Regular Article

Psychother Psychosom 2015;84:241–249 Received: September 18, 2014

Accepted after revision: February 2, 2015
DOI: 10.1159/000377705
Published online: May 23, 2015

Are Group Psychotherapeutic Treatments

Effective for Patients with Schizophrenia?
A Systematic Review and Meta-Analysis
Stavros Orfanos Ciara Banks Stefan Priebe 
Unit for Social and Community Psychiatry, Queen Mary University of London, London, UK

Key Words (standard mean difference = –0.37, 95% confidence interval

Schizophrenia · Group psychotherapy · Systematic review · –0.60, –0.14; p < 0.01, I2 = 59.8%) only when compared to
Meta-analysis treatment as usual and not to active sham groups. Improved
social functioning was reported as a treatment outcome in
the majority of studies compared to treatment as usual. The
Abstract ‘group effect’ on negative symptoms was positively related
Background: Different psychotherapeutic treatments for to ‘treatment intensity’ (β = 0.32, standard error = 0.121; p <
schizophrenia are delivered in groups. However, little is 0.05). Conclusion: Group psychotherapeutic treatments can
known about the effectiveness of these group therapies for improve negative symptoms and social functioning deficits
people with schizophrenia across different treatments with in the treatment of schizophrenia. The effect occurs across
varying therapeutic orientations. This review aimed to (1) es- different treatments and appears to be non-specific. Future
timate the effect of different group psychotherapeutic treat- research should identify the underlying mechanisms for the
ments for schizophrenia and (2) explore whether any overall positive effect of participating in groups and explore how
‘group effect’ is moderated by treatment intensity, diagnos- they can be maximised to increase the therapeutic benefit.
tic homogeneity and therapeutic orientation. Methods: A © 2015 S. Karger AG, Basel
systematic search of randomised controlled trials exploring
the effectiveness of group psychotherapeutic treatments for
people with schizophrenia was conducted. Random-effect Introduction
meta-analyses on endpoint symptom scores compared
group psychotherapeutic treatments with treatment as usu- In accordance with guidelines from the National Insti-
al and active sham groups. Findings on social functioning tute for Health and Care Excellence in the United King-
were described narratively, and meta-regression analyses on dom [1] and the Schizophrenia Patient Outcomes Re-
group characteristics were carried out. Results: Thirty-four search Team in the United States [2], psychotherapeutic
eligible trials were included. A weak-to-moderate significant treatments are widely regarded as a necessary interven-
between-group difference in favour of group psychothera- tion for schizophrenia. In particular, there has been a
peutic treatments was found for negative symptom scores growing interest in the development and delivery of psy-
128.103.149.52 - 5/26/2015 11:10:41 AM

© 2015 S. Karger AG, Basel Stavros Orfanos

0033–3190/15/0844–0241$39.50/0 Unit for Social and Community Psychiatry
Harvard University

Queen Mary University of London, Newham Centre for Mental Health

Downloaded by:

E-Mail karger@karger.com
London E13 8SP (UK)
www.karger.com/pps
E-Mail s.orfanos @ qmul.ac.uk
chotherapeutic treatments in a group format for this pop- more, whilst attempts have been made to summarise
ulation [3]. findings from controlled trials exploring the effectiveness
From an economic perspective, a group setting is seen of group psychotherapeutic treatments for schizophrenia
as a useful approach, as it allows for one therapist to treat [8, 22, 24, 25], the conclusions from these studies are lim-
several people at the same time [4]. From a clinical per- ited in scope. For example, the most recent attempt by
spective, group psychotherapeutic treatments are also be- Segredou et al. [8] does not include evidence from non-
lieved to offer social advantages relevant to this popula- verbal creative group arts therapies (including music
tion [4–8], who often have smaller social networks and therapy, body psychotherapy and art therapy), which
less satisfactory interpersonal relationships compared to have been shown to be effective in reducing negative
a healthy population [9]. Seminal work on group thera- symptoms [1]. Furthermore, their findings are limited to
peutic processes [10] (including group cohesion, instilla- a descriptive analysis of the literature.
tion of hope, interpersonal learning and sharing of infor- To date, no attempt has been made to statistically pool
mation) supports the notion that the group setting can be the existing evidence using meta-analytical techniques.
utilised as an agent of change in group psychotherapeutic Consequently, it is unclear whether group psychothera-
treatments. peutic treatments have an effect across different treat-
Evidence from randomised controlled trials on group ment models for schizophrenia with varying therapeutic
cognitive behavioural therapy (CBT) [6], group social orientations. Therefore, this review aimed to establish
skills training [11], group music therapy [12] and group whether there is an overall ‘group effect’ across a range of
psycho-education [13] suggests that group psychothera- group psychotherapeutic treatments as compared to
peutic treatments, with different therapeutic orienta- treatment as usual (TAU) [26]. If people with schizophre-
tions, can be effective in improving a number of clinical nia benefit from a non-specific ‘group experience’, one
outcomes for people with schizophrenia. In their review would expect to see clinical improvements in participants
of controlled studies for schizophrenia conducted be- across a range of group psychotherapeutic treatments. If
tween 1986 and 2006, Segredou et al. [8] found that, de- this effect was in fact due to processes in the ‘group’, it
scriptively, all of the 23 studies they identified showed a might not be apparent when compared to an active sham
positive effect on either symptom or skills outcomes. group [1]. In the literature, active sham groups are de-
In the treatment of positive symptoms, including hal- fined as a group condition aimed at controlling for non-
lucinations and delusions, a group format has been sug- specific effects of the ‘group’ (for example, therapist atten-
gested to provide an opportunity for participants to share tion, therapeutic rationale and therapeutic alliance) and
experiences and reflect on similarities, which, in turn, can strictly does not involve any of the unique psychothera-
aid the restructuring of false beliefs [7, 14, 15]. In the peutic techniques under investigation [27, 28]. Therefore,
treatment of negative symptoms, such as lack of speech, we also assessed whether there is an effect of group psy-
social withdrawal, blunted affect and social functioning chotherapeutic approaches compared to active sham
deficits, it has been argued that group members serve as groups. Finally, we aimed to explore which group charac-
models and reinforcers for each other, which, in turn, can teristics contribute to any potential group effect. In par-
help the development of relationships [11, 16]. An im- ticular, we considered the therapeutic orientation, num-
proved understanding of how to treat negative symptoms ber of sessions/length of intervention [22] and/or diag-
is of particular importance, given that these symptoms nostic homogeneity [7] as potentially important factors
are more resistant to medication than positive symptoms for the impact of group psychotherapeutic treatments [8,
[17] and highly related to poor social functioning [18] 24].
and a poor quality of life [19].
Despite the potential cost benefits and clinical advan-
Methods
tages of a group setting, little methodologically robust re-
search has explored whether group psychotherapeutic Search Strategy
treatments have a benefit for people with schizophrenia A protocol was developed using the Preferred Reporting Items
[20–22] and whether they are effective across specific for Systematic Reviews and Meta-Analyses Statement (PRISMA)
[29]. The electronic databases searched included PsychINFO
therapeutic orientations [23]. At present, too few studies
(1806 to March 2014), Medline (1946 to March 2014), Embase
are available to test the effectiveness of group treatments (1974 to March 2014) and AMED (1985 to March 2014). MeSH
as compared to individual treatments for each psycho- and text word search terms relating to ‘group psychotherapeutic
therapeutic treatment of schizophrenia [22]. Further- therapies’ AND ‘randomised controlled trials’ AND ‘schizophre-
128.103.149.52 - 5/26/2015 11:10:41 AM

