Psychological Medicine Cognitive behaviour therapy for depression

cambridge.org/psm
in primary care: systematic review and
meta-analysis
Fredrik Santoft1, *, Erland Axelsson1*, Lars-Göran Öst1,2, Maria Hedman-Lagerlöf1,
Review Article
Jens Fust3 and Erik Hedman-Lagerlöf1,4
*Joint first authorship.
1
Division of Psychology, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden;
Cite this article: Santoft F, Axelsson E, 2
Öst L-G, Hedman-Lagerlöf M, Fust J, Hedman- Department of Psychology, Stockholm University, Stockholm, Sweden; 3Neuro, Department of Clinical
Lagerlöf E (2019). Cognitive behaviour therapy Neuroscience, Karolinska Institutet, Stockholm, Sweden and 4Osher Center for Integrative Medicine, Department
for depression in primary care: systematic of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
review and meta-analysis. Psychological
Medicine 49, 1266–1274. https://doi.org/
Abstract
10.1017/S0033291718004208
Depression is common in primary care, and most patients prefer psychological treatment over
Received: 27 August 2018 pharmacotherapy. Cognitive behaviour therapy (CBT) is an effective treatment, but there are
Revised: 12 December 2018
gaps in current knowledge about CBT in the primary care context, especially with regard to
Accepted: 22 December 2018
First published online: 28 January 2019 long-term effects and the efficacy of specific delivery formats. This is an obstacle to the inte-
gration of primary care and specialist psychiatry. We conducted a systematic review and meta-
Key words: analysis of randomised controlled trials of CBT for primary care patients with depression to
Cognitive behaviour therapy; depression;
investigate the effect of CBT for patients with depression in primary care. A total of 34 studies,
major depression; meta-analysis; primary care;
psychological treatment; psychotherapy; with 2543 patients in CBT and 2815 patients in control conditions, were included. CBT was
systematic review more effective than the control conditions [g = 0.22 (95% confidence interval (CI) 0.15–0.30)],
and the effect was sustained at follow-up [g = 0.17 (95% CI 0.10–0.24)]. CBT also led to a
Author for correspondence: higher response rate [odds ratio (OR) = 2.47 (95% CI 1.60–3.80)] and remission rate [OR =
Erland Axelsson, E-mail: [email protected]
1.56 (95% CI 1.15–2.14)] than the control conditions. Heterogeneity was moderate. The con-
trolled effect of CBT was significant regardless of whether patients met diagnostic criteria for
depression, scored above a validated cut-off for depression, or merely had depressive symp-
toms. CBT also had a controlled effect regardless of whether the treatment was delivered as
individual therapy, group therapy or therapist-guided self-help. We conclude that CBT
appears to be effective for patients with depression in primary care, and recommend that
patients with mild to moderate depression be offered CBT in primary care.

Introduction
Unipolar depressive disorders are the leading cause of disease burden in middle- to high-
income countries, and have been predicted to become the worldwide leading cause of disease
burden by 2030 (World Health Organization, 2008). Depression is associated not only with
suffering, disability and impaired health (Moussavi et al., 2007; Ormel et al., 2008), but also
considerable societal costs (Cuijpers et al., 2007). Nevertheless, most people who suffer
from depression never receive adequate treatment (e.g. Thornicroft et al., 2017). The majority
of patients with mood disorders are found in primary care (e.g. ESEMeD⁄MHEDEA 2000
investigators, 2004). In the treatment of depression, the primary care context offers both
advantages – for example, in terms of a low threshold for health care seeking, and the tradition
of a lifetime perspective on health – and challenges, for example, with regard to the gap in
mental health competency, and the integration with specialist care services. Most patients
with depression prefer psychological treatment over pharmacotherapy (McHugh et al.,
2013), and the psychological treatment that has been most studied in the treatment of depres-
sion is cognitive behaviour therapy (CBT) (National Collaborating Centre for Mental Health,
2010). Individual one-to-one CBT constitutes the gold standard format, but the treatment can
also be delivered as group therapy or guided self-help, for example, via the Internet (Hedman
et al., 2012) or as bibliotherapy (Cuijpers, 1997). Meta-analyses have provided preliminary evi-
dence that CBT for depression may be effective in the primary care context, but there is a rapid
development of this research field and several key questions for implementation in routine care
remain unanswered (Cuijpers et al., 2009; Linde et al., 2015; Twomey et al., 2015). For
example, long-term effects have not been investigated, the relative efficacy of delivery formats
© Cambridge University Press 2019 is largely unknown (Linde et al., 2015), potential moderators of treatment effect like the num-
ber of sessions and therapist qualifications have not been investigated for CBT specifically, and
it is unclear whether CBT is suitable for primary care patients who have depressive symptoms
but do not meet full diagnostic criteria for depression. Increased knowledge in these areas is
likely to facilitate the implementation of treatment in primary care, and the integration with
Psychological Medicine 1267

