmetrical.
The network plot of the included studies for acceptability Sensitivity analyses where the three studies with the
analysis was similar. early treatment with antidepressants for PSD (online
supplementary file 5) were excluded did not substan-
Risk of bias in included studies
tially change the results for efficacy and acceptability,
Figure 3 and 4 reveal the risk bias in all 12 studies.
Five studies described random sequence genera- excepting that citalopram and sertraline exchanged the
tion and adequate allocation concealment.33 36 37 39 41 ranking order in SUCRA rank for efficacy, and sertraline
Eight studies described blinding of participants and and nortriptyline exchanged the ranking order in SUCRA
personnel,33 35–37 39–42 while one study had a high risk rank for acceptability. In other words, sertraline became
of bias since it was an open label.34 The study, an open- more effective but less acceptable after removal of the
label trial, was also considered as a high risk of bias about three studies with the early treatments.

Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499 7

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

Discussion the subtype of PSD into account, because the inconsis-

Although PSD is one of the most common compli- tency test revealed no significant difference even after
cations after stroke and has been recognised by excluding those studies which identified the subtypes
psychiatrists for more than 100 years, controlled system- of the participants.37 39 It should be noted that the
atic studies did not begin until the 1970s.43 However, response of the patients with PSD  might be related
there is still no consensus on the efficacy of antidepres- to the subtype of depression, similarly to what was
sants to treat PSD, and the relevant RCTs are few. At observed for primary depression. 44 45 Therefore, it is
present, the Guidelines for Adult Stroke Rehabilitation important to emphasise that in this review, reboxetine
and Recovery: A Guideline for Healthcare Profes- was administered to patients with ‘retarded’ PSD.39
sionals from the American Heart Association/American With respect to sertraline, the indirect comparison
Stroke Association only indicated that treatment with was consistent with the direct comparison, although
heterocyclic antidepressant medications (tricyclic and the direct comparison of sertraline versus placebo
tetracyclic antidepressants (TCAs)) and selective sero- had been undertaken in patients with stroke  and
tonin reuptake inhibitors (SSRIs) were stated as viable minor depression and less severe major depression.37
options for PSD.13 However, we found that not every For this reason, we speculated that the result of
TCA or SSRI is viable. In our study, reboxetine, a novel sertraline may not be related to the severity of depres-
selective norepinephrine reuptake inhibitor (NARI) sion. However, sertraline became more effective but
was ranked first for efficacy followed by paroxetine less acceptable after removal of the three studies with
(SSRI), doxepin (TCA), duloxetine (serotonin and the early treatments. We cannot judge whether the
norepinephrine reuptake inhibitors (SNRI)), trazo- outcomes of early treatment is related to the severity
done (serotonin receptor antagonists and serotonin of depression. Furthermore, some studies have shown
reuptake inhibitors), nortriptyline (TCA), citalopram that PSD was related with other factors such as sex, 46
(SSRI), sertraline (SSRI), placebo, nefiracetam (nico- stroke location,47 which may influence the results.
tinic acetylcholine receptors (nAChR)), fluoxetine But it was not possible to carry out the relevant anal-
(SSRI). In terms of acceptability, paroxetine (SSRI), yses due to the lack of individual data. Finally, we did
placebo, sertraline (SSRI), nortriptyline (TCA) and not investigate important outcomes of acceptability,
reboxetine (NARI) were better tolerated than trazo- such as side effects, toxic effects and discontinuation
done (serotonin receptor antagonists and serotonin symptoms, as only a few studies reported this kind of
reuptake inhibitors), nefiracetam (nAChR), duloxe- data. Therefore, the treatment discontinuation may
tine (SNRI), doxepin (TCA), fluoxetine (SSRI) and not have been attributable only to the antidepressant
citalopram (SSRI). When weighing the efficacy and itself.
acceptability, it seems that paroxetine might be the best Conventional pairwise meta-analyses of the treatment
choice when starting an acute treatment for PSD, and of antidepressants in PSD were less,7 48–52 few of which
fluoxetine might be the worst choice. Here, acute treat- conducted meta-analysis based on different antidepres-
ment referred to the treatment duration of 6–12 weeks. sants. Tan et al48 showed that citalopram was superior to
There are several studies suggesting that the results the other SSRIs and TCAs in improving the total HDRS
of comparison between different drugs would change scores of PSD with acute treatment. The only other SSRIs
as the treatment duration became longer,33 34 36 40 41 included in that study were fluoxetine and sertraline.
which indicated that the results of the ranking may also Nonetheless, the results were consistent with our rank
change. Clinicians need to know whether (and to what order. The TCAs included also amitriptyline and imip-
extent) antidepressant drug treatments work within a ramine, which were not included in our analysis since
clinically reasonable period. we excluded Chinese articles. Xu et al7 demonstrated a
However, the results of the ranking are only for significant advantage associated with antidepressants in
reference and are subject to the following limita- comparison with placebo treatment in PSD (SMD=−0.96;
tions. First, although the examined related RCTs were 95%  CI −1.41  to −0.51; p<0.0001). In that previous
heterogeneous, typically they had small sample sizes. 1 meta-analysis, the subgroup analyses demonstrated
Due to the variety of antidepressants, the research that when compared with placebo, other antidepres-
objectives were somewhat fragmented, and the sants (SMD=−2.01; 95% CI −3.13 to −0.89; p=0.10) were
number of studies on the same kind of antidepressant better than TCAs (SMD=−1.41; 95% CI −2.51 to −0.31;
was too small, which meant that most of the compar- p=0.02) and SSRIs (SMD=−0.53; 95% CI −0.97 to −0.09;
isons in this review included only one study. Second, p<0.0001), which are not at odds with the results found
the results of inconsistency test for efficacy when here.
examined by the loop-specific approach indicated All in all, in the treatment of PSD, there are many
that the loop of fluoxetine versus nortriptyline versus unresolved factors that could influence the choice of
placebo was heterogeneous, and the inconsistency the optimal antidepressant compound that is, subtype
factor (IF) was obvious (IF=1.23) even though the of depression, sex, stroke location. One topic for future
loop was consistent. So we should cautiously explain exploration would be to determine whether the dura-
the result of the loop. In addition, we did not take tion of treatment affects the response. There is an urgent