242 Psychother Psychosom 2015;84:241–249 Orfanos/Banks/Priebe

DOI: 10.1159/000377705
Harvard University
Downloaded by:
nia’ (online suppl. table  1 for Medline search terms; see www. end of treatment, measured as a continuous variable. As mea-
karger.com/doi/10.1159/000377705) were used for each database. sured by the Positive and Negative Symptom Scale (PANSS) [31],
Search terms were modified for each database. Where outcome positive symptoms include delusions, grandiosity, suspicious-
data were not fully reported, first and second authors were con- ness, hostility and hallucinations; negative symptoms include
tacted via e-mail requesting any missing information. Hand emotional withdrawal, poor rapport, difficulty in abstract think-
searching of the following key journals was conducted: Group ing, blunt affect and social withdrawal; general symptoms include
Therapy, Behavioural Group Therapy, The Clinical Psychologist, anxiety, depression, insight and guilt, while total symptom scores
Group Analysis, and the Journal of Contemporary Psychotherapy. are the sum of positive, negative and general symptoms. The orig-
A grey literature search of the Cochrane database and websites inal authors’ definitions of symptoms were followed, rather than
including the Health Technology Assessment, the National Insti- a predefined operationalised definition. Social functioning scores
tute of Mental Health, the Wellcome Trust and the Medical Re- were measured as a secondary outcome and examined descrip-
search Council was also conducted. Additionally, studies cited in tively.
relevant reviews on psychotherapeutic treatments for schizophre-
nia were hand searched. Data Analysis
For each study, means and standard deviations were extracted.
Eligibility Criteria Standard mean differences with 95% confidence intervals were cal-
Studies at the title and abstract phase were screened for the fol- culated from the data extracted. Data were pooled using a random-
lowing inclusion criteria: (1) randomised controlled trial; (2) psy- effects meta-analysis in STATA version 12. Endpoint scores from
chotherapeutic treatments provided in treatment condition, and (3) both the treatment and control conditions were used to assess the
included participants with a diagnosis of schizophrenia and related impact of group psychotherapeutic treatments on symptoms of
disorders. Studies were excluded if they: (1) involved individualised schizophrenia. Data were pooled in such a way that a standard
treatment; (2) involved family therapy and/or family intervention, mean difference <0 favoured the treatment condition. Heteroge-
and (3) included participants aged ≤16 years. Studies that were only neity was assessed visually and by the I2 statistic [32].
abstract publications and/or protocols were not included. The first set of meta-analyses explored group psychotherapeu-
At the full paper review stage, studies were further excluded ac- tic treatments compared to TAU for positive, negative, general and
cording to the following criteria: (1) a sample with <85% of par- total symptom scores. The second set of meta-analyses explored
ticipants diagnosed with schizophrenia, schizotypal, schizoaffec- group psychotherapeutic treatments compared to active sham
tive and/or other non-affective psychotic disorders outlined in the groups for positive, negative, general and total symptom scores.
Diagnostic Statistical Manual and International Classification of Planned sensitivity analyses were conducted to explore the robust-
Diseases; (2) did not measure either symptoms of schizophrenia ness of the results. In these analyses, studies with a high risk of bias
(either positive, negative, general or total symptoms) or social and studies where baseline mean symptom scores varied across the
functioning; (3) did not make a clear reference to a group format treatment and control condition were excluded.
in the treatment condition; (4) was not published in a language us- Post hoc meta-regression analyses were used to explore which
ing Latin-based characters; (5) the control condition was delivered factors were driving significant group effects found across the
as a group psychotherapeutic treatment rather than an active sham main meta-analyses and planned sensitivity analyses. Meta-re-
group (i.e. active discussion group, support group, counselling gression analyses were therefore only conducted on studies that
group, occupational therapy group or problem-solving discussion compared a group psychotherapeutic treatment to TAU and not
group) or TAU, where ‘waitlist control group’ (no treatment of- active sham groups. The first two meta-regression analyses ex-
fered until the intervention condition has received their treatment) plored the effect of therapeutic orientation by dichotomising psy-
and ‘standard psychiatric care’ are considered as TAU. chotherapeutic treatments as (1) non-verbal arts therapies (in-
cluding music therapy, body-oriented psychotherapy and art
Study Selection and Data Extraction therapy) versus non-arts therapies and (2) cognitive-behavioural
The first author (S.O.) conducted the initial screening of all the approaches (including cognitive-behavioural social skills train-
titles and abstracts and all the studies at the full paper review phase. ing and compensatory cognitive training) versus other therapeu-
The second author (C.B.) re-extracted 50% of the studies at full tic approaches. The second meta-regression analysis explored the
paper review and 20% of abstracts, randomly selected using a ran- effect of treatment ‘intensity’, calculated as a continuous variable
dom number generator. Any ambiguity was resolved with the third from the duration of a session (in hours) multiplied by the num-
author (S.P.). All included studies were independently extracted by ber of sessions offered in the treatment. A log transformation was
two reviewers (S.O. and C.B.) using a structured format (online conducted on this variable to ensure that it was normally distrib-
suppl. table 2). The Cochrane risk of bias tool was used to assess uted. The third meta-regression analysis explored the effect of
the studies [30]. It was agreed by all authors to exclude the ‘blind- ‘diagnosis’ as a dichotomised variable, comparing studies that in-
ing of personnel’ category, given that in trials examining the effec- cluded ‘schizophrenia’ and ‘schizophrenia and related disorders’.
tiveness of group psychotherapeutic treatments, it is not possible Due to the varied range of assessments used to measure social
to keep participants blind to their treatment allocation. ‘High’-risk functioning, it was decided a priori to not conduct a meta-analysis
studies were identified as those that scored a ‘high risk’ for at least on this outcome. As outlined by Higgins et al. [30], a meta-analysis
4 of the 6 categories prior to data extraction. should only be conducted if outcomes share similar clinical char-
acteristics. Instead, outcomes on social functioning deficits were
Outcomes discussed descriptively in a narrative synthesis, which included a
The primary outcome was the mean symptom scores (includ- description of statistical outcomes and author conclusions.
ing positive, negative, general and/or total symptom scores) at the
128.103.149.52 - 5/26/2015 11:10:41 AM

