The British Journal of Psychiatry (2020)

Page 1 of 8. doi: 10.1192/bjp.2020.198

Review

Psychological interventions as an
alternative and add-on to
antidepressant medication to prevent
depressive relapse: systematic review
and meta-analysis
Josefien Johanna Froukje Breedvelt, Maria Elisabeth Brouwer, Mathias Harrer, Maria Semkovska,
David Daniel Ebert, Pim Cuijpers and Claudi Louisa Hermina Bockting

Background Results
After remission, antidepressants are often taken long term to Among 11 included trials (n = 1559), we did not observe an
prevent depressive relapse or recurrence. Whether psycho- increased risk of relapse for participants receiving a psycho-
logical interventions can be a viable alternative or addition to logical intervention while tapering antidepressants versus anti-
antidepressants remains unclear. depressants alone (RR = 1.02, 95% CI 0.84–1.25; P = 0.85).
Psychological interventions added to antidepressants signifi-
Aims cantly reduced the risk of relapse (RR = 0.85, 95% CI 0.74–0.97;
P = 0.01) compared with antidepressants alone.
To compare the effectiveness of psychological interventions as
an alternative (including delivered when tapering antidepres- Conclusions
sants) or addition to antidepressants alone for preventing
This study found no evidence to suggest that adding a psycho-
depressive relapse.
logical intervention to tapering increases the risk of relapse when
compared with antidepressants alone. Adding a psychological
Method intervention to antidepressant use reduces relapse risk signifi-
Embase, PubMed, the Cochrane Library and PsycINFO were cantly versus antidepressants alone. As neither strategy is rou-
searched from inception until 13 October 2019. Randomised tinely implemented these findings are relevant for patients,
controlled trials (RCTs) with previously depressed patients in clinicians and guideline developers.
(partial) remission where preventive psychological interventions
with or without antidepressants (including tapering) were com- Keywords
pared with antidepressant control were included. Data were Depressive disorders; meta-analysis; pharmacotherapy; relapse
extracted independently from published trials. A random-effects prevention; recurrent depression.
meta-analysis on time to relapse (hazard ratio, HR) and risk
of relapse (risk ratio, RR) at the last point of follow-up was Copyright and usage
conducted. PROSPERO ID: CRD42017055301. © The Author(s), 2020. Published by Cambridge University Press
on behalf of the Royal College of Psychiatrists.

Major depressive disorder (MDD) is often characterised by a chronic the risk of relapse of depression when added to tapering or combined
and recurrent nature.1,2 After the first episode of MDD, there is a 40– with antidepressants.7,10,11 Yet, to the best of our knowledge, a meta-
60% risk that a person relapses,1,2 and this risk rises by 16% following analysis comparing psychological intervention as an alternative or
each successive episode.3 Ultimately, the cumulative lifetime risk may additive to antidepressants alone is lacking. The three reviews10–12
increase up to 90% after three episodes or more.1,3 It is of increasing published thus far were conducted on a restricted sample of studies
importance to prevent relapse of depression in clinical practice, given (n = 4) comparing psychological interventions without antidepres-
its substantial impact on public health. sants with antidepressants alone,11 included studies with unmasked
International clinical guidelines recommend long-term anti- (unblinded) outcome assessments10–12 or did not report on the spe-
depressant medication beyond 9 months (USA)4 or 2 years (UK)5 cific comparison, that is, psychological interventions with or without
for patients with at least two or three previous episodes. antidepressants compared with antidepressants alone.12
Psychological interventions, that is, cognitive–behavioural therapy In this study, we aim to remediate the above limitations and
(CBT) or mindfulness-based cognitive therapy (MBCT), are recom- conduct an up-to-date systematic review and meta-analysis to
mended as an addition to antidepressants for high-risk populations, compare the effects of psychological interventions as an alternative
including people with multiple previous episodes.4,5 In clinical practice, (psychological intervention alone) and as an addition to antidepres-
MDD appears to be mostly managed with antidepressant medication sants (psychological intervention plus maintenance antidepres-
alone.6 Taking antidepressants long-term is not a panacea. Relapse sants) versus long-term antidepressant use on relapse/recurrence
rates on antidepressants vary by population but can rise to 60% after in depression.
24 months.7 Antidepressant use is associated with side-effects,4,5 and
patients might often prefer psychological treatments to continuing
antidepressants.8 Patients who attempt to taper off antidepressants Method
after remission are at an increased risk of depressive relapse.9
Thus, alternatives to long-term antidepressant use should be Search strategy
explored to optimise outcomes for patients. Recent studies suggest The PRISMA guidance was followed in the reporting of this
that psychological interventions appear to be effective in reducing review13 (PROSPERO ID: CRD42017055301). Search strings were