8 Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

need for more high-quality research on antidepressant 14. Salanti G, Higgins JP, Ades AE, et al. Evaluation of networks of
randomized trials. Stat Methods Med Res 2008;17:279–301.
treatment in PSD. 15. Ades AE, Sculpher M, Sutton A, et al. Bayesian methods for evidence
synthesis in cost-effectiveness analysis. Pharmacoeconomics
Contributors  YS designed the study, extracted the data, wrote and approved 2006;24:1–19.
the manuscript. YL ran the network meta-analysis, revised and approved the 16. Eskes GA, Lanctôt KL, Herrmann N, et al.
manuscript. YJ and JL identified the relevant outcomes, discussed the validity Canadian Stroke Best Practice Recommendations: Mood, Cognition
and Fatigue Following Stroke practice guidelines, update 2015. Int J
of the trials, revised and approved the manuscript. HQ extracted the data,
Stroke 2015;10:1130–40.
proofread the results and approved the manuscript. JX reanalysed data to match 17. Intercollegiate Stroke Working Party. National clinical guideline for
inclusion/exclusion criteria, revised and approved the manuscript. CZ conducted stroke. 4th edition: Royal College of Physicians, 2012.
the systematic review, evaluated the quality of the studies, participated in the 18. Higgins J, Green S. Cochrane Handbook for Systematic Reviews
elaboration of the manuscript and approved the manuscript. of Interventions Version 5.1.0: The Cochrane Collaborationk, 2011.
www.​cochrane-​handbook.​org
Funding  This work was supported by The Program of Liaoning Province Education 19. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
Administration (no LK201605 to YS), The Shenyang Population and Health Trials 1986;7:177–88.
Technical Critical Special Project (no F16520659501 to CZ) and The Program of the 20. Higgins JPT, Green S. Cochrane handbook for systematic reviews of
Distinguished Professor of Liaoning Province, Neurology (to CZ). interventions, version 5.1.0, 2011. http://www.​cochrane-​handbook.​
org (accessed 9 Apr 2015).
Competing interests  None declared.
21. White IR. Multivariate random-effects meta-regression: updates to
Provenance and peer review  Not commissioned; externally peer reviewed. mvmeta. The STATA Journal 2011;11:255–70.
22. White IR, Barrett JK, Jackson D, et al. Consistency and inconsistency
Data sharing statement  No additional data are available. in network meta-analysis: model estimation using multivariate meta-
Open Access  This is an Open Access article distributed in accordance with the regression. Res Synth Methods 2012;3:111–25.
Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which 23. Chaimani A, Higgins JP, Mavridis D, et al. Graphical tools for network
meta-analysis in STATA. PLoS One 2013;8:e76654.
permits others to distribute, remix, adapt, build upon this work non-commercially,
24. Salanti G. Indirect and mixed-treatment comparison, network, or
and license their derivative works on different terms, provided the original work is multiple-treatments meta-analysis: many names, many benefits,
properly cited and the use is non-commercial. See: http://​creativecommons.​org/​ many concerns for the next generation evidence synthesis tool. Res
licenses/​by-​nc/4​ .​0/ Synth Methods 2012;3:80–97.
25. Veroniki AA, Vasiliadis HS, Higgins JP, et al. Evaluation of inconsistency
© Article author(s) (or their employer(s) unless otherwise stated in the text of the
in networks of interventions. Int J Epidemiol 2013;42:332–45.
article) 2017. All rights reserved. No commercial use is permitted unless otherwise 26. Salanti G, Marinho V, Higgins JP, et al. A case study of multiple-
expressly granted. treatments meta-analysis demonstrates that covariates should be
considered. J Clin Epidemiol 2009;62:857–64.
27. Robinson RG, Morris PL, Fedoroff JP, et al. Depression and
cerebrovascular disease. J Clin Psychiatry 1990;51:26–33.
References 28. Parikh RM, Robinson RG, Lipsey JR, et al. The impact of poststroke
1. Towfighi A, Ovbiagele B, El Husseini N, et al. Poststroke Depression: depression on recovery in activities of daily living over a 2-year
A Scientific Statement for Healthcare Professionals From the follow-up. Arch Neurol 1990;47:785–9.
American Heart Association/American Stroke Association. Stroke 29. Morris PL, Robinson RG, Andrzejewski P, et al. Association of