psychiatric care for this large patient group. We therefore con- from working in primary care is that patients with elevated
ducted a systematic review and meta-analysis of randomised con- depressive symptoms who do not meet full criteria for a diag-
trolled trials of CBT for adult primary care patients with nosis of depression make up a substantial portion of those
depression. offered a treatment focusing on reducing depression symp-
toms. This inclusion criterion was therefore deliberately
vague so as to capture studies of high ecological validity,
Method
and enable moderator analysis to assess if the manner in
Search strategy which the eligibility criterion was formulated (diagnosis v.
cut-off v. depressive symptoms) was predictive of outcome.
A systematic review and meta-analysis was conducted in accord-
No restrictions were made with regard to comorbidity, but
ance with PRISMA guidelines (Moher et al., 2009). We searched
we did not include studies of bipolar disorders, seasonal
PubMed and PsycINFO for randomised controlled trials where
affective disorders or pre-/postnatal depression.
CBT was compared with a control condition in the treatment of
(c) Only studies where the entire sample consisted of adult
adults with depression in primary care. Our strategy was to con-
patients (⩾18 years) were included.
duct a relatively broad search and combine terms for adult
(d) Studies were required to have a primary care focus, meaning
patients, depression, CBT and primary care (see the online
that either (I) more than half of the sample was recruited
Supplementary material for complete search terms). We began
from primary care or (II) CBT was delivered in primary
work on study selection in October 2014, and last searched data-
care. This was arguably the criterion most difficult to assess,
bases on 6 November 2018. In order to identify both early works
not least due to notable international differences in organisa-
and recent articles not yet categorised in the databases, no filters
tion of the health care system, and also because articles were
or restrictions (e.g. with regard to study time of publication) were
not explicit about the setting where treatments were con-
applied. We also read the reference lists of all included studies,
ducted. Whenever necessary, authors were contacted and
and considered studies found in the process of data extraction
asked if the majority of their sample had been recruited
and in previous meta-analyses. We did not search for unpublished
through primary care, and/or whether they would character-
studies. Because all data were collected at the study level, we did
ise the setting where patients received treatment as primary
not deem it necessary to obtain ethics approval for this review.
care. In this study, we regarded primary care as being non-
specialised, i.e. concerned with most common disease states,
Selection of studies relatively accessible and commonly serving the role of a first
step or ‘gatekeeper’ in relation to secondary care.
The eligibility of all unique search hits was assessed by one of the
(e) All studies had to be randomised controlled trials where CBT
authors (FS or EA) in three stages. First, publications were excluded
was compared with a control condition. The control condition
based on titles, then on abstracts and finally on full texts. Reason for
could be treatment as usual (TAU), antidepressants, another
exclusion was defined as the first exclusion criterion identified. In
psychological treatment, waiting-list or a placebo (psycho-
order to validate the selection process, all studies that reached the
logical or pharmacological). We excluded studies where the
stage of full-text evaluation were also read by a second assessor
patient was not the unit of randomisation, and studies which
(EA, MHL or JF). Whenever there was disagreement on whether
solely compared different forms of CBT against each other.
to include a study, a third author (EHL or FS) was consulted and
(f) Studies had to be published in an English-language journal
a decision made in consensus after discussion. If vital information
with peer-review.
to assess study eligibility was missing, corresponding authors were
(g) Studies were only included if the effect of CBT on its
contacted and asked to provide that information. This occurred in
own could be estimated. In other words, if the experimental
33 cases, of which 24 replied.
condition involved substantial structured interventions in
addition to CBT (e.g. the addition of antidepressant medica-
Eligibility criteria tion), the study was excluded. If the CBT condition included
(a) We required studies to have investigated the effect of CBT for access to TAU, the study was included because this did not
depression. CBT was defined as either cognitive therapy (i.e. constitute a structured parallel treatment.
where the treatment is designed to work through cognitive
restructuring), behaviour therapy (i.e. behavioural activation
Data extraction
focusing on increasing positive reinforcement), or a combin-
ation of the two. Treatments were required to have as their Most of the data extraction was done in parallel by two independ-
principal aim to reduce symptoms of depression, and to ent assessors (FS and EA). For most studies, the primary outcome
last more than 1 week. Third-wave approaches to CBT such was included in the meta-analysis. If a primary outcome was not
as acceptance and commitment therapy, dialectical behaviour specified, or if the primary outcome was not a measure of depres-
therapy and mindfulness-based cognitive therapy were not sion, the first measure of depression reported in the article was
included. CBT could be delivered in any format (e.g. as chosen. For studies that used both self-reported and independ-
face-to-face therapy, online treatment or bibliotherapy), as ently assessed measures of depression, we followed the procedure
long as there was support from a clinician. The amount of of previous meta-analyses (Cuijpers et al., 2013) and based our
support could, however, be minimal, for example, consisting estimates on the mean effect size from these two measures. We
of one phone call only. also assembled data on responder and remission rates, where
(b) We required all patients to either (I) meet diagnostic criteria the former was operationalised as the proportion of patients
for a unipolar depressive disorder such as DSM-IV major who achieved a clinically significant symptom reduction and the
depressive disorder, (II) score above a recognised cut-off for latter was operationalised as the proportion of patients who did
depression or (III) have depressive symptoms. Our experience not meet criteria for depression or who scored below an adequate
1268 Fredrik Santoft et al.