1
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
Breedvelt et al

developed with a health sciences librarian (supplementary baseline, co-interventions, adherence and similar timing of outcome
Appendix, available at https://doi.org/10.1192/bjp.2020.198). assessment. Studies were rated as ‘low risk’, ‘high risk’ or ‘unclear
MeSH terms, Boolean operators and free text were used to identify risk’. Studies were scored with a total ‘low risk’ rating and qualified
studies in Embase, the Cochrane Library, PubMed and PsychInfo. as at overall low risk of bias (an indication of high quality) if six or
The last search was conducted on the 13 October 2019. more risk of bias variables were assessed as low risk. In other cases,
Covidence14 was used to manage the study screening and selection the study was scored ‘high risk’.
process. Duplicates were removed before the title and abstract
screening. Reference lists and prior meta-analyses were searched
for relevant literature. Titles and abstracts were screened independ- Meta-analysis
ently by J.B., V.Z., E.B. and A.S.. After screening, two independent Comprehensive Meta-Analysis version 3.0 (Biostat.org) for
researchers (J.B. and M.B.) conducted the final inclusion of Windows20 was used to analyse the data extracted from included
studies. Disagreements were resolved by consulting a third studies and calculate pooled effect sizes, forest plots, heterogeneity
researcher (C.B.). and funnel plots. For recurrence of depression, 2 × 2 tables (events
and non-events for intervention and control) were converted into
Study criteria risk ratios (RR) and 95% confidence intervals (CI). HR outcome
Randomised controlled trials (RCTs) with participants aged 18–65 data were calculated separately, as HRs measure time to relapse,
who had at least one prior diagnosis of MDD were included. Studies and RR and HR outcomes are non-comparable in meta-analysis.
comparing psychological interventions with and without antide- Between-study heterogeneity was expected to be high; thus, a
pressants versus antidepressant control were included. For the random-effects model was applied using the DerSimonian & Laird
studies evaluating psychological interventions without antidepres- method used to estimate the between-study variance τ2.21 A mixed-
sants versus antidepressants alone, psychological interventions effects model was used to perform subgroup analysis on categorical
could be delivered either on their own or while tapering antidepres- variables for differences between groups. Subgroup analysis included
sants. The participants had to be randomised to a relapse prevention the different treatment combinations of psychological interventions
intervention after response, remission or recovery.15,16 Remission with or without (including tapering) antidepressants and compari-
was defined as a period of at least 8 weeks during which participants sons between different psychological intervention types. Meta-regres-
had no or subclinical symptoms (i.e. Hamilton Rating Scale for sion analysis was conducted on follow-up duration to assess whether
Depression score ≤7, Beck Depression Inventory-II score ≤13, the time point of follow-up affected the results of the meta-analysis.
Patient Health Questionnaire score ≤10).15 Relapse was defined as
an increase in symptomatology during remission, and recurrence
as an increase in symptomatology during recovery.15,16 In this Multiple study arms
paper we use relapse to describe both recurrence and relapse. Two studies included psychological intervention both without and
Sequential treatment combinations15 were included provided that with antidepressants versus antidepressant control.7,22 To avoid
participants in the intervention group achieved remission or double counting,19 we halved the control group relapse rates and
response, according to the authors of the study. Studies were sample size for these comparisons only in the subgroup analysis
included if participants’ primary presenting problem was depres- comparing psychological interventions with or without antidepres-
sion (as opposed to depression secondary to other (mental) health sants versus control.
conditions). Participants were excluded if they were in active treat-
ment for another mental disorder as classified by DSM-IV.17
Outcome measurement (relapse or recurrence of a depressive Number needed to treat
episode) had to be conducted by an independent assessor via a clin-
The number needed to treat (NNT) was calculated from the inverse
ical diagnostic assessment.
of the risk difference to indicate the numbers needed to treat with
the relapse prevention intervention to prevent one relapse.
Data extraction
Study data were extracted in a pre-piloted extraction table by three
reviewers (J.B., C.M. and M.H.). Discrepancies were discussed and Heterogeneity
resolved by consensus. Study characteristics, sample characteristics, Heterogeneity was assessed using I2, with 0–24% indicating no sig-
intervention characteristics (including tapering or continuation of nificant heterogeneity, 25–49% low heterogeneity, 50–74% moder-
antidepressants), delivery information and the definition of recur- ate heterogeneity and 75–100% high heterogeneity.24 The 95%
rence or relapse were extracted. Extracted outcome data included confidence intervals25 of I2 were calculated using the formula
the length of follow-up, rates of relapse or recurrence in intervention from Stata’s HETEROGI module.26
and control group at the last point of follow-up and the hazard ratio
(HR) for each relevant comparison.
Risk of bias and publication bias
Risk of bias assessment Risk of bias was investigated by a subgroup analysis of the overall
Study quality was assessed independently by three reviewers. J.B. risk of bias score for each of the main comparisons, i.e. a psycho-
and M.B. conducted the risk of bias assessments independently. logical intervention with antidepressants versus a psychological
Six criteria for risk of bias (on nine domains) were used from intervention without antidepressants (including tapering).
the risk of bias assessment tool developed by the Cochrane Publication bias or small sample bias was assessed by visual inspec-
Collaboration.18,19 The following criteria were applied: sequence tion of the funnel plot. Egger’s test of the intercept27 was applied to
generation; allocation concealment; masking of participants, per- test for asymmetry of the funnel plot, which would imply that bias
sonnel and outcome assessors; incomplete outcome data drop-out may be present. Duval & Tweedie’s28 trim and fill procedure was
and intention-to-treat analysis; selective outcome reporting; and applied to assess whether imputation of studies to address potential
other threats to validity, including similarity of the groups at publication bias would change the effect size of the meta-analysis.