Group Psychotherapy for Schizophrenia Psychother Psychosom 2015;84:241–249 243

DOI: 10.1159/000377705
Harvard University
Downloaded by:
 Results an upper limit. In total, 2,634 patients were represented
in the 32 data sets included in this review, of whom 1,334
Search Results participants were represented in the treatment condition
A total of 5,078 studies were identified in the electron- and 1,300 were represented in the control condition.
ic database search. Following the exclusion of duplica-
tions (n = 1,962) and the removal of studies at the title Risk of Bias
screening phase (n = 1,564), 1,552 abstract articles were In total, 3 studies scored ‘high risk of bias’ for at least
reviewed (online suppl. fig. 1). Of the 324 studies identi- 4 of the 6 categories, and, therefore, were rated as low
fied for full paper review, 34 studies were included. Seven quality (online suppl. fig. 2). With the exception of the
studies [33–39] used data from 3 data sets, 1 study [40] funnel plot on positive symptoms, all plots are slightly
included data from 2 separate trials, and 1 study [41] had asymmetric, with an absence of data on the lower right
2 control arms. Hence, in total, 32 data sets were included hand side of the plot (online suppl. fig. 3). Egger tests of
in the final meta-analysis. publication bias found no statistical evidence of publica-
tion bias for negative, positive or general symptom scores.
Study Characteristics However, there was statistical evidence of publication
The study characteristics of the studies from which the bias for the studies included in the meta-analyses com-
32 data sets were included are summarised in online sup- paring group therapeutic treatments with TAU and active
plementary table 3. In total, 13 data sets compared a group sham groups for total symptoms (β = 0.975, p = 0.01) and
psychotherapeutic treatment to an active sham group [37, for studies included in the total symptoms planned sen-
39, 41–51], and 19 data sets compared a group psycho- sitivity analyses (β = 0.999, p = 0.02).
therapeutic treatment to TAU [7, 12, 13, 34, 40, 41, 52–
63]. Thirty-one percent of the interventions were cogni- Impact of Group Psychotherapeutic Treatments on
tive-behavioural approaches (including cognitive-behav- Symptoms
ioural social skills training and compensatory cognitive Table 1 summarises findings from the meta-analyses
training), 19% came under the umbrella term ‘non-verbal comparing group psychotherapeutic treatments with
arts therapies’ (including music therapy, body-oriented TAU and active sham groups and from sensitivity analy-
psychotherapy and art therapy), and the remaining 50% ses (which excluded studies with a high risk of bias and
included a range of therapeutic orientations such as cog- studies where baseline mean symptom scores varied
nitive remediation therapy, psycho-education and inte- across the treatment and control condition; online suppl.
grated approaches. These approaches were varied in table 6) of endpoint outcomes for positive, negative and
terms of their therapeutic focus and therapeutic outcome general symptoms. Separate analyses were conducted for
(online suppl. table  4), including positive symptoms studies that compared a group psychotherapeutic treat-
(13%), social functioning (22%), cognitive functioning ment to TAU and those that compared a group psycho-
(22%) and negative symptoms (9%), a range of outcomes therapeutic treatment to an active sham group.
(22%) or outcomes that did not fit into any of these cat- In the meta-analyses comparing group psychothera-
egories (22%). The most common measure of symptoms peutic treatments to TAU, there was a significant be-
was the PANSS (81, 63, 92.9 and 68.4% for positive, neg- tween-group difference for endpoint negative symptom
ative, general and total symptom scores, respectively; on- scores, endpoint general symptom scores and endpoint
line suppl. table 5). Twenty-two studies (71%) were con- total symptom scores in favour of the treatment condi-
ducted in an outpatient setting, 12 studies (38%) stated tion. No main effect was found for positive symptom
the use of an intention-to-treat design, and 9 studies scores. Findings were robust across planned sensitivity
(28%) included a sample size calculation. The average fol- analyses for both negative and positive symptoms. How-
low-up rate was 8% for studies that compared a group ever, the effects on general and total symptoms were no
psychotherapeutic treatment to an active sham group and longer significant in the planned sensitivity analyses, fol-
7% for studies that compared group psychotherapeutic lowing the removal of studies rated as high risk of bias.
treatment to TAU. On average, 38 and 34% of the treat- Forest plots for group psychotherapeutic treatments
ment and control groups were female, respectively. The compared to TAU and active sham groups are shown in
age of the participants ranged from 17 to 78 years, and the online supplementary figure 4a and b.
mean age reported was 39 years; 4 studies did not have There was no evidence of a significant between-group
any information on age range, and 4 studies did not state difference for endpoint negative symptoms, positive
128.103.149.52 - 5/26/2015 11:10:41 AM