cut-off score for depression. We did not include estimates which of cognitive interventions, session length). Our ambition was to
conflated symptom reduction (i.e. response) with endpoint score only analyse putative moderators where at least 75% of the studies
(i.e. remission). Remission rates were based only on those studies reported data. Two exceptions from this rule were however made
where it was clear that no patient met the criterion for remission because we found these tests important. One was a comparison of
at baseline. In order to enable moderator analyses, data were also studies with high and low ratings on the allocation concealment
collected about study design and patient characteristics (e.g. coun- risk-of-bias criterion, and the other was a comparison based on
try, mean age, type of control), as well as the characteristics of the setting where CBT was delivered (primary care v. specialist
CBT (e.g. delivery format, number of sessions, therapist qualifica- setting with patients from primary care). Because p values for
tions). CBT protocols were classified as being in individual format the Q statistic were often of limited value due to low power,
if the majority of the treatment content was delivered through and also in order to quantify the pooled effects in subgroups of
extensive one-to-one contact with a therapist (face-to-face, via studies (e.g. differentiate between CBT formats), we also con-
telephone or online), in group format if the majority of the treat- ducted a series of secondary subgroup analyses. These corre-
ment content was delivered through sessions with more than one sponded to the levels of putative moderators, so that we, for
patient, and as guided self-help if most of the content was example, could present separate effect sizes for individual CBT,
intended to be conveyed by the means of a text or didactic mater- group CBT and guided self-help CBT. Key estimates based only
ial, with little (typically <3 h) therapist support. on the subsample of studies where CBT was not delivered outside
of primary care are provided as online Supplementary material.
Heterogeneity was estimated based on the Q- and I 2 statistics,
Risk of bias assessment
where the latter stands for the proportion of the total variance
We assessed study risk of bias based on the Cochrane collabora- between studies that can be attributed to true study differences
tion’s tool (Higgins et al., 2011), and rated the following dimen- in effects, rather than random (sampling) error. I 2 is measured
sions: (I) random sequence generation, (II) allocation concealment, in percentage, where values of 25, 50 and 75% represent low,
(III) blinding of outcome assessment, (IV) incomplete outcome moderate and high heterogeneity, respectively (Higgins et al.,
data and (V) selective reporting. Studies were rated in terms of 2003). Publication bias was assessed based on visual inspection
high risk of bias, low risk of bias or insufficient information of funnel plots in combination with Egger’s intercept test
(when a criterion could not be assessed). Criterion III (‘blinding (Egger et al., 1997), and the Duval and Tweedie trim and fill pro-
of outcome assessment’) was rated as ‘not applicable’ for studies cedure with the R0 estimator (Duval and Tweedie, 2000).
where all depression measures were self-reported. We did not
rate criterion ‘blinding of participants and personnel’ because it
Results
is not possible for those who administer psychological treatments
to be blinded with regard to the treatment that they are delivering. After reading 372 articles in full text, assessors agreed to include
Ratings for criteria III, IV and V focused on those continuous 29 studies (κ = 0.81), and disagreed on 12 studies, of which three
outcomes which were used for the meta-analysis. were included after discussion with a third author. These three
studies were discussed due to the most common reason for dis-
agreement, which was different views on whether studies had a
Statistical analysis
primary care focus (6/12). The second most common topic of dis-
Analyses were done in R 3.4.4 (R Core Team, 2016) with metafor cussion was whether CBT protocols targeted depression specific-
2.0-0 (Viechtbauer, 2010). Controlled effect sizes on continuous ally (3/12). The remaining three differences concerned whether
measures were quantified as Hedges’ g; g > 0 favouring CBT the patients had elevated symptoms of depression, whether the
(Hedges, 1981). In cases where studies did not report means sample was mixed with other disorders and whether the treat-
and standard deviations to allow for conventional computation ment was to be regarded as CBT. Two additional studies were
of effect sizes, approximations were used (Lipsey and Wilson, found through reference lists and during data extraction, which
2001). Absolute values for g of 0.2 are usually regarded as resulted in 34 randomised controlled trials included in
small, 0.5 as moderate and 0.8 as large (Cohen, 1988). meta-analysis (Fig. 1) (Teasdale et al., 1984; Ross and Scott,
Controlled effect sizes for responder and remission rates were 1985; Scott and Freeman, 1992; Scott et al., 1997; Ward et al.,
instead reported as odds ratios (OR), i.e. the ratio of the odds of a 2000; Watson et al., 2003; Willemse et al., 2004; Dalgard, 2006;
beneficial outcome in CBT and the corresponding odds in the Smit et al., 2006; González González et al., 2007; Spek et al.,
control condition, where OR>1 is in favour of CBT. We also 2007; Laidlaw et al., 2008; Lovell et al., 2008; Wiles et al., 2008;
reported the number needed to treat (NNT), which is the inverse Kessler et al., 2009; Serfaty et al., 2009; Hegerl et al., 2010;
of the absolute risk difference. The NNT stands for the average Naylor et al., 2010; Cramer et al., 2011; Dwight-Johnson et al.,
number of patients necessary to assign to CBT in order to achieve 2011; Ekers et al., 2011; Joling et al., 2011; Levin et al., 2011;
one beneficial case (i.e. one case of response or remission) that Casañas et al., 2012; Power and Freeman, 2012; Sørensen
would not have been achieved had the patients instead been Høifødt et al., 2013; Wiles et al., 2013; Williams et al., 2013;
assigned to the control group. Husain et al., 2014; Kivi et al., 2014; Gilbody et al., 2015;
We based all aggregation of effect sizes on random-effects Kanter et al., 2015; Chowdhary et al., 2016; Gilbody et al., 2017).
models fitted with the restricted maximum likelihood estimator
(linear regression for g and NNT, logistic regression for OR).
Study characteristics
For continuous outcome measures, we first did a primary com-
parison of CBT (any form) and control conditions (any form). From the 34 randomised controlled trials of CBT for depression
Second, we explored possible moderators of this between-group in primary care, we analysed post-treatment data from 5358
effect based on Q-tests and meta-regression (most pre-specified, patients; 2543 in CBT (35 conditions) and 2815 controls (45 con-
except: therapist adherence reported, expert-level therapists, use ditions). In terms of CBT formats, 17 conditions were individual
Psychological Medicine 1269

Fig. 1. Flowchart of study selection process. CBT, cognitive

behaviour therapy; RCTs, randomised controlled trials.