2
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
 Psychological interventions in the prevention of depressive relapse

Identification Records identified Duplicates removed

(n = 22 067) (n = 7404)

Records screened
(n = 14 663) Records excluded
Screening

(n = 14 438)

Full-text articles assessed

for eligibility Full-text articles excluded, with
Eligibility

(n = 225) reasons (n = 214)

50 Wrong patient population
28 Wrong outcomes
19 Duplicate
25 Wrong study design
17 Protocol
14 Wrong intervention
11 Review
29 Wrong comparator
8 Not original / preliminary
Included

Studies included in findings

quantitative synthesis (n = 11) 8 Paediatric population
2 Poster
2 Wrong language
1 Bookchapter

Fig. 1 PRISMA study selection process.

the psychological interventions studied is given in the supplementary

Results
Appendix. In five of the comparisons, antidepressants were tapered in
the psychological intervention group, and in one comparison, antide-
Characteristics of included studies
pressants were not taken or tapered. In the remaining comparisons,
After removing duplicates, 14 663 records were screened. Out of 225 continuation antidepressant medication was delivered in combin-
full-text articles, 11 studies with 1559 participants (n = 832 inter- ation with a psychological intervention.
vention group and n = 727 control group) met the inclusion criteria
(Fig. 1). The studies were conducted in the USA (n = 4), the UK
(n = 3), The Netherlands (n = 2), Canada (n = 1) and Germany
(n = 1). The follow-up periods in the studies ranged from 28 to Treatment strategies to prevent relapse
235 weeks, with an average of 97 weeks. Two tables summarising Psychological interventions without antidepressants (including
study and participant characteristics of included studies can be tapering of antidepressants) versus antidepressants alone
found in the supplementary Appendix. Most of the included Six studies (n = 948) compared psychological interventions without
patients were at high risk of relapse; 54% of the studies included par- antidepressants versus antidepressants alone (RR = 1.02; 95% CI
ticipants with at least three previous episodes of depression; 27% of 0.4–1.25; P = 0.85 (Table 1 and Fig. 2)). This implies that there
the studies included participants with at least two previous episodes; was no significant difference in effect between psychological inter-
and only two studies (18%) included participants who had experi- ventions without antidepressants and antidepressants alone.
enced at least one previous episode. Most studies (63%) were Heterogeneity was moderate: I2 = 51 (95% CI 0–80). One compari-
judged to have a low risk of bias (supplementary Appendix). son evaluating the psychological intervention IPT versus antide-
In two studies,7,22 two relapse prevention intervention groups pressants alone could be considered an outlier because of the high
were compared with antidepressants alone. Therefore, 13 compari- RR found (RR = 2.87; 95% CI 1.33–6.21).22 After removing IPT,
sons in total were possible for the analyses. Four of these comparisons the overall effect on five comparisons changed to RR = 0.97 (95%
studied CBT or MBCT, two studied interpersonal psychotherapy CI 0.85–1.10; P = 0.61) and heterogeneity lowered to I2 = 0 (95%
(IPT) or preventive cognitive therapy (PCT) and one studied con- CI 0–79). Although the effect size changed, the conclusion from
tinuation cognitive therapy (C-CT) versus control. A description of the main effect analysis did not change. The overall risk of bias

3
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
Breedvelt et al

Table 1 Risk ratios (RR) for the effects of psychological interventions with or without antidepressant medication versus antidepressants alone

Number P for
Number of needed to subgroup
Comparison comparisons RR (95% CI) I2 (95% CI) P treat analysisa
(1) Psychological intervention without 6 1.02 (0.84–1.25) 51 (0–80) 0.85 −36 0. 17
antidepressants v. antidepressants
(2) Psychological intervention combined with 7 0.85 (0.74–0.97) 0 (0–71) 0.01 15
antidepressants v. antidepressants
v., versus; n.a., not applicable.
a. P-value for subgroup analysis on risk ratios comparing psychological interventions with and without antidepressants versus antidepressants alone.

was low in five out of the six comparisons, and the effect size did not Hazard ratio results
vary significantly by the risk of bias score. Six of the 11 included studies (seven comparisons) provided data on
Five of the six studies (n = 840) compared antidepressant taper- HR (Table 2 and Fig. 4). The results were in line with findings from
ing when receiving a psychological intervention with antidepres- the RR analysis in our previous paragraphs, although a smaller
sants alone. A main-effect meta-analysis on these studies found number of studies was available. Four comparisons were available
no significant difference between either approach (RR = 1.01; 95% where psychological interventions without antidepressants (includ-
CI 0.79–1.29; P = 0.94; NNT = −37). Heterogeneity was moderate ing antidepressant tapering) were compared with antidepressant
(I2 = 59; 95% CI 1–83) and the risk of bias was low in all medication (HR = 0.84; 95% CI 0.69–1.03; P = 0.09; I2 = 0; 95% CI
comparisons. One study compared a psychological intervention 0–79). Two comparisons were available where antidepressants
without antidepressants versus antidepressants alone. We did not were combined with a psychological intervention and compared
further compare the effects of psychological interventions with with antidepressants alone (HR = 0.65; 95% CI 0.44–0.96;
tapering or psychological interventions alone owing to the small P = 0.03; I2 = 0). Only one study had a high risk of bias; no
sample size. further subgroup analyses were conducted owing to the small
number of studies available.