244 Psychother Psychosom 2015;84:241–249 Orfanos/Banks/Priebe

DOI: 10.1159/000377705
Harvard University
Downloaded by:
Table 1. Summary of meta-analyses for positive, negative, general and total symptom scores, comparing group psychotherapeutic treat-
ments to TAU and active sham groups

Outcome Analysis Studies Participants SMD p value I², %

Negative symptoms Treatment vs. TAU 15 893 –0.37 (–0.60, –0.14) 0.002 59.8
Treatment vs. sham 12 783 –0.09 (–0.36, 0.19) 0.542 68.5
Sensitivity (treatment vs. TAU) 12 762 –0.40 (–0.67, –0.13) 0.004 66.6
Sensitivity (treatment vs. sham) 10 687 –0.09 (–0.33, 0.16) 0.504 56.3
Positive symptoms Treatment vs. TAU 11 730 –0.06 (–0.25, 0.13) 0.553 29.8
Treatment vs. sham 9 654 0.07 (–0.32, 0.18) 0.590 49.7
Sensitivity (treatment vs. TAU) 8 654 –0.02 (–0.21, 0.18) 0.877 27.5
Sensitivity (treatment vs. sham) 9 654 0.07 (–0.32, 0.18) 0.590 49.7
General symptoms Treatment vs. TAU 9 625 –0.22 (–0.43, –0.02) 0.035 27.8
Treatment vs. sham 6 521 0.17 (–0.76, 0.42) 0.575 87.9
Sensitivity (treatment vs. TAU) 7 593 –0.13 (–0.29, –0.03) 0.120 0
Sensitivity (treatment vs. sham) 4 425 –0.16 (–0.46, 0.14) 0.303 49.8
Total symptoms Treatment vs. TAU 9 651 –0.41 (–0.69, –0.13) 0.004 60.3
Treatment vs. sham 10 812 0.12 (–0.74, 0.35) 0.479 91.8
Sensitivity (treatment vs. TAU) 5 514 –0.33 (–0.66, 0.01) 0.052 66.3
Sensitivity (treatment vs. sham) 8 538 –0.48 (–1.10, 0.11) 0.108 88.5

Figures in parentheses are 95% confidence intervals. Treatment vs. TAU = Meta-analysis comparing group psychotherapeutic treat-
ments with TAU, waitlist control or standard psychiatric care; treatment vs. sham = meta-analysis comparing group psychotherapeutic
treatments with active sham groups; sensitivity = sensitivity analysis; SMD = standardised mean difference; I² = heterogeneity.

symptoms, general symptoms or total symptoms for on social functioning (see online suppl. table 7). Six of the
studies that compared a group psychotherapeutic treat- 11 studies [7, 12, 34, 52, 58, 63] found a statistically sig-
ment and an active sham group. nificant improvement favouring the group psychothera-
peutic treatments over the control condition, and 5 stud-
Meta-Regression Analyses ies did not [41, 53, 57, 59, 61]. Nine different measures of
Meta-regression analyses were limited to outcomes on social functioning were reported in the 11 studies.
negative symptoms, given that no effect of group psycho-
therapeutic treatments was found on positive symptoms
and that findings on the impact of general and total symp- Discussion
toms were inconsistent across the planned sensitivity
analyses. The effect of group psychotherapeutic treat- This review found that group psychotherapeutic treat-
ments on negative symptoms was not moderated by the ments were more effective in reducing negative symp-
therapeutic orientation or diagnostic homogeneity (ta- toms than TAU across a diverse range of psychothera-
ble 2). However, the effect size on negative symptoms was peutic orientations. This effect was apparent only when
positively moderated by the treatment intensity of the these group psychotherapeutic treatments were com-
group psychotherapeutic treatments (β = 0.32, SE = 0.121, pared to TAU, not to active sham groups. There was no
p < 0.05). The adjusted R2 value indicated that 31% of the evidence that group psychotherapeutic treatments im-
variance in this model was accounted for by the intensity proved positive symptoms across a range of group psy-
of sessions, measured as number of sessions available in chotherapeutic treatments compared to TAU or active
the group psychotherapeutic treatments. sham groups. Furthermore, any evidence that general
symptoms and total symptoms improved in favour of the
Impact of Group Psychotherapeutic Treatments on group psychotherapeutic treatment condition compared
Social Functioning to TAU was no longer significant when eliminating stud-
In total, 11 of the 19 studies which compared a group ies rated as high risk of bias. The narrative summary of
psychotherapeutic treatment to TAU reported outcomes studies indicated that overall, participants in group psy-
128.103.149.52 - 5/26/2015 11:10:41 AM

Group Psychotherapy for Schizophrenia Psychother Psychosom 2015;84:241–249 245

DOI: 10.1159/000377705
Harvard University
Downloaded by:
 Table 2. Summary of meta-regression analysis on endpoint negative symptom standardised mean difference
scores comparing group psychotherapeutic treatments and TAU

Characteristics Coefficient SE I2, % Adjusted p value

R², %

Therapeutic orientation: arts and others 0.220 0.201 61.39 –1.41 0.282
Therapeutic orientation: cognitive-behavioural and others –0.004 0.185 63.46 –6.20 0.985
Intensity (log transformed) 0.320 0.121 55.10 31.02 0.014
Diagnostic homogeneity –0.001 0.176 61.65 –6.94 0.994

SE = Standard error; adjusted R2 = variance; I2 = heterogeneity.