CBT with extensive support either face-to-face or remotely, seven Post-treatment effects on depressive symptoms
were group therapies, 10 were guided self-help CBT with little
Based on 46 randomised comparisons of CBT to control conditions
therapist support and one was a mixed individual and group ther-
in the treatment of depression in primary care, the pooled effect size
apy sample. As to therapists, 31% (11/35) of CBT conditions
was g = 0.22 [95% confidence interval (CI) 0.15–0.30] in favour of
employed psychologists or psychotherapists, 14% (5/35) unquali-
CBT. Heterogeneity was significant and in the low-to-moderate
fied personnel, 11% (4/35) nurses and 43% (15/35) mixed profes-
range (I 2 = 40%; Q45 = 78, p = 0.002). The pooled effect size of
sions or other/unspecified. The mean number of sessions was 9.8
those five studies with lowest risk of bias (see above) was g =
(S.D. = 3.8), and the mean session length was 58.1 min (S.D = 21.6).
0.19 (95% CI 0.06–0.32). Effect sizes based on the subsample of
The most common control condition was TAU, which consisted
comparisons (40/46) where CBT was not delivered outside of pri-
of a wide range of treatments usually offered in primary care
mary care were also similar (online Supplementary material).
such as visits with a general practitioner, antidepressant medica-
Figure 3 displays a forest plot of study effect sizes with CI.
tion, counselling or referral for psychological treatment. As to
country of origin, 50% (17/34) of the trials were based in the
UK, 12% (4/34) in the Netherlands, 12% (4/34) in the USA, 6%
(2/34) in Norway, 6% (2/34) in Spain and the remaining 15%
(5/34) in other countries. Additional study and condition charac- Moderators and subgroups
teristics are presented in the online Supplementary material. Moderator and subgroup analyses are presented in Table 1 and
online Supplementary Tables DS4 and DS5. As to study design,
CBT had a significant controlled effect regardless of whether
Study risk of bias
patients were included based on a diagnosis of depression, a cut-
The risk of bias varied considerably between studies (Fig. 2, online off score or depressive symptoms only (p = 0.347). The choice of
Supplementary Table DS3). While random sequence generation control condition was associated with outcome (p = 0.041). For
and allocation concealment were adequate in the majority of example, the three studies which compared CBT against a waiting
cases, the most common reason for risk of bias was incomplete list reported a moderate pooled effect size (g = 0.48), whereas the
outcome data. Three studies were given a low risk of bias rating pooled difference between CBT and other psychological treat-
on all applicable criteria (Wiles et al., 2013; Husain et al., 2014; ments was close to zero (g = −0.02). There was no significant dif-
Gilbody et al., 2017). Two additional studies (Smit et al., 2006; ference in effect between studies from Europe and studies from
Joling et al., 2011) fell just short of this mark because no pre- other parts of the world (p = 0.896). The following variables
registered study protocol could be found (selective reporting also did not moderate the controlled effect of CBT: mean baseline
unclear). depression severity, mean patient age, proportion female,
1270 Fredrik Santoft et al.

Fig. 2. Study risk of bias based on the Cochrane collaboration’s tool. Please note that these ratings apply to the outcome aggregated in the primary meta-analysis.

Fig. 3. Forest plot of all comparisons (k = 46) of CBT and control conditions. ADM, antidepressant medication; CBT, cognitive behaviour therapy; GSH, guided self-
help; TAU, treatment as usual.

publication year, weeks to the primary endpoint and the Cochrane associated with a larger effect than studies where CBT was
risk of bias criteria. delivered in a primary care setting (g = 0.43 v. 0.22; p = 0.009).
As to the characteristics of CBT, there was no significant The following variables did not moderate the effect of CBT:
difference in effect between delivery formats (ps = 0.234– number of sessions, session length, if an adherence check was
0.765). Individual CBT (g = 0.24), group CBT (g = 0.28) and reported, if expert-level therapists (i.e. psychologists/psy-
guided self-help CBT (g = 0.15) were all more effective than chotherapists or equivalent) delivered the treatment and if the
the control condition. Studies where CBT was conducted with treatment was based on both cognitive techniques and behavioural
primary care patients in a research or specialist setting were activation or the latter only.
Psychological Medicine 1271

Table 1. Moderator analyses of cognitive behaviour therapy and control groups in the treatment of depression and subclinical depression in primary care:
categorical variables

Putative moderator p Subgroup k g 95% CI I 2 (%)

Pooled total 46 0.22 0.15–0.30 40

Main inclusion criterion 0.347 Diagnosis of depression 21 0.29 0.16–0.41 40
Cut-off on depression scale 19 0.17 0.07–0.27 34
Depressive symptoms 6 0.21 0.03–0.39 36
Baseline depression severity 0.418 Mild 12 0.20 0.10–0.29 0
Moderate 27 0.20 0.10–0.29 41
Severe 5 0.38 0.05–0.70 67
Outcome 0.475 Self-rated 34 0.21 0.13–0.29 44
Clinician-rated or both 12 0.28 0.12–0.44 24
Control group 0.041 Treatment as usual 29 0.27 0.18–0.36 44
Psychological 7 −0.02 −0.26 to 0.23 49
Antidepressant 3 0.05 −0.21 to 0.32 0
Waiting-list 3 0.48 0.15–0.81 29
Other 4 0.15 −0.09 to 0.39 14
Study site 0.896 Europe 39 0.22 0.14–0.31 45
Not Europe 7 0.26 0.09–0.43 5
a
CBT delivery format 0.467 Individual 24 0.24 0.13–0.34 38
Group 11 0.28 0.16–0.40 0
Guided self-help 10 0.15 0.02–0.28 48
CBT delivery setting <0.001 Primary care 15 0.22 0.12–0.32 0
Unclear 25 0.16 0.07–0.25 29
Specialist or mixed 6 0.43 0.31–0.55 0
Therapist adherence check 0.423 Yes, reported and acceptable 19 0.26 0.15–0.38 27
No, not reported 27 0.20 0.11–0.30 46
Expert-level therapists 0.611 Yes 20 0.25 0.09–0.41 52
Mixed or unclear 13 0.21 0.13–0.30 4
No 13 0.23 0.09–0.37 54
Interventions 0.885 Cognitive and behavioural 31 0.23 0.15–0.31 25
Behavioural only 5 0.31 0.00–0.62 58
CBT, cognitive behaviour therapy.
a
One comparison excluded from this analysis due to mixed CBT format. The ‘individual’ category included one-to-one treatment via the Internet or telephone.

Follow-up effects on depressive symptoms 49% (95% CI 42–56) in CBT and 26% (95% CI 15–37) in the con-
trol groups, which corresponded to OR = 2.47 (95% CI 1.60–3.80)
Controlled follow-up effects of CBT were reported for 27 compar-
and NNT = 4.60 (95% CI 3.25–7.84). Heterogeneity of the OR was
isons from 21 studies. The pooled last follow-up assessment (M =
moderate (I 2 = 54%; Q8 = 18, p = 0.024). Remission rates were
10.2 months, S.D. = 9.4; Mdn = 8) effect size was g = 0.17 (95% CI
reported for 20 comparisons from 17 studies, and cut-off scores
0.10–0.24) in favour of CBT. Heterogeneity was low (I 2 = 18%;
clustered around θ = 55–66 on the latent depression metric pub-
Q26 = 30, p = 0.269). Approximately 70% (19/27) of these compar-
lished by Wahl et al. (2014). The aggregate remission rate was
isons were against a TAU control condition, and had a pooled
45% (95% CI 39–51) in CBT and 35% (95% CI 27–42) in the con-
controlled effect size of g = 0.19 (95% CI 0.10–0.28; I 2 = 30%;
trol conditions, which corresponded to OR = 1.56 (95% CI 1.15–
Q18 = 22, p = 0.216).
2.14) and NNT = 10.08 (95% CI 5.83–36.90). Again, heterogeneity
of the OR was moderate (I 2 = 65%; Q19 = 54, p < 0.001).
Responder and remission rates
Publication bias
Responder rates were reported for nine comparisons from nine
studies, and the most common criterion for response was a symp- As judged from visual inspection of the funnel plot there
tom reduction of at least 50%. The aggregate responder rate was appeared to be no asymmetry of effect sizes in relation to the
1272 Fredrik Santoft et al.