Psychological interventions with antidepressants versus

Psychological interventions with or without antidepressants versus
antidepressants alone
antidepressants alone
Overall, seven studies (n = 611) compared a psychological interven-
Subgroup analysis conducted between psychological interventions
tion with antidepressants versus antidepressants alone. There was a
with and without antidepressants did not identify a significant dif-
significant difference (RR = 0.85; 95% CI 0.74–0.97; P = 0.01), with
ference between either approach for RRs (Table 1) and HRs
combination therapy (psychological intervention plus antidepres-
(Table 2). We did not conduct subgroup analyses on the different
sants) resulting in a 15% lower relative risk of relapse compared
psychological intervention types owing to the small numbers avail-
with antidepressants alone (Table 1 and Fig. 3). The NNT was 15
able for some comparisons.
and heterogeneity was not significant (I2 = 0; 95% CI 0–71). The
overall quality of evidence was mixed, with four comparisons
showing a high risk of bias and three a low risk. Subgroup analysis Sensitivity analyses
showed no different effect size in relation to the risk of bias score at To study whether previous treatment exposure (i.e. electroconvulsive
P < 0.10. therapy, psychotherapy) and study design (sequential or

Study name Time point Comparison Risk ratio and 95% CI

Risk
ratio

Jarrett (2013)30 104 CT vs. ADM 1.111

1.111
Bockting (2018)7 104 PCT/-ADM vs. ADM 1.055
Frank (1990)22 156 IPT/-ADM vs. ADM 2.872
Kuyken (2008)38 65 MBCT /-ADM vs. ADM 0.797
Kuyken (2015)39 104 MBCT /-ADM vs. ADM 0.940
Segal (2010) 40
78 MBCT /-ADM vs. ADM 0.826
1.009
1.026
0.1 0.2 0.5 1 2 5 10

Favours intervention Favours control

Fig. 2 Forest plot of the effects of psychological interventions alone or with tapering versus antidepressants alone on risk ratios.
CT, cognitive therapy; ADM, antidepressant medication; PCT/-ADM, preventive cognitive therapy with tapering of antidepressant medication; IPT/-ADM, interpersonal therapy with
tapering of antidepressant medication; MBCT/-ADM, mindfulness-based cognitive therapy with tapering of antidepressant medication.

4
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
 Psychological interventions in the prevention of depressive relapse

Study name Time point Comparison Risk ratio and 95% CI

Risk
ratio

Bockting (2018)7 104 PCT+ ADM vs. ADM 0.710

Frank (1990)22 156 IPT + ADM vs. ADM 1.120
Huijbers (2015)34 65 MBCT + ADM vs. ADM 0.979
Paykel (1999)33 and (2005)34 235 CBT+ ADM vs. ADM 0.900
Perlis (2002)29 28 CBT + ADM vs. ADM 0.800
Petersen (2007)41 80 CBT + ADM vs. ADM 1.273
Brakemeier (2014)31 26 CT + ADM vs. ADM 0.529
0.848
0.1 0.2 0.5 1 2 5 10

Favours intervention Favours control

Fig. 3 Forest plot of the effects of psychological interventions with antidepressants versus antidepressants alone on risk ratios.
PCT + ADM, preventive cognitive therapy with antidepressant medication; ADM, antidepressant medication; PCT + ADM, preventive cognitive therapy with antidepressant
medication; MBCT + ADM, mindfulness-based cognitive therapy with antidepressant medication; CBT + ADM, cognitive–behavioural therapy with antidepressant medication;
CT + ADM, cognitive therapy with antidepressant medication.

maintenance) affected outcomes, we conducted a sensitivity analysis on the left side of the mean, suggesting that non-significant study
to compare results with and without studies with a maintenance or results favouring the control group were not published.
continuation design.22,29–33 No difference in results was found.
Moreover, for both HR and RR analyses, meta-regression analyses
were conducted to assess whether the risk of relapse could be pre- Risk of bias
dicted on the basis of time to follow-up. In both analyses, as fewer Overall, the risk of bias in the 11 studies was low, with 7 studies
than ten studies were available the results are highly preliminary. having low risk and 4 studies scoring high risk. All individual
For RRs, a covarying trend towards significance for time to follow- ratings and overall ratings on the risk of bias domains can be
up was found in the psychological interventions without antidepres- found in Tables 3 and 4 in the supplementary Appendix. It is note-
sants versus antidepressants alone comparison ((Q(1,5) = 10.24, P = worthy that more recent trials (published since 2010) had low risk as
0.07)). The effect was small; over time the risk of relapse increased, they incorporated procedures that decreased the risk, including pre-
with 0.01 more for patients who received a psychological interven- registering the trial protocol. For all studies, risk of masking bias for
tion (with or without antidepressants) compared with those who both participants and personnel was scored as ‘high’; this is due to
did not for each additional week of follow-up. For HRs, no covarying the nature of this type of research, where masking is often not pos-
effect of time of follow-up on outcome was found. sible. Other studies had problems pertinent to the trial: the trial of
Huijbers et al (2015)34 had no allocation concealment; the trial of
Jarrett et al (2013)30 had differences in length of follow-up and
Publication bias number of sessions attended; and Petersen et al (2007)41 was in
Although the funnel plot suggested some asymmetry, Egger’s test and part funded by a pharmaceutical company.
Duval & Tweedie’s trim and fill indicated no potential publication
bias in studies comparing psychological interventions without antide-
pressants versus those with antidepressants (studies trimmed: 0). Discussion
Potential publication bias was found in comparisons combining a
psychological intervention with antidepressants versus antidepres- This is, to the best of our knowledge, the first systematic review to
sants alone (studies trimmed: 1), yet the change in the adjusted RR assess whether psychological interventions can be delivered as an
was minimal (adjusted RR = 0.84; 95% CI 0.74–0.96). For HR com- alternative to antidepressants, or whether a combination of psycho-
paring psychological interventions without antidepressants versus logical intervention and antidepressants may be more effective than
antidepressants alone, there was some indication of publication medication alone. Two key findings can be reported. First, we found
bias, and 1 study was trimmed. Owing to the limited number of no evidence for an increased risk of relapse in the groups where a
studies (n = 2), it was not possible to analyse publication bias for psychological intervention was delivered when participants were
the comparison of psychological intervention with antidepressants tapering antidepressants versus where participants continued
versus antidepressant medication alone. All trimmed studies were taking antidepressants. Second, adding a psychological intervention