chotherapeutic treatments benefited more in terms of re- tion bias was found for positive, negative or general
duced social functioning deficits in the treatment condi- scores. However, there was statistical evidence of bias for
tion compared to TAU. No evidence was found for an total symptom scores.
effect of therapeutic orientation or diagnostic homogene- Furthermore, I2 scores from meta-analyses on nega-
ity. However, there was a significant positive relationship tive symptoms indicate a moderate to high level of het-
between treatment intensity and reduced negative symp- erogeneity, i.e. I2 scores between 50 and 75% [32]. How-
toms. ever, visual examination of the forest plots (online suppl.
This study has a number of strengths. To our knowl- fig. 4a and b) indicated a consistent overlap between the
edge, this is the first systematic review to explore the ef- confidence intervals of the effect sizes in the majority of
fectiveness of psychotherapeutic treatments delivered in the studies, hence a minimal heterogeneity between stud-
groups using meta-analytic techniques. We used rigorous ies. It is therefore likely that the high heterogeneity is be-
methods and a wide array of search terms encompassing ing driven by a minority of outliers – Vreeland et al. [62]
a broad range of verbal and non-verbal psychotherapeu- and Levine et al. [46] in the TAU and active sham group
tic group treatments. Stringent measures controlled for analyses, respectively – rather than by the significant vari-
study quality. For example, all studies were independent- ation between studies.
ly extracted and assessed for risk of bias. Low-quality This review is also limited to symptom and social func-
studies were excluded in planned sensitivity analyses tioning outcomes. Given that group psychotherapeutic
rather than being rated on a quality scale and controlled treatments have been implicated with a variety of im-
for statistically [64]. proved outcomes [8, 22, 66], conclusions on their effec-
There are also a number of potential limitations. The tiveness are therefore incomplete. To address this limita-
majority of the samples represented were outpatients tion, separate analyses were conducted on major symp-
(71%) and male (64%), which may limit the generalisabil- tom domains.
ity of the findings. However, as noted by Jane-Wit et al. Most studies were not reported as intention-to-treat
[65], an important factor contributing to different results analyses. Since a dropout is unlikely to be due to random
between randomised controlled trials is the difference in factors and since only few studies reported reasons for a
patient characteristics. Hence, the clinical validity of the dropout, this may introduce completer-only bias. Given
findings is strengthened by the homogeneous population that too few studies carried out an intention-to-treat anal-
across the studies. ysis, a further sensitivity analysis on this sub-group of
There is also the possibility of publication bias. Visual studies was not deemed suitable. However, encouraging-
examination of funnel plots (online suppl. fig. 3) for neg- ly, the follow-up assessment rate at the end of treatment
ative, general and total symptoms indicate that there are was high across both treatment versus active sham groups
slightly fewer trials with small samples favouring the con- and treatment versus TAU comparisons.
trol condition represented in this review. This may have Finally, group psychotherapeutic treatments have not
biased the results of the review against the control condi- been assessed against individual psychotherapeutic treat-
tion. To account for this, statistical tests of publication ments. Without controlling for the specific factors poten-
bias were conducted. No statistical evidence of publica- tially relevant to the psychotherapeutic treatment itself, it
128.103.149.52 - 5/26/2015 11:10:41 AM

246 Psychother Psychosom 2015;84:241–249 Orfanos/Banks/Priebe

DOI: 10.1159/000377705
Harvard University
Downloaded by:
is difficult to make firm conclusions about the benefits of more, results from this review are not consistent with the
non-specific group effects. Whilst Wykes et al. [67] found notion that group processes can be effective in aiding the
no difference in the two treatment modalities, the valid- restructuring of false beliefs around delusions or halluci-
ity of this comparison is limited by the fact that only 7 nations [15, 71] in the treatment of positive symptoms
group CBT studies were compared to 26 studies on indi- [20, 23, 26]. As suggested by Wykes et al. [7], it might be
vidual CBT. difficult for therapists to flexibly respond to a wide variety
Overall, evidence from this review supports the view of individual therapeutic needs when addressing positive
that group mechanisms underpinning different group symptoms in groups. Hence, a ‘group effect’ for positive
psychotherapeutic treatments can be clinically advanta- symptoms might be specific only to highly homogenous
geous for people with schizophrenia [8, 26, 66] in the groups, such as hearing voices groups [14], rather than a
treatment of negative symptoms [16] and social function- non-specific shared effect [23].
ing deficits [11]. As argued by Kanas [66], ‘the group ex- In the meta-regression analyses, there was no evidence
perience itself’ (p. 10) appears to be clinically useful for that the ‘therapeutic orientation’, in terms of arts versus
this population who are often isolated and relate poorly non-arts and CBT versus non-CBT studies, moderated
with others. Given the effectiveness of group psychother- the group effect on negative symptoms. This supports the
apeutic treatments across different therapeutic orienta- idea that the benefit of group psychotherapeutic treat-
tions as compared to TAU and the absence of a significant ments, in terms of negative symptoms at least, is indepen-
effect as compared to active sham groups, the findings are dent of a particular therapeutic approach [20]. Further-
are consistent with the hypothesis that beneficial group more, there was no evidence to suggest that the degree of
mechanisms are non-specific [20, 26]. In support of the ‘diagnostic homogeneity’ moderated the effect of group
‘contextual’ model of psychotherapy of Wampold [23], psychotherapeutic treatments on negative symptoms.
these findings support the view that the benefit of group Therefore, the non-specific effect of negative symptoms
therapeutic mechanisms is due to common factors. held true for groups consisting of patients with a diagno-
The group effect is shared across different approach- sis of schizophrenia and related disorders [10]. However,
es and, potentially, also with sham groups, which inevi- the more ‘intense’ treatments were related to a greater dif-
tably have some group processes in common with psy- ference in negative symptom scores. This supports the
chotherapeutic groups. The fact that a group condition hypothesis that longer group psychotherapeutic treat-
is meant to be a sham condition in a trial can be obvious ments for schizophrenia are more effective than shorter
to researchers, but is often not evident to participants treatments [22]. This result further refines the impor-
taking part in the trial. However, with respect to sham tance of the ‘length’ of treatment to the ‘number of ses-
conditions, we cannot establish whether they are also ef- sions in a given space of time’ [72] as a more precise factor
fective in improving negative symptoms. Whilst we did that may influence the effectiveness of this treatment mo-
not find a difference with psychotherapeutic groups, the dality [22]. Effective group mechanisms may therefore
data do not allow us to test for non-inferiority, and a di- have a dose-response association, where short-term
rect comparison of sham groups with TAU was not pos- groups with few sessions do not exhaust the full potential
sible. of these mechanisms.
Whilst the effect size was only small to moderate for In conclusion, findings from this review suggest that
negative symptoms, it is bigger than the standardised group psychotherapeutic therapies, irrespective of their
mean difference scores for negative symptoms reported therapeutic approach, can improve negative symptoms
in meta-analyses of CBT for schizophrenia [64]. Further- and social functioning deficits in the treatment of schizo-
more, the effect size is comparable with those in studies phrenia. In support of the contextual model of psycho-
of social skills training for schizophrenia [68], cognitive therapy, the impact of group mechanisms on negative
remediation therapy for overall symptoms of schizophre- symptoms appear to be non-specific and shared across a
nia [69] and scores from meta-analyses of first- and sec- wide range of psychotherapeutic treatments delivered in
ond-generation anti-psychotics [70]. a group setting. Future research should identify the non-
In contrast to the review of Segredou et al. [8] on group specific mechanisms that explain the effect of group par-
psychotherapeutic treatments for schizophrenia, there ticipation on negative symptoms and explore ways to
was no evidence of improved positive symptoms. The in- strengthen them so that the therapeutic benefit is maxi-
clusion of non-verbal therapies and more precise statisti- mised.
cal techniques may account for this difference. Further-
128.103.149.52 - 5/26/2015 11:10:41 AM