standard error (online Supplementary Fig. DS1). Egger’s test was than primary care were more likely to be highly qualified.
not significant (p = 0.323). Based on Duval and Tweedie’s proced- Non-expert therapists were employed by 60% (9/15) of CBT con-
ure, no study was missing on the left side of the graph (P = .500), ditions in primary care v. 0% (0/6) of CBT conditions delivered
and no imputation was indicated. elsewhere.

Discussion
Limitations
This systematic review and meta-analysis based on 34 randomised
controlled trials found that CBT has a significant though small The primary limitation of this meta-analysis was that only 9% (3/
controlled effect on depression for adult primary care patients 34) of the included studies had low risk of bias on all eligible bias
(g = 0.22), and that the effect is sustained at follow-up (g = criteria. Approximately half of the studies reported outcomes sus-
0.17). The body of studies was moderately heterogeneous, with ceptible to bias due to missing outcome data or inadequate hand-
most CBT conditions being individual therapy (24/46 = 52%), ling of missing data (e.g. large proportion of missing data, high
and most comparisons against TAU (29/46 = 63%). CBT was risk of data missing not at random or extensive imputation
more effective than the control also for patients who had depres- based on last-observation-carried-forward). It is conceivable that
sive symptoms without meeting full criteria for depression. All the high rate of missing data distorted the outcome.
CBT formats, i.e. individual therapy, group therapy and guided Another limitation is that the inclusion of multiple compari-
self-help, were significantly more effective than the control condi- sons from one randomised controlled trial may be argued to
tions. Based on the NNT statistic, it appears that it is typically lead to biased estimates and artificially reduced heterogeneity,
necessary to treat approximately five patients with CBT rather given that conditions are ‘counted twice’. Although there is
than the control condition to achieve one additional responder, some truth to this, we also regard it as important not to arbitrarily
and to treat approximately 10 patients with CBT instead of the disregard comparisons from trials which incorporated more than
control condition to achieve one additional case in remission. two conditions of interest.
The validity of our findings is indicated by our use of broad Because there was a significant heterogeneity in most analyses,
search criteria in combination with a reliable inclusion procedure it is probably important to differentiate between studies and con-
with independent assessors. This is to date, and to our knowledge, ditions in order to generalise our findings and make valid infer-
the largest meta-analysis of CBT for depression in primary care, ences about real-world situations. It has, for example, been
and the first meta-analysis to investigate follow-up effects in demonstrated that the components of TAU, which was the most
this context. common comparator, tend to vary considerably from study to
study (Watts et al., 2015).
Comparison with prior research We also wish to point out two caveats related to our reporting
of responder and remission rates. First, the OR presented here do
The remission rates of the present study were similar to those not approximate risk ratios because response and remission were
reported in studies of CBT for depression outside of primary common outcomes (Cummings, 2009). Use of the OR is never-
care (Cuijpers et al., 2014). Our findings are also in line with pre- theless commonplace, and facilitates comparisons with previous
vious meta-analyses which have demonstrated the efficacy of CBT meta-analyses (e.g. Linde et al., 2015). Second, use of the NNT
as compared with TAU for depression in the primary care context in meta-analyses rests on the strong assumption that the differ-
(Cuijpers et al., 2009; Linde et al., 2015; Twomey et al., 2015). ence between the response rate in the CBT condition and the
Controlled effect sizes were however smaller than in previous response rate in the control condition are approximately equal
meta-analyses, possibly because two previous meta-analyses of over studies. An advantage of the NNT is, however, that it is rela-
CBT in primary care had a higher proportion of comparisons tively easy to understand.
with average severe depressive symptoms at baseline (29%/56%
v. 11% in this sample) (Cuijpers et al., 2009; Linde et al., 2015).
Our finding that CBT has a similar effect on depression as
Clinical implications and recommendations for future research
other psychological interventions in primary care (g = −0.02) is
tentative given the small subsample (k = 7), though also consistent Despite notable uncertainty about exact effects, this meta-analysis
with a previous meta-analysis of CBT in both primary care and indicates that CBT for depression is effective in the primary care
other settings (Cuijpers et al., 2013). However, it is probably context. Considering the high prevalence of the disorder and high
important to differentiate between psychological treatments, strain on psychiatric care, we regard this as an important finding
which could not be done here due to the small number of com- because it suggests that it is feasible, as a first step, to offer patients
parisons against psychological interventions. with mild to moderate depression treatment in primary care. In
Based on the present study, most aspects of the treatment for- contrast, treatment for severe depression could be argued to fall
mat, such as treatment length, do not appear to be important for outside of the generalist tradition of primary health care, and
the controlled effect of CBT in primary care. It must be noted, has rarely been studied in this context.
however, that the present study had insufficient power to detect As for future research on CBT for depression in primary care,
small to moderate effects, or to explore the inter-relationships we advise investigators to adhere to common guidelines for the
between putative moderators based on multivariate analyses. conduct and publication of clinical trials (Schulz et al., 2010;
Another noteworthy finding was that if the treatment was Higgins et al., 2011). Of particular importance, we urge that miss-
delivered in primary care, this was associated with a smaller con- ing data be properly addressed, and that future publications be
trolled effect (g = 0.22) than if the patients were from primary care more explicit about the setting of treatment delivery. For the
but the treatment took place elsewhere (g = 0.43). A possible field to move forward, we also encourage more direct compari-
explanation for this is that clinicians working in other contexts sons between active structured treatments.
Psychological Medicine 1273