Table 2 Hazard ratios (HR) for the effects of psychological interventions with or without antidepressant medication versus antidepressants alone

P for subgroup
Comparison Number of comparisons HR (95% CI) I2 (95% CI) P analysisa
(1) Psychological intervention without antidepressants v. antidepressants 5 0.84 (0.69–1.03) 0 (0–79) 0.09 0.25
(2) Psychological intervention with antidepressants v. antidepressants 2 0.65 (0.44–0.96) 0 n.a. 0.03
v., versus; n.a., not applicable.
a. P-value for subgroup analysis on hazard ratios comparing psychological interventions with and without antidepressants versus antidepressants alone.

5
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
Breedvelt et al

Study name Time point Subgroup within study Statistics for each study Hazard ratio and 95% CI
Hazard Lower Upper
ratio limit limit Z-Value p-Value

Kuyken (2008)38 65 MBCT /- ADM vs. ADM 0.630 0.386 1.029 –1.847 0.065
Kuyken (2015)39 104 MBCT /- ADM vs. ADM 0.890 0.671 1.181 –0.807 0.420
Bockting (2018)7 104 PCT /-ADM vs. ADM 0.860 0.560 1.320 –0.690 0.491
Jarrett (2013)30 104 C-CT vs. ADM 0.930 0.542 1.596 –0.263 0.792
Segal (2010)40 78 MBCT /- ADM vs. ADM 1.070 0.252 4.535 0.092 0.927
0.844 0.694 1.027 –1.697 0.090
Bockting (2018)7 104 PCT+ ADM vs. ADM 0.590 0.375 0.928 –2.284 0.022
Huijbers (2015)34 65 MBCT + ADM vs. ADM 0.870 0.399 1.896 –0.350 0.726
0.651 0.440 0.963 –2.150 0.032
0.785 0.625 0.986 –2.079 0.038
0.1 0.2 0.5 1 2 5 10

Favours intervention Favours control

Fig. 4 Forest plot of meta-analysis comparing psychological interventions with and without antidepressants versus antidepressants alone on
hazard ratios.
MBCT/-ADM, mindfulness-based cognitive therapy with tapering of antidepressant medication; ADM, antidepressant medication; PCT/-ADM, preventive cognitive therapy with
tapering of antidepressant medication; C-CT, continuation cognitive therapy; PCT + ADM, preventive cognitive therapy with antidepressant medication; MBCT + ADM, mindfulness-
based cognitive therapy with antidepressant medication.

to antidepressants after remission significantly reduced the risk of further RCT studies.9 Besides, before suggesting or recommending
relapse compared with taking antidepressants alone. psychological interventions, the availability of evidence-based psy-
Although previous meta-analyses found that patients have a chological interventions for relapse prevention (i.e. CBT, MBCT,
decreased risk of relapse when they receive psychological interven- PCT) should be considered, as they may not be available or rou-
tion without antidepressants compared with clinical management tinely implemented.35
or maintenance antidepressant medication,11,12 we found no signifi-
cant difference between either condition. A potential explanation
might be different inclusion and exclusion criteria, as we only Strengths and limitations
included studies where outcomes were assessed using diagnostic This review has several strengths. It is the first review assessing this
interview by an independent assessor and we also included more comparison specifically among high-quality RCTs, including
recent publications.7,34 In line with previous literature, we found studies from a range of settings where most patients had two or
that the combination of psychological intervention with antidepres- more previous episodes, improving the translatability of these
sants significantly reduces the risk of relapse compared with antide- results to a high-risk population who are currently mostly offered
pressants alone.11,12 antidepressant maintenance.
This review also has some limitations. In terms of generalisabil-
ity, included studies were only from high-income countries, meaning
Implications for clinical practice that there is still more research needed in the translation of these
These findings are highly relevant to current clinical practice as they results to low- and middle-income countries. In our meta-analyses,
differ from current clinical guidelines for depression. For patients we did not include adverse outcomes due to tapering or the approach
with two or more previous depressive episodes, clinical guidelines taken to tapering (i.e. duration, dose) nor were these commonly
currently recommend long-term antidepressant use after remis- reported; we recommend that future studies monitor these and
sion4,5 and no routine offering of psychological intervention (PCT describe the tapering procedure in detail, so as to inform tapering
or MBCT) for patients who are tapering off their antidepressant practices when delivering a preventive psychological intervention.
use. We found no evidence to suggest that there is a difference Multiple studies had various follow-up points; we selected the
between patients receiving a psychological intervention while taper- longest time of follow-up for each study and assumed proportional
ing off antidepressants versus antidepressant continuation alone. hazards, and then we conducted a sensitivity analysis to assess
Thus, the combination of tapering and psychological interventions whether the time to follow-up might be associated with the risk of
might be a viable alternative to long-term antidepressant use for relapse found.19 Although this approach may increase inconsistency
patients with multiple previous episodes who wish to taper. On across studies and heterogeneity,19 we did not have further data
the other hand, should patients deem continuation of antidepres- available to study this in more depth by, for example, testing RR
sants acceptable, adding a psychological intervention significantly or HR at different follow-up points.
reduces the risk of relapse compared with antidepressants alone. Finally, the number of studies in total is still small, which limits
Given that our results suggest that short-term preventive psy- the subgroup analyses we were able to conduct and power to detect a
chological interventions, in particular MBCT or PCT, might be an significance between subgroups.
alternative to long-term antidepressants, we wish to add a note of Overall, uncertainties remain that require further research. In
caution. In general, tapering of antidepressants requires an indivi- particular, it is important to note previous treatment combinations
dualised approach, and symptoms need to be monitored closely.9 during the acute phase and what effect these might have had on out-
Novel ways of tapering, i.e. over a longer time and with more comes in remission. For instance, a recent trial by deRubeis and col-
gradual reduction, may help mitigate withdrawal effects but requires leagues36 showed that the combination of antidepressant