Group Psychotherapy for Schizophrenia Psychother Psychosom 2015;84:241–249 247

DOI: 10.1159/000377705
Harvard University
Downloaded by:
 Acknowledgements Disclosure Statement

This study was supported by a grant from the East London NHS The authors declare no conflicts of interest.
Foundation Trust.

References
1 National Collaborating Centre for Mental 13 Chien WT, Lee IY: The mindfulness-based group treatment for social phobia: compari-
Health: Core interventions in the treatment psychoeducation program for Chinese pa- son with a credible placebo control. Cogn
and management of schizophrenia in adults tients with schizophrenia. Psychiatr Serv Ther Res 1990;14:1–23.
in primary and secondary care (Clinical 2013;64:376–379. 28 Safer DL, Hugo EM: Designing a control for a
Guideline CG82). London, National Institute 14 McLeod T, Morris M, Birchwood M, Dovey behavioral group therapy. Behav Ther 2006;
for Health and Clinical Excellence, 2009. A: Cognitive behavioural therapy group work 37:120–130.
2 Kreyenbuhl J, Buchanan RW, Dickerson FB, with voice hearers. Part 1. Br J Nurs 2007;16: 29 Moher D, Liberati A, Tetzlaff J, Altman DG:
Dixon LB: The Schizophrenia Patient Out- 248–252. Preferred reporting items for systematic re-
comes Research Team (PORT): updated 15 Guaiana G, Morelli AC, Chiodo D: Cognitive views and meta-analyses: the PRISMA state-
treatment recommendations 2009. Schizophr behavioural therapy (group) for schizophre- ment. Ann Intern Med 2009;151:264–269.
Bull 2010;36:94–103. nia – protocol. Cochrane Library 2012, DOI: 30 Higgins JPT, Altman DG, Gotzsche PC, Juni
3 Caruso R, Grassi L, Biancosino B, Marmai L, 10.1002/14651858. P, Moher D, Oxman AD: The Cochrane Col-
Bonatti L, Moscara M, Rigatelli M, Carr C, 16 Elis O, Caponigro JM, Kring AM: Psychoso- laboration’s tool for assessing risk of bias in
Priebe S: Exploration of experiences in thera- cial treatments for negative symptoms in randomised trials. BMJ 2011;343:d5928.
peutic groups for patients with severe mental schizophrenia: current practices and future 31 Kay SR, Fiszbein A, Opler LA: The Positive
illness: development of the Ferrara group ex- directions. Clin Psychol Rev 2013; 33: 914– and Negative Syndrome Scale (PANSS) for
periences scale (FE-GES). BMC Psychiatry 928. schizophrenia. Schizophr Bull 1987; 13: 261–
2013;13:1–9. 17 Velligan D, Alphs L: Negative symptoms in 276.
4 Kahn M, Kahn E: Group treatment assign- schizophrenia: the importance of identifica- 32 Higgins JPT, Green S: Cochrane handbook
ment for outpatients with schizophrenia. tion and treatment. Psychiatr Times 2008;25: for systematic reviews of interventions. 2013.
Community Ment Health J 1992;28:539–548. 39–45. http://www.cochrane.org/handbook.
5 DeChavez MG, Gutierrez M, Ducaju M, 18 Hunter R, Barry S: Negative symptoms and 33 Granholm E, McQuaid JR, Link PC, Fish S,
Fraile JC: Comparative study of the therapeu- psychosocial functioning in schizophrenia: Patterson T, Jeste DV: Neuropsychological
tic factors of group therapy in schizophrenic neglected but important targets for treatment. predictors of functional outcome in Cognitive
inpatients and outpatients. Group Anal 2000; Eur Psychiatry 2012;27:432–436. Behavioral Social Skills Training for older
33:251–264. 19 Buckley PF, Stahl SM: Pharmacological treat- people with schizophrenia. Schizophr Res
6 Wykes TA, Parr A, Landau S: Group treat- ment of negative symptoms of schizophrenia: 2008;100:133–143.
ment of auditory hallucinations: exploratory therapeutic opportunity or cul-de-sac? Acta 34 Granholm E, McQuaid JR, McClure FS, Aus-
study of effectiveness. Br J Psychiatry 1999; Psychiatr Scand 2007;115:93–100. lander LA, Perivoliotis D, Pedrelli P, Patter-
175:180–185. 20 Garcia-Cabeza I, Ducaju M, Chapela E, Gon- son T, Jeste D: A randomized, controlled trial
7 Wykes TA, Hayward P, Thomas N, Green N, zalez de Chavez M: Therapeutic factors in pa- of cognitive behavioral social skills training
Surguladze S, Fannon D, Landau S: What are tient groups with psychosis. Group Anal for middle-aged and older outpatients with
the effects of group cognitive behaviour ther- 2011;44:421–438. chronic schizophrenia. Am J Psychiatry 2005;
apy for voices? A randomised control trial. 21 Kanas N, Barr MA: Process and content in a 162:520–529.
Schizophr Res 2005;77:201–210. short-term inpatient schizophrenic group. 35 Granholm E, McQuaid JR, McClure FS, Link
8 Segredou I, Xenitidis K, Panagiotopoulou M, Small Group Res 1986;17:355–363. PC, Perivoliotis D, Gottlieb JD, Patterson TL,
Bochtsou V, Antoniadou O, Livaditis M: 22 Lockwood CT, Page T, Conroy-Hiller T: Ef- Jeste DV: Randomized controlled trial of cog-
Group psychosocial interventions for adults fectiveness of individual therapy and group nitive behavioral social skills training for old-
with schizophrenia and bipolar illness: the ev- therapy in the treatment of schizophrenia. JBI er people with schizophrenia: 12-month fol-
idence base in the light of publications be- Rep 2004;2:309–338. low-up. J Clin Psychiatry 2007;68:730–737.
tween 1986 and 2006. Int J Soc Psychiatry 23 Wampold BE: The Great Psychotherapy De- 36 Johnson DP, Penn DL, Bauer DJ, Meyer P, Ev-
2012;58:229–238. bate: Models, Methods, and Findings. Lon- ans E: Predictors of the therapeutic alliance in
9 Macdonald EM, Hayes RL, Baglioni AJ: The don, Routledge, 2001, vol 9, pp 98–113. group therapy for individuals with treatment-
quantity and quality of the social networks of 24 Kanas N: Group therapy with schizophrenics: resistant auditory hallucinations. Br J Clin
young people with early psychosis compared a review of controlled studies. Int J Group Psychol 2008;47:171–183.
with closely matched controls. Schizophr Res Psychother 1986;36:339–351. 37 Ojeda N, Pena J, Bengoetxea E, Garcia A, San-
2000;46:25–30. 25 Huxley NA, Rendal MM, Sederer L: Psychso- chez P, Elizagarate E, Segarra R, Ezcurra J,
10 Yalom I, Leszcz M: The Theory and Practice cial treatments in schizophrenia: a review of Gutierrez-Fraile M: Evidence of the effective-
of Group Psychotherapy B, ed 5. New York, the past 20 years. J Nerv Ment Dis 2000; 188: ness of cognitive rehabilitation in psychosis
Basic Books, 2005, pp 309. 187–201. and schizophrenia with the REHACOP pro-
11 Kopelowicz A, Liberman RP, Zarate R: Recent 26 Bernard H, Burlingame G, Flores P, Greene L, gramme. Rev Neurol 2012;54:577–586.
advances in social skills training for schizo- Joyce A, Kobos JC, Leszcz M, MacNair-Se- 38 Ojeda N, Pena J, Sanchez P, Bengoetxea E,
phrenia. Schizophr Bull 2006;32:12–23. mands RR: Clinical practice guidelines for Elizagarate E, Ezcurra J, Gutierrez-Fraile M:
12 Ulrich G, Houtmans T, Gold C: The addition- group psychotherapy. Int J Group Psychother Efficiency of cognitive rehabilitation with
al therapeutic effect of group music therapy for 2008;58:445–542. REHACOP in chronic treatment resistant
schizophrenic patients: a randomized study. 27 Heimberg RG, Dodge CS, Hope DA Kennedy Hispanic patients. Neurorehabilitation 2012;
Acta Psychiatr Scand 2007;116:362–370. CR, Zollo LJ, Becker RE: Cognitive behavioral 30:65–74.
128.103.149.52 - 5/26/2015 11:10:41 AM