Conclusion Duval S and Tweedie R (2000) Trim and fill: a simple funnel-plot-based
method of testing and adjusting for publication bias in meta-analysis.
CBT is an effective treatment for adult depressed primary care Biometrics 56, 455–463.
patients, and effects are sustained over time. Effect sizes in com- Dwight-Johnson M, Aisenberg E, Golinelli D, Hong S, O’Brien M and
parison to TAU are typically small, but may be clinically import- Ludman E (2011) Telephone-based cognitive-behavioral therapy for
ant given the high prevalence of depression in the primary care Latino patients living in rural areas: a randomized pilot study. Psychiatric
setting, patient demand for psychological treatment and high Services 62, 936–942.
strain on specialist psychiatry. Egger M, Davey Smith G, Schneider M and Minder C (1997) Bias in
meta-analysis detected by a simple, graphical test. BMJ 315, 629–634.
Author ORCIDs. Fredrik Santoft 0000-0003-0673-6521, Erland Axelsson Ekers D, Richards D, McMillan D, Bland J and Gilbody S (2011)
0000-0003-2562-2925, Lars-Göran Öst 0000-0002-4351-2810, Maria Hedman- Behavioural activation delivered by the nonspecialist: phase II randomised
Lagerlöf 0000-0002-3581-099X, Jens Fust 0000-0002-4706-092X, Erik Hedman- controlled trial. British Journal of Psychiatry 198, 66–72.
Lagerlöf 0000-0002-7939-9848. ESEMeD⁄MHEDEA 2000 investigators (2004) Use of mental health services
in Europe: results from the European Study of the Epidemiology of Mental
Supplementary material. The supplementary material for this article can Disorders (ESEMeD) project. Acta Psychiatrica Scandinavica 109, 47–54.
be found at https://doi.org/10.1017/S0033291718004208 Gilbody S, Littlewood E, Hewitt C, Brierley G, Tharmanathan P, Araya R,
Barkham M, Bower P, Cooper C, Gask L, Kessler D, Lester H, Lovell K,
Author contributions. FS and EA contributed equally to this study. FS, EA, Parry G, Richards DA, Andersen P, Brabyn S, Knowles S, Shepherd C,
LGÖ and EHL designed the study. All authors took part in the data collection. Tallon D and White D (2015) Computerised cognitive behaviour therapy
EA conducted the statistical analysis. All authors contributed to the publica- (cCBT) as treatment for depression in primary care (REEACT trial): large
tion. EA had full access to all data, and takes responsibility for the integrity scale pragmatic randomised controlled trial. BMJ 351, h5627.
of the data and the accuracy of the presented analysis. Gilbody S, Lewis H, Adamson J, Atherton K, Bailey D, Birtwistle J,
Bosanquet K, Clare E, Delgadillo J, Ekers D, Foster D, Gabe R,
Financial support. This research was funded by Karolinska Institutet and
Gascoyne S, Haley L, Hamilton J, Hargate R, Hewitt C, Holmes J,
Stockholm County Council, none of which had any role in the design, execu-
Keding A, Lilley-Kelly A, Meer S, Mitchell N, Overend K, Pasterfield M,
tion or publication of the study.
Pervin J, Richards DA, Spilsbury K, Traviss-Turner G, Trepel D,
Conflict of interest. None. Woodhouse R, Ziegler F and McMillan D (2017) Effect of collaborative
care vs usual care on depressive symptoms in older adults with subthreshold
depression: the CASPER randomized clinical trial. JAMA 317, 728–737.
González González S, Fernández Rodríguez C, Pérez Rodríguez J and
Amigo I (2007) Secondary prevention of depression in primary care.
References Psychology in Spain 11, 24–32.
Casañas R, Catalán R, del Val JL, Real J, Valero S and Casas M (2012) Hedges LV (1981) Distribution theory for Glass’s estimator of effect size
Effectiveness of a psycho-educational group program for major depression and related estimators. Journal of Educational and Behavioral Statistics 6,
in primary care: a randomized controlled trial. BMC Psychiatry 12, 230. 107–128.
Chowdhary N, Anand A, Dimidjian S, Shinde S, Weobong B, Balaji M, Hedman E, Ljótsson B and Lindefors N (2012) Cognitive behavior therapy
Hollon SD, Rahman A, Wilson G, Verdeli H, Araya R, King M, via the Internet: a systematic review of applications, clinical efficacy and
Jordans MJ, Fairburn C, Kirkwood B and Patel V (2016) The healthy cost-effectiveness. Expert Review of Pharmacoeconomics & Outcomes
activity program lay counsellor delivered treatment for severe depression Research 12, 745–764.
in India: systematic development and randomised evaluation. British Hegerl U, Hautzinger M, Mergl R, Kohnen R, Schutze M, Scheunemann W,
Journal of Psychiatry 208, 381–388. Allgaier A-K, Coyne J and Henkel V (2010) Effects of pharmacotherapy
Cohen J (1988) Statistical Power Analysis for the Behavioral Sciences. Hillsdale, and psychotherapy in depressed primary-care patients: a randomized, con-
NJ: Lawrence Erlbaum Associates. trolled trial including a patients’ choice arm. International Journal of
Cramer H, Salisbury C, Conrad J, Eldred J and Araya R (2011) Group cog- Neuropsychopharmacology 13, 31–44.
nitive behavioural therapy for women with depression: pilot and feasibility Higgins JP, Thompson SG, Deeks JJ and Altman DG (2003) Measuring
study for a randomised controlled trial using mixed methods. BMC inconsistency in meta-analyses. BMJ 327, 557–560.
Psychiatry 11, 82. Higgins JP, Altman DG, Gotzsche PC, Juni P, Moher D, Oxman AD,
Cuijpers P (1997) Bibliotherapy in unipolar depression: a meta-analysis. Savovic J, Schulz KF, Weeks L, Sterne JA, Cochrane Bias Methods
Journal of Behavior Therapy and Experimental Psychiatry 28, 139–147. Group and Cochrane Statistical Methods Group (2011) The Cochrane
Cuijpers P, Smit F, Oostenbrink J, de Graaf R, Ten Have M and Beekman A Collaboration’s tool for assessing risk of bias in randomised trials. BMJ
(2007) Economic costs of minor depression: a population-based study. Acta 343, d5928.
Psychiatrica Scandinavica 115, 229–236. Husain N, Chaudhry N, Fatima B, Husain M, Amin R, Chaudhry IB,
Cuijpers P, van Straten A, van Schaik A and Andersson G (2009) Ur Rahman R, Tomenson B, Jafri F, Naeem F and Creed F (2014)
Psychological treatment of depression in primary care: a meta-analysis. Antidepressant and group psychosocial treatment for depression: a rater
British Journal of General Practice 59, e51–e60. blind exploratory RCT from a low income country. Behavioural and
Cuijpers P, Berking M, Andersson G, Quigley L, Kleiboer A and Cognitive Psychotherapy 42, 693–705.
Dobson KS (2013) A meta-analysis of cognitive-behavioural therapy for Joling KJ, van Hout HP, van’t Veer-Tazelaar PJ, van der Horst HE,
adult depression, alone and in comparison with other treatments. Cuijpers P, van de Ven PM and van Marwijk HW (2011) How effective
Canadian Journal of Psychiatry 58, 376–385. is bibliotherapy for very old adults with subthreshold depression? A rando-
Cuijpers P, Karyotaki E, Weitz E, Andersson G, Hollon SD and van mized controlled trial. American Journal of Geriatric Psychiatry 19, 256–265.
Straten A (2014) The effects of psychotherapies for major depression in Kanter JW, Santiago-Rivera AL, Santos MM, Nagy G, Lopez M,
adults on remission, recovery and improvement: a meta-analysis. Journal Hurtado GD and West P (2015) A randomized hybrid efficacy and effect-
of Affective Disorders 159, 118–126. iveness trial of behavioral activation for Latinos with depression. Behavior
Cummings P (2009) The relative merits of risk ratios and odds ratios. Archives Therapy 46, 177–192.
of Pediatrics and Adolescent Medicine 163, 438–445. Kessler D, Lewis G, Kaur S, Wiles N, King M, Weich S, Sharp DJ, Araya R,
Dalgard OS (2006) A randomized controlled trial of a psychoeducational Hollinghurst S and Peters TJ (2009) Therapist-delivered internet psycho-
group program for unipolar depression in adults in Norway therapy for depression in primary care: a randomised controlled trial. The
(NCT00319540). Clinical Practice & Epidemiology in Mental Health 2, 15. Lancet 374, 628–634.
1274 Fredrik Santoft et al.