6
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
 Psychological interventions in the prevention of depressive relapse

medication and CBT may interfere with any prophylactic effect

Data availability
CBT may have during the relapse prevention stage. Also, Barlow
et al (2000)37 studied CBT combined with antidepressants or The data that support the findings of this study are available from the corresponding author on
request.
alone in the treatment of panic disorder and observed that combin-
ation therapy resulted in little benefit over monotherapy (antide-
pressants or CBT). Although these findings are to be confirmed Acknowledgements
further, it is also important to note that they do not hold for starting
The authors thank Alicia Segovia, MSc, and Emma Boers, BSc, for support with title and abstract
psychotherapy after remission to prevent relapse and recurrence. screening, and Catherine Moore, MSc, for conducting the second-rater extractions on the
This meta-analysis suggests that adding a brief psychological inter- included studies.

vention to antidepressant continuation is more effective at reducing

the risk of depressive relapse compared to continuing antidepres-
Funding
sants alone, suggesting greater focus should be placed on this
approach. The authors thank the Amsterdam Public Health research institute for providing a small grant
which has funded software and travel costs.
Further research could also assess which specific intervention
types might be more effective (MBCT, PCT, IPT) than others,
depending on patient profiles. Finally, it would be of interest Author contributions
whether there are particular patient profiles associated with a
J.J.F.B. co-formulated the research question, co-designed the study, led on extractions, data
more or less favourable response to tapering or a combination of analysis and drafting the article prior to submission. M.E.B. co-formulated the research ques-
a psychological intervention with antidepressant medication. tion, co-designed the study, co-conducted study selections and extractions, provided analysis
advice and critically reviewed the article prior to submission. M.H. supported with the design
More research in the form of individual participant data and and extractions, analysis of the manuscript and preparing the risk of bias tables and critically
network meta-analysis could assess what works for whom and reviewed the article prior to submission. M.S. supported with the design, and analysis of the
study as well as critically reviewed the article prior to submission. D.D.E. supported with the
allow for even more targeted recommendations. design of the study and critically reviewed the article prior to submission. P.C. supported
In terms of future research, it would also be beneficial to assess with the design and analysis of the study and critically reviewed the article prior to submission.
C.L.H.B. co-formulated the research question, co-designed the study and extractions,
how these results can be best translated into clinical practice and supported with analysis advice and critically reviewed the article prior to submission.
clinical guidelines. Current guidelines for depression in the UK
and the USA do not recommend a psychological intervention
being delivered while tapering but recommend it as an add-on for Declaration of interest
patients at high risk of relapse.4,5 Our results suggest that the
C.L.H.B. is co-developer of the Dutch multidisciplinary clinical guideline for anxiety and depres-
addition of a psychological intervention when tapering would sion, for which she receives no remuneration; is a member of the scientific advisory board of
improve patient outcomes, and it would generally improve out- the National Health Care Institute, the Netherlands, for which she receives an honorarium,
although this role has no direct relation to this study; and has developed a preventive cognitive
comes for patients who are on long-term antidepressant medica- therapy based on the cognitive model of A.T. Beck.
tion. Developing our findings into recommendations in clinical ICMJE forms are in the supplementary material, available online at https://doi.org/10.1192/
bjp.2020.198.
guidelines, with substantial input from patients, families and