2017;48. depression with 10-year poststroke mortality. Am J Psychiatry
2. Wulsin L, Alwell K, Moomaw CJ, et al. Comparison of two depression 1993;150:124–9.
measures for predicting stroke outcomes. J Psychosom Res 30. Angeleri F, Angeleri VA, Foschi N, et al. The influence of depression,
2012;72:175–9. social activity, and family stress on functional outcome after stroke.
3. Willey JZ, Disla N, Moon YP, et al. Early depressed mood after stroke Stroke 1993;24:1478–83.
predicts long-term disability: the Northern Manhattan Stroke Study 31. Andersen G, Vestergaard K, Lauritzen L, et al. Effective treatment of
(NOMASS). Stroke 2010;41:1896–900. poststroke depression with the selective serotonin reuptake inhibitor
4. Kim JH, Park EY. The factor structure of the center for epidemiologic citalopram. Stroke 1994;25:1099–104.
studies depression scale in stroke patients. Top Stroke Rehabil 32. Chen W, Wang G-F, Chen X-H, et al. Effects of paroxetine on function
2012;19:54–62. recovery in patients with poststroke depression. Chinese J Clin
5. Jørgensen TS, Wium-Andersen IK, Wium-Andersen MK, et al. Rehabil 2002;6:2014–5.
Incidence of Depression After Stroke, and Associated Risk Factors 33. Fruehwald S, Gatterbauer E, Rehak P, et al. Early fluoxetine treatment
and Mortality Outcomes, in a Large Cohort of Danish Patients. JAMA of post-stroke depression--a three-month double-blind placebo-
Psychiatry 2016;73:1032–40. controlled study with an open-label long-term follow up. J Neurol
6. Meader N, Moe-Byrne T, Llewellyn A, et al. Screening for poststroke 2003;250:347–51.
major depression: a meta-analysis of diagnostic validity studies. J 34. Karaiskos D, Tzavellas E, Spengos K, et al. Duloxetine versus
Neurol Neurosurg Psychiatry 2014;85:198–206. citalopram and sertraline in the treatment of poststroke depression,
7. Xu XM, Zou DZ, Shen LY, et al. Efficacy and feasibility of anxiety, and fatigue. J Neuropsychiatry Clin Neurosci
antidepressant treatment in patients with post-stroke depression. 2012;24:349–53.
Medicine 2016;95:e5349. 35. Kimura M, Robinson RG, Kosier JT, et al. Treatment of cognitive
8. Hackett ML, Anderson CS, House A, et al. Interventions for impairment after poststroke depression : a double-blind treatment
preventing depression after stroke. Cochrane Database Syst Rev trial. Stroke 2000;31:1482–6.
2008:CD003689. 36. Lipsey JR, Robinson RG, Pearlson GD, et al. Nortriptyline
9. Holloway RG, Arnold RM, Creutzfeldt CJ, et al. Palliative and end- treatment of post-stroke depression: a double-blind study. Lancet
of-life care in stroke: a statement for healthcare professionals from 1984;1:297–300.
the American Heart Association/American Stroke Association. Stroke 37. Murray V, von Arbin M, Bartfai A, et al. Double-blind comparison
2014;45:1887–916. of sertraline and placebo in stroke patients with minor depression
10. Hackett ML, Yapa C, Parag V, et al. Frequency of depression and less severe major depression. J Clin Psychiatry
after stroke: a systematic review of observational studies. Stroke 2005;66:708–16.
2005;36:1330–40. 38. Raffaele R, Rampello L, Vecchio I, et al. Trazodone therapy of
11. Hackett ML, Pickles K. Part I: frequency of depression after stroke: the post-stroke depression. Arch Gerontol Geriatr 1996;22(Suppl
an updated systematic review and meta-analysis of observational 1):217–20.
studies. Int J Stroke 2014;9:1017–25. 39. Rampello L, Alvano A, Chiechio S, et al. An evaluation of efficacy
12. Hackett ML, Anderson CS, House A, et al. Interventions for and safety of reboxetine in elderly patients affected by "retarded"
treating depression after stroke. Cochrane Database Syst Rev post-stroke depression. A random, placebo-controlled study. Arch
2008:CD003437. Gerontol Geriatr 2005;40:275–85.
13. Winstein CJ, Stein J, Arena R, et al. Guidelines for Adult 40. Robinson RG, Schultz SK, Castillo C, et al. Nortriptyline versus
Stroke Rehabilitation and Recovery: A Guideline for Healthcare fluoxetine in the treatment of depression and in short-term recovery
Professionals From the American Heart Association/American Stroke after stroke: a placebo-controlled, double-blind study. Am J
Association. Stroke 2016;47:e98–e169. Psychiatry 2000;157:351–9.

Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499 9

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Open Access

41. Robinson RG, Jorge RE, Clarence-Smith K, et al. Double-blind 47. Wei N, Yong W, Li X, et al. Post-stroke depression and lesion
randomized treatment of poststroke depression using nefiracetam. J location: a systematic review. J Neurol 2015;262:81–90.
Neuropsychiatry Clin Neurosci 2008;20:178–84. 48. Tan S, Huang X, Ding L, et al. Efficacy and safety of citalopram
42. Wiart L, Petit H, Joseph PA, et al. Fluoxetine in early poststroke in treating post-stroke depression: a meta-analysis. Eur Neurol
depression: a double-blind placebo-controlled study. Stroke 2015;74:188–201.
2000;31:1829–32. 49. Bhogal SK, Teasell R, Foley N, et al. Heterocyclics and selective
43. Robinson RG, Jorge RE. Post-Stroke Depression: a Review. Am J serotonin reuptake inhibitors in the treatment and prevention of
Psychiatry 2016;173:221–31. poststroke depression. J Am Geriatr Soc 2005;53:1051–7.
44. Rampello L, Nicoletti G, Raffaele R, et al. Comparative effects 50. Hackett ML, Anderson CS, House AO, et al. Management of
of amitriptyline and amineptine in patients affected by anxious depression after stroke: a systematic review of pharmacological
depression. Neuropsychobiology 1995;31:130–4. therapies. Stroke 2005;36:1092–7.
45. Rampello L, Nicoletti G, Raffaele R, et al. Dopaminergic hypothesis 51. Chen Y, Guo JJ. Meta-analysis of antidepressant treatment for
for retarded depression: a symptom profile for predicting patients with poststroke depression. Stroke 2006;37:1365–6.
therapeutical responses. Acta Psychiatr Scand 1991;84:552–4. 52. Chen Y, Guo JJ, Zhan S, et al. Treatment effects of antidepressants
46. Alajbegovic A, Djelilovic-Vranic J, Nakicevic A, et al. Post stroke in patients with post-stroke depression: a meta-analysis. Ann
depression. Med Arch 2014;68:47–50. Pharmacother 2006;40:2115–22.

10 Sun Y, et al. BMJ Open 2017;7:e016499. doi:10.1136/bmjopen-2017-016499

 Downloaded from http://bmjopen.bmj.com/ on August 6, 2017 - Published by group.bmj.com

Comparative efficacy and acceptability of

antidepressant treatment in poststroke
depression: a multiple-treatments
meta-analysis
Yefei Sun, Yifan Liang, Yang Jiao, Jueying Lin, Huiling Qu, Junjie Xu and
Chuansheng Zhao

BMJ Open 2017 7:

doi: 10.1136/bmjopen-2017-016499

Updated information and services can be found at:

http://bmjopen.bmj.com/content/7/8/e016499

These include:

References This article cites 46 articles, 11 of which you can access for free at:
http://bmjopen.bmj.com/content/7/8/e016499#BIBL
Open Access This is an Open Access article distributed in accordance with the Creative
Commons Attribution Non Commercial (CC BY-NC 4.0) license, which
permits others to distribute, remix, adapt, build upon this work
non-commercially, and license their derivative works on different terms,
provided the original work is properly cited and the use is
non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Email alerting Receive free email alerts when new articles cite this article. Sign up in the
service box at the top right corner of the online article.

Topic Articles on similar topics can be found in the following collections

Collections Evidence based practice (722)
Mental health (690)
Neurology (404)

Notes

To request permissions go to:

http://group.bmj.com/group/rights-licensing/permissions

To order reprints go to:

http://journals.bmj.com/cgi/reprintform

To subscribe to BMJ go to:

http://group.bmj.com/subscribe/