Kivi M, Eriksson MC, Hange D, Petersson EL, Vernmark K, Johansson B Serfaty MA, Haworth D, Blanchard M, Buszewicz M, Murad S and King M
and Björkelund C (2014) Internet-based therapy for mild to moderate (2009) Clinical effectiveness of individual cognitive behavioral therapy for
depression in Swedish primary care: short term results from the depressed older people in primary care: a randomized controlled trial.
PRIM-NET randomized controlled trial. Cognitive Behaviour Therapy 43, Archives of General Psychiatry 66, 1332–1340.
289–298. Smit A, Kluiter H, Conradi HJ, van der Meer K, Tiemens BG, Jenner JA,
Laidlaw K, Davidson K, Toner H, Jackson G, Clark S, Law J, Howley M, van Os TW and Ormel J (2006) Short-term effects of enhanced treatment
Bowie G, Connery H and Cross S (2008) A randomised controlled trial for depression in primary care: results from a randomized controlled trial.
of cognitive behaviour therapy vs treatment as usual in the treatment of Psychological Medicine 36, 15–26.
mild to moderate late life depression. International Journal of Geriatric Sørensen Høifødt R, Lillevoll KR, Griffiths KM, Wilsgaard T, Eisemann M,
Psychiatry 23, 843–850. Waterloo K and Kolstrup N (2013) The clinical effectiveness of web-based
Levin W, Campbell DR, McGovern KB, Gau JM, Kosty DB, Seeley JR and cognitive behavioral therapy with face-to-face therapist support for
Lewinsohn PM (2011) A computer-assisted depression intervention in depressed primary care patients: randomized controlled trial. Journal of
primary care. Psychological Medicine 41, 1373–1383. Medical Internet Research 15, 301–322.
Linde K, Sigterman K, Kriston L, Rucker G, Jamil S, Meissner K and Spek V, Nyklicek I, Smits N, Cuijpers P, Riper H, Keyzer J and Pop V
Schneider A (2015) Effectiveness of psychological treatments for depressive (2007) Internet-based cognitive behavioural therapy for subthreshold
disorders in primary care: systematic review and meta-analysis. Annals of depression in people over 50 years old: a random controlled clinical trial.
Family Medicine 13, 56–68. Psychological Medicine 37, 1797–1806.
Lipsey MW and Wilson DB (2001) Practical Meta-Analysis. Thousand Oaks, Teasdale JD, Fennell MJ, Hibbert GA and Amies PL (1984) Cognitive ther-
CA: Sage Publications, Inc. apy for major depressive disorder in primary care. British Journal of
Lovell K, Bower P, Richards D, Barkham M, Sibbald B, Roberts C, Davies L, Psychiatry 144, 400–406.
Rogers A, Gellatly J and Hennessy S (2008) Developing guided self-help Thornicroft G, Chatterji S, Evans-Lacko S, Gruber M, Sampson N,
for depression using the Medical Research Council complex interventions Aguilar-Gaxiola S, Al-Hamzawi A, Alonso J, Andrade L, Borges G,
framework: a description of the modelling phase and results of an explora- Bruffaerts R, Bunting B, de Almeida JM, Florescu S, de Girolamo G,
tory randomised controlled trial. BMC Psychiatry 8, 91. Gureje O, Haro JM, He Y, Hinkov H, Karam E, Kawakami N, Lee S,
McHugh RK, Whitton SW, Peckham AD, Welge JA and Otto MW (2013) Navarro-Mateu F, Piazza M, Posada-Villa J, de Galvis YT and
Patient preference for psychological vs pharmacologic treatment of psychi- Kessler RC (2017) Undertreatment of people with major depressive dis-
atric disorders: a meta-analytic review. Journal of Clinical Psychiatry 74, order in 21 countries. British Journal of Psychiatry 210, 119–124.
595–602. Twomey C, O’Reilly G and Byrne M (2015) Effectiveness of cognitive behav-
Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Preferred reporting items ioural therapy for anxiety and depression in primary care: a meta-analysis.
for systematic reviews and meta-analyses: the PRISMA statement. BMJ 339, Family Practice 32, 3–15.
b2535. Viechtbauer W (2010) Conducting meta-analyses in R with the metafor pack-
Moussavi S, Chatterji S, Verdes E, Tandon A, Patel V and Üstün B (2007) age. Journal of Statistical Software 36, 1–48.
Depression, chronic diseases, and decrements in health: results from the Wahl I, Löwe B, Bjorner JB, Fischer F, Langs G, Voderholzer U, Aita SA,
World Health Surveys. The Lancet 370, 851–858. Bergemann N, Brähler E and Rose M (2014) Standardization of depres-
National Collaborating Centre for Mental Health (2010) Depression: The sion measurement: a common metric was developed for 11 self-report
Treatment and Management of Depression in Adults (Update). Leicester depression measures. Journal of Clinical Epidemiology 67, 73–86.
and London: The British Psychological Society and the Royal College of Ward E, King M, Lloyd M, Bower P, Sibbald B, Farrelly S, Gabbay M,
Psychiatrists. Tarrier N and Addington-Hall J (2000) Randomised controlled trial of
Naylor EV, Antonuccio DO, Litt M, Johnson GE, Spogen DR, Williams R, non-directive counselling, cognitive-behaviour therapy, and usual general
McCarthy C, Lu MM, Fiore DC and Higgins DL (2010) Bibliotherapy as a practitioner care for patients with depression. I: clinical effectiveness. BMJ
treatment for depression in primary care. Journal of Clinical Psychology in 321, 1383–1388.
Medical Settings 17, 258–271. Watson JC, Gordon LB, Stermac L, Kalogerakos F and Steckley P (2003)
Ormel J, Petukhova M, Chatterji S, Aguilar-Gaxiola S, Alonso J, Comparing the effectiveness of process-experiential with cognitive-
Angermeyer MC, Bromet EJ, Burger H, Demyttenaere K, de behavioral psychotherapy in the treatment of depression. Journal of
Girolamo G, Haro JM, Hwang I, Karam E, Kawakami N, Lepine JP, Consulting and Clinical Psychology 71, 773–781.
Medina-Mora ME, Posada-Villa J, Sampson N, Scott K, Ustun TB, Watts SE, Turnell A, Kladnitski N, Newby JM and Andrews G (2015)
Von Korff M, Williams DR, Zhang M and Kessler RC (2008) Disability Treatment-as-usual (TAU) is anything but usual: a meta-analysis of CBT versus
and treatment of specific mental and physical disorders across the world. TAU for anxiety and depression. Journal of Affective Disorders 175, 152–167.
British Journal of Psychiatry 192, 368–375. Wiles N, Hollinghurst S, Mason V, Musa M, Burt V, Hyde J, Jerrom B,
Power MJ and Freeman C (2012) A randomized controlled trial of IPT versus Lewis G and Kessler D (2008) A randomized controlled trial of cognitive
CBT in primary care: with some cautionary notes about handling missing behavioural therapy as an adjunct to pharmacotherapy in primary care
values in clinical trials. Clinical Psychology & Psychotherapy 19, 159–169. based patients with treatment resistant depression: a pilot study.
R Core Team (2016) R: A Language and Environment for Statistical Behavioural and Cognitive Psychotherapy 36, 21–33.
Computing. Vienna, Austria: R Foundation for Statistical Computing. Wiles N, Thomas L, Abel A, Ridgway N, Turner N, Campbell J, Garland A,
Ross M and Scott M (1985) An evaluation of the effectiveness of individual Hollinghurst S, Jerrom B, Kessler D, Kuyken W, Morrison J, Turner K,
and group cognitive therapy in the treatment of depressed patients in an Williams C, Peters T and Lewis G (2013) Cognitive behavioural therapy as
inner city health centre. Journal of the Royal College of General an adjunct to pharmacotherapy for primary care based patients with treat-
Practitioners 35, 239–242. ment resistant depression: results of the CoBalT randomised controlled
Schulz KF, Altman DG, Moher D and Group C (2010) CONSORT 2010 trial. The Lancet 381, 375–384.
statement: updated guidelines for reporting parallel group randomised Willemse GR, Smit F, Cuijpers P and Tiemens BG (2004) Minimal-contact
trials. BMC Medicine 8, 18. psychotherapy for sub-threshold depression in primary care. British Journal
Scott AI and Freeman CP (1992) Edinburgh primary care depression study: of Psychiatry 185, 416–421.
treatment outcome, patient satisfaction, and cost after 16 weeks. BMJ 304, Williams C, Wilson P, Morrison J, McMahon A, Walker A, Allan L,
883–887. McConnachie A, McNeill Y and Tansey L (2013) Guided self-help cogni-
Scott C, Tacchi MJ, Jones R and Scott J (1997) Acute and one-year outcome tive behavioural therapy for depression in primary care: a randomised con-
of a randomised controlled trial of brief cognitive therapy for major trolled trial. PLoS ONE 8, e52735.
depressive disorder in primary care. British Journal of Psychiatry 171, World Health Organization (2008) The Global Burden of Disease: 2004
131–134. Update. Geneva, Switzerland: World Health Organization.
Copyright © Cambridge University Press 2019