clinicians, would be a recommended next step.
In this meta-analysis, we found no evidence to suggest that psy- References
chological interventions added to tapering antidepressants increase
the risk of relapse versus antidepressants alone for patients with
1 Eaton WW, Shao H, Nestadt G, Lee BH, Bienvenu OJ, Zandi P. Population-based
multiple previous episodes of depression. Adding a psychological study of first onset and chronicity in major depressive disorder. Arch Gen
intervention to maintenance antidepressants significantly reduces Psychiatry 2008; 65: 513–20.
the risk of relapse, making this an option that one should consider 2 Moffitt T, Caspi A, Taylor A, Kokaua J, Milne BJ, Polanczyk G, et al. How common
adding routinely in clinical practice where possible. Healthcare are common mental disorders? Evidence that lifetime prevalence rates are
doubled by prospective versus retrospective ascertainment. Psychol Med
providers and clinical guideline developers can explore how psycho-
2010; 6: 899–909.
logical interventions can be routinely recommended and implemen-
3 Solomon DA, Keller MB, Leon AC, Mueller TI, Lavori PW, Shea MT, et al. Multiple
ted alongside (the tapering of) antidepressants to improve patient recurrences of major depressive disorder. Am J Psychiatry 2000; 157: 229–33.
choice and reduce the risk of depressive relapse. 4 American Psychiatric Association. Practice Guideline for the Treatment of
Patients With Major Depressive Disorder (3rd edn). APA, 2010.
Josefien Johanna Froukje Breedvelt , MSc, Amsterdam University Medical 5 National Institute for Health and Care Excellence. Depression in Adults:
Centers, Department of Psychiatry, University of Amsterdam, The Netherlands; and the Recognition and Management (Clinical Guideline CG90). NICE, 2009 (https://
Mental Health Foundation, London, UK; Maria Elisabeth Brouwer , PhD, Amsterdam www.nice.org.uk/guidance/cg90).
University Medical Centers, Department of Psychiatry, University of Amsterdam, The
Netherlands; Mathias Harrer, MSc, Department of Clinical Psychology and 6 Moore M, Yuen H, Dunn N, Mullee MA, Maskell J, Kendrick T. Explaining the rise
Psychotherapy, Friedrich-Alexander-University Erlangen-Nuremberg, Erlangen, in antidepressant prescribing: a descriptive study using the general practice
Germany; Maria Semkovska, PhD, Department of Psychology, University of Southern research database. BMJ 2009; 339: 956.
Denmark, Odense, Denmark; David Daniel Ebert, PhD, Department of Clinical, Neuro-
7 Bockting CLH, Klein NS, Elgersma HJ, van Rijsbergen GD, Slofstra C, Ormel J,
and Developmental Psychology, Vrije Universiteit Amsterdam, The Netherlands;
et al. Effectiveness of preventive cognitive therapy while tapering antide-
Pim Cuijpers , PhD, Department of Clinical, Neuro- and Developmental Psychology,
Vrije Universiteit Amsterdam, The Netherlands; Claudi Louisa Hermina Bockting, PhD, pressants versus maintenance antidepressant treatment versus their com-
Amsterdam University Medical Centers, Department of Psychiatry, University of bination in prevention of depressive relapse or recurrence (DRD study): a
Amsterdam; and Institute for Advanced Study, Amsterdam, The Netherlands three-group, multicentre, randomised control. Lancet Psychiatry 2018; 5:
401–10.
Correspondence: Claudi Bockting. Email: c.l.h.bockting@amsterdamumc.nl
8 McHugh RK, Whitton SW, Peckham AD, Welge JA, Otto MW. Patient preference
First received 21 Apr 2020, final revision 31 Jul 2020, accepted 22 Sep 2020 for psychological vs pharmacologic treatment of psychiatric disorders: a meta-
analytic review. J Clin Psychiatry 2013; 74: 595–602.
9 Horowitz MA, Taylor D. Tapering of SSRI treatment to mitigate withdrawal
symptoms. Lancet Psychiatry 2019; 6: 538–46.
10 Guidi J, Fava GA, Fava M, Papakostas GI. Efficacy of the sequential integration
of psychotherapy and pharmacotherapy in major depressive disorder: A pre-
liminary meta-analysis. Psychol Med 2011; 41: 321–31.
Supplementary material
11 Guidi J, Tomba E, Fava GA. The sequential integration of pharmacotherapy and
Supplementary material is available online at https://doi.org/10.1192/bjp.2020.198. psychotherapy in the treatment of major depressive disorder: A meta-analysis