248 Psychother Psychosom 2015;84:241–249 Orfanos/Banks/Priebe

DOI: 10.1159/000377705
Harvard University
Downloaded by:
39 Penn DL, Meyer PS, Evans E, Wirth R, Cai K, 49 Patterson TL, Mausbach, BT, McKibbin C, 61 Twamley EW, Vella L, Burton CZ, Heaton
Burchinal M: A randomized controlled trial Goldman S, Bucardo J, Jeste DV: Functional RK, Jeste DV: Compensatory cognitive train-
of group cognitive-behavioral therapy vs en- Adaptation Skills Training (FAST): a ran- ing for psychosis: effects in a randomized con-
hanced supportive therapy for auditory hal- domized trial of a psychosocial intervention trolled trial. J Clin Psychiatry 2012; 73: 1212–
lucinations. Schizophr Res 2009;109:52–59. for middle-aged and older patients with 1219.
40 Lecardeur L, Stip E, Giguere M, Blouin G, Ro- chronic psychotic disorders. Schizophr Res 62 Vreeland B, Minsky S, Yanos PT, Menza M,
driguez JP, Champagne-Lavau M: Effects of 2006;86:291–299. Gara M, Kim E, Toto AM, Allen L: Efficacy of
cognitive remediation therapies on psychotic 50 Rohricht F, Priebe S: Effect of body-oriented the team solutions program for educating pa-
symptoms and cognitive complaints in pa- psychological therapy on negative symptoms tients about illness management and treat-
tients with schizophrenia and related disor- in schizophrenia: a randomized controlled ment. Psychiatr Serv 2006;57:822–828.
ders: a randomized study. Schizophr Res trial. Psychol Med 2006;36:669–678. 63 Zimmer M, Duncan AV, Laitano D, Ferreira
2009;111:153–158. 51 Xiang Y, Weng Y, Li W, Gao L, Chen G, Xie EE, Belmonte-de-Abreu P: A twelve-week
41 Crawford MJ, Killaspy H, Barnes TR, Barrett L, Chang Y, Tang WK, Ungvari GS: Training randomized controlled study of the cognitive-
B, Byford S, Clayton K, Dinsmore J, Floyd S, patients with schizophrenia with the commu- behavioral Integrated Psychological Therapy
Hoadley A, Johnson T, Kalaitzaki E, King M, nity re-entry module: a controlled study. Soc program: positive effect on the social func-
Leurent B, Maratos A, O’Neill FA, Osborn Psychiatry Psychiatr Epidemiol 2006;41:464– tioning of schizophrenic patients. Rev Bras
DP, Patterson S, Soteriou T, Tyrer P, Waller 469. Psiquiatr 2007;29:140–147.
D: Group art therapy as an adjunctive treat- 52 Atkinson JM, Coia DA, Gilmour H, Harper 64 Jauhar S, McKenna PJ, Radua J, Fung E, Sal-
ment for people with schizophrenia: multi- JP: The impact of education groups for people vador R, Laws KR: Cognitive-behavioural
centre pragmatic randomised trial. BMJ 2012; with schizophrenia on social functioning and therapy for the symptoms of schizophrenia:
344:e846. quality of life. Br J Psychiatry 1996; 168: 199– systematic review and meta-analysis with ex-
42 Farreny A, Aguado J, Ochoa S, Huerta-Ramos 204. amination of potential bias. Br J Psychiatry
E, Marsa F, Lopez-Carrilero R, Carral V, Haro 53 Barrowclough C, Haddock G, Lobban F, 2014;204:20–29.
JM, Usall J: REPYFLEC cognitive remedia- Jones S, Siddle R, Roberts C, Gregg L: Group 65 Jane-Wit D, Horwitz RI, Concato J: Variation
tion group training in schizophrenia: looking cognitive-behavioural therapy for schizo- in results from randomized, controlled trials:
for an integrative approach. Schizophr Res phrenia: randomised controlled trial. Br J stochastic or systematic? J Clin Epidemiol
2012;142:137–144. Psychiatry 2006;189:527–532. 2010;63:56–63.
43 Fries A, Pfammatter M, Andres K, Brenner H: 54 Borras L, Boucherie M, Mohr S, Lecomte T, 66 Kanas N: Group Therapy for Schizophrenic
Outcome and process characteristics of a psy- Perroud N, Huguelet P: Increasing self-es- Patients. Washington, DC, American Psychi-
choeducational, coping oriented group thera- teem: efficacy of a group intervention for in- atric Association Press, 1996, vol 39, p 10.