7
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.
Breedvelt et al

of the sequential model and a critical review of the literature. Am J Psychiatry relapse and residual depressive symptoms in continuation treatment of major
2016; 173: 128–37. depressive disorder. J Clin Psychopharmacol 2002; 22: 474–80.
12 Biesheuvel-Leliefeld KEM, Kok GD, Bockting CLH, Cuijpers P, Hollon SD, van 30 Jarrett RB, Minhajuddin A, Gershenfeld H, Friedman ES, Thase ME. Preventing
Marwijk HWJ, et al. Effectiveness of psychological interventions in preventing depressive relapse and recurrence in higher-risk cognitive therapy respon-
recurrence of depressive disorder: meta-analysis and meta-regression. ders: a randomized trial of continuation phase cognitive therapy, fluoxetine, or
J Affect Disord 2015; 174: 400–10. matched pill placebo. JAMA Psychiatry 2013; 70: 1152–60.
13 Moher D, Shamseer L, Clarke M, Ghersi D, Liberatî A, Petticrew M, et al. 31 Brakemeier E-L, Merkl A, Wilbertz G, Quante A, Regen F, Buhrsch N, et al.
Preferred reporting items for systematic review and meta-analysis protocols Cognitive-behavioral therapy as continuation treatment to sustain response
(PRISMA-P) 2015 statement. Syst Rev 2015; 4(1): 1. after electroconvulsive therapy in depression: a randomized controlled trial.
14 Covidence. Covidence. Veritas Health Innovation, 2016 (https://www. Biol Psychiatry 2014; 76: 194–202.
covidence.org/). 32 Paykel ES, Scott J, Cornwall PL, Abbott R, Crane C, Pope M, et al. Duration of
15 Bockting HS, Jarrett RB, Kuyken W, Dobson K. A lifetime approach to major relapse prevention after cognitive therapy in residual depression: follow-up of
depressive disorder: the contributions of psychological interventions in pre- controlled trial. Psychol Med 2005; 35: 59–68.
venting relapse and recurrence. Clin Psychol Rev 2015; 41: 16–26. 33 Paykel ES, Scott J, Teasdale JD, Johnson AL, Garland A, Moore R, et al.
16 Frank E, Prien RF, Jarret RB, Keller MB, Kupfer DJ, Lavori PW, et al. Prevention of relapse in residual depression by cognitive therapy: a controlled
Conceptualization and rationale for consensus definitions of terms in major trial. Arch Gen Psychiatry 1999; 56: 829–35.
depressive disorder: remission, recovery, relapse, and recurrence. Arch Gen 34 Huijbers MJ, Spinhoven P, Spijker J, Ruhe HG, van Schaik DJF, van Oppen P,
Psychiatry 1991; 48: 851–5. et al. Adding mindfulness-based cognitive therapy to maintenance anti-
17 American Psychiatric Association. Diagnostic and Statistical Manual of Mental depressant medication for prevention of relapse/recurrence in major
Disorders (4th edn text revision) (DSM-IV-TR): 373–4. APA, 2000. depressive disorder: Randomised controlled trial. J Affect Disord 2015; 187:
54–61.
18 Furlan AD, Pennick V, Bombardier C, van Tulder M. 2009 updated method
guidelines for systematic reviews in the Cochrane Back Review Group. Spine 35 Crane RS, Kuyken W. The implementation of mindfulness-based cognitive
2009; 34: 1929–41. therapy: learning from the UK health service experience. Mindfulness (NY)
2013; 4: 246–54.
19 Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al. (eds).
Cochrane Handbook for Systematic Reviews of Interventions version 6.0 36 DeRubeis RJ, Zajecka J, Shelton RC, Amsterdam JD, Fawcett J, Xu C, et al.
(updated July 2019). Cochrane Collaboration, 2019. Prevention of recurrence after recovery from a major depressive episode with
antidepressant medication alone or in combination with cognitive behavioral
20 Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive Meta-Analysis therapy: phase 2 of a 2-phase randomized clinical trial. JAMA Psychiatry 2019;
Version 3.0. Biostat, 2013. 77: 237–45.
21 DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 37 Barlow DH, Gorman JM, Shear MK, Woods SW. Cognitive-behavioral therapy,
7: 177–88. imipramine, or their combination for panic disorder: a randomized controlled
22 Frank E, Kupfer DJ, Perel JM, Cornes C, Jarrett DB, Mallinger AG, et al. Three- trial. JAMA 2000; 283: 2529–36.
year outcomes for maintenance therapies in recurrent depression. Arch Gen 38 Kuyken W, Byford S, Taylor RS, Watkins E, Holden E, White K, et al. Mindfulness-
Psychiatry 1990; 47: 1093–9. based cognitive therapy to prevent relapse in recurrent depression. J Consult
23 Altman DG, Andersen PK. Calculating the number needed to treat for trials Clin Psychol 2008; 76: 966–78.
where the outcome is time to an event. 1999; 319: 1492–5. 39 Kuyken W, Hayes R, Barrett B, Byng R, Dalgleish T, Kessler D, et al. Effectiveness
24 Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in and cost-effectiveness of mindfulness-based cognitive therapy compared with
meta-analyses. BMJ 2003; 327: 557–60. maintenance antidepressant treatment in the prevention of depressive
relapse or recurrence (PREVENT): A randomised controlled trial. Lancet 2015;
25 Ioannidis JPA, Patsopoulos NA, Evangelou E. Uncertainty in heterogeneity
386: 63–73.
estimates in meta-analyses. BMJ 2007; 335: 914–6.
40 Segal ZV, Bieling P, Young T, MacQueen G, Cooke R, Martin L, et al.
26 Orsini N, Bottai M, Higgins J, Buchan I. HETEROGI: Stata Module to Quantify
Antidepressant monotherapy vs sequential pharmacotherapy and mindful-
Heterogeneity in a Meta-Analysis (Stata Software Components S449201).
ness-based cognitive therapy, or placebo, for relapse prophylaxis in recurrent
Boston College Department of Economics, 2006.
depression. Arch Gen Psychiatry 2010; 67: 1256–64.
27 Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a
41 Petersen TJ, Pava JA, Buchin J, Matthews JD, Papakostas GI, Nierenberg AA,
simple, graphical test measures of funnel plot asymmetry. BMJ 1997; 315:
et al. The role of cognitive-behavioral therapy and fluoxetine in prevention of
629–34.
recurrence of major depressive disorder. Cognit Ther Res 2007; 34: 13–23.
28 Duval S, Tweedie R, Taylor S, Tweedie R. Trim and fill: a simple funnel plot
based method of testing and adjusting for publication bias in meta-analysis.
Biometrics 2000; 56: 455–63.
29 Perlis RH, Nierenberg AA, Alpert JE, Pava J, Matthews JD, Buchin J, et al. The
effects of adding cognitive therapy to fluoxetine dose increase on risk of

8
Downloaded from https://www.cambridge.org/core. 21 Nov 2020 at 03:28:43, subject to the Cambridge Core terms of use.