py for schizophrenia and schizoaffective pa- dividuals with severe mental disorders. Eur 67 Wykes T, Steel C, Everitt B, Tarrier N: Cogni-
tients. Verhaltenstherapie 2003;13:237–243. Psychiatry 2009;24:307–316. tive behavior therapy for schizophrenia: effect
44 Gohar SM, Hamdi E, El Ray LA, Horan WP, 55 Daniels L: A group cognitive behavioural and sizes, clinical models, and methodological rig-
Green MF: Adapting and evaluating a social process-oriented approach to treating the so- or. Schizophr Bull 2008;34:523–537.
cognitive remediation program for schizo- cial impairment and negative symptoms as- 68 Kurtz MM, Richardson CL: Social cognitive
phrenia in Arabic. Schizophr Res 2013; 148: sociated with chronic mental illness. J Psy- training for schizophrenia: a meta-analyt-
12–17. chother Pract Res 1998;7:167–176. ic investigation of controlled research.
45 Granholm E, Holden J, Link PC, McQuaid JR, 56 Fardig R, Lewander T, Melin L, Folke F, Schizophr Bull 2012;38:1092–1104.
Jeste DV: Randomized controlled trial of cog- Fredriksson A: A randomized controlled trial 69 Wykes T, Huddy V, Cellard C, McGurk SR,
nitive behavioral social skills training for old- of the illness management and recovery pro- Czobor P: A Meta-analysis of cognitive reme-
er consumers with schizophrenia: defeatist gram for persons with schizophrenia. Psychi- diation for schizophrenia: methodology and
performance attitudes and functional out- atr Serv 2011;62:606–612. effect sizes. Am J Psychiatry 2011; 168: 472–
come. Am J Geriatr Psychiatry 2013; 21: 251– 57 Richardson P, Jones K, Evans C, Stevens P, 485.
262. Rowe A: Exploratory RCT of art therapy as an 70 Leucht S, Corves C, Arbter D, Engel RR, Li C,
46 Levine J, Barak Y, Granek I: Cognitive group adjunctive treatment in schizophrenia. J Ment Davis JM: Second-generation versus first-
therapy for paranoid schizophrenics: apply- Health 2007;16:483–491. generation anti-psychotic drugs for schizo-
ing cognitive dissonance. J Cogn Psychother 58 Roncone R, Mazza M, Frangou I, De Risio A, phrenia: a meta-analysis. Lancet 2009; 373:
1998;12:3–12. Ussoria D, Tozzini C, Casacchia M: Rehabili- 31–41.
47 Mazza M, Lucci G, Pacitti F, Pino MC, Mari- tation of theory of mind deficit in schizophre- 71 Lecomte T, Leclerg C, Wykes T: Group CBT
ano M, Casacchia M, Roncone R: Could nia: a pilot study of metacognitive strategies for early psychosis – are there still benefits one
schizophrenic subjects improve their social in group treatment. Neuropsychol Rehabil year later? Int J Group Psychother 2012; 62:
cognition abilities only with observation and 2004;14:421–435. 309–321.
imitation of social situations? Neuropsychol 59 Rus-Calafell M, Gutierrez-Maldonado J, 72 Erkkila J, Punkanen M, Fachner J, Ala-Ruona
Rehabil 2010;20:675–703. Ortega-Bravo M, Ribas-Sabate J, Caqueo- E, Pontio I, Tervaniemi M, Vanhala M, Gold
48 Ng RMK, Cheung MSL: Social skills training Urizar A: A brief cognitive-behavioural social C: Individual music therapy for depression:
in Hong Kong Chinese patients with chronic skills training for stabilised outpatients with randomised controlled trial. Br J Psychiatry
schizophrenia. Hong Kong J Psychiatry 2007; schizophrenia: a preliminary study. Schizophr 2011;199:132–139.
16:14–20. Res 2013;143:327–336.
60 Tang W, Yao X, Zheng Z: Rehabilitative effect
of music therapy for residual schizophrenia: a
one-month randomised controlled trial in
Shanghai. Br J Psychiatry 1994;24:38–44.
128.103.149.52 - 5/26/2015 11:10:41 AM

Group Psychotherapy for Schizophrenia Psychother Psychosom 2015;84:241–249 249

DOI: 10.1159/000377705
Harvard University
Downloaded by: