Articles

Clinical effectiveness and acceptability of structured group

psychoeducation versus optimised unstructured peer
support for patients with remitted bipolar disorder
(PARADES): a pragmatic, multicentre, observer-blind,
randomised controlled superiority trial
Richard Morriss, Fiona Lobban, Lisa Riste, Linda Davies, Fiona Holland, Rita Long, Georgia Lykomitrou, Sarah Peters, Christopher Roberts,
Heather Robinson, Steven Jones, on behalf of the NIHR PARADES Psychoeducation Study Group

Summary
Background Group psychoeducation is a low-cost National Institute for Health and Care Excellence-recommended Lancet Psychiatry 2016;
treatment for bipolar disorder. However, the clinical effectiveness and acceptability of this intervention are unclear 3: 1029–38

compared with unstructured peer support matched for delivery and aim of treatment, and for previous bipolar Published Online
September 26, 2016
history. We aimed to assess the clinical effectiveness and acceptability of structured group psychoeducation versus
http://dx.doi.org/10.1016/
optimised unstructured peer support for patients with remitted bipolar disorder. S2215-0366(16)30302-9
See Comment page 1000
Methods We did this pragmatic, multicentre, parallel-group, observer-blind, randomised controlled superiority Department of Psychiatry and
trial at eight community sites in two regions in England. Participants aged 18 years or older with bipolar disorder Applied Psychology, University
and no episode in the preceding 4 weeks were recruited via self-referral or secondary care referral. Participants of Nottingham, Nottingham,
were individually randomly assigned (1:1), via a computer-generated stochastic allocation sequence, to attend UK (Prof R Morriss MD,
G Lykomitrou MA); Spectrum
21 2-h weekly sessions of either structured group psychoeducation or optimised unstructured peer support. Centre, University of Lancaster,
Randomisation was minimised by number of previous episodes (one to seven, eight to 19, or ≥20) and stratified by Lancaster, UK
clinical site. Outcome assessors were masked to group allocation. The primary outcome was time from (Prof F Lobban PhD, R Long,
Prof S Jones PhD,
randomisation to next bipolar episode, with planned moderator analysis of number of previous bipolar episodes
H Robinson PhD); and School of
and qualitative interview of participant experience. We did analysis by intention to treat. This trial is registered Psychological Sciences
with the International Standard Randomised Controlled Trial registry, number ISRCTN62761948. (L Riste PhD, S Peters PhD) and
Institute of Population Health
(Prof L Davies MSc,
Findings Between Sept 28, 2009, and Jan 9, 2012, we randomly assigned 304 participants to receive psychoeducation
F Holland MSc,
(n=153) or peer support (n=151); all (100%) participants had complete primary outcome data. Attendance at Prof C Roberts PhD), University
psychoeducation groups was higher than at peer-support groups (median 14 sessions [IQR three to 18] vs nine of Manchester, Manchester, UK
sessions [two to 17]; p=0·026). At 96 weeks, 89 (58%) participants in the psychoeducation group had experienced a Correspondence to:
next bipolar episode compared with 98 (65%) participants in the peer-support group; time to next bipolar episode Prof Richard Morriss, Institute of
Mental Health, University of
did not differ between groups (hazard ratio [HR] 0·83, 95% CI 0·62–1·11; p=0·217). Planned moderator analysis
Nottingham, Nottingham
showed that psychoeducation was most beneficial in participants with few (one to seven) previous bipolar episodes NG7 2TU, UK
(χ²; HR 0·28, 95% CI 0·12–0·68; p=0·034). Four (1%) participants (one in the psychoeducation group and three in richard.morriss@nottingham.
the peer-support group) died during follow-up; these deaths were deemed unrelated to the study interventions ac.uk
or procedures.

Interpretation Structured group psychoeducation was no more clinically effective than similarly intensive
unstructured peer support, but was more acceptable and improved outcome in participants with fewer previous
bipolar episodes. Optimum provision of structured psychological interventions, such as group psychoeducation,
early in the course of bipolar disorder might have important benefits on the course of illness, and merits
further research.

Funding National Institute for Health Research.

Introduction the UK, such information and support is widely

Bipolar disorder is a common relapsing lifelong mental
health disorder presenting in adolescence or early
adulthood.1 Provision of information and emotional
support is a National Institute for Health and Care
Excellence (NICE)-supported key recommendation and
quality standard for all mental health service users.2 In
available to people with bipolar disorder through For more on Bipolar UK
unstructured peer-support groups run by both the support groups see
http://www.bipolaruk.org/find-
National Health Service (NHS) and the third sector (eg, a-support-group
through Bipolar UK, which includes a national
network of more than 130 support groups). Across
Europe, people can access information and support

www.thelancet.com/psychiatry Vol 3 November 2016 1029

 Articles
Research in context
Evidence before this study
Morriss and colleagues’ 2007 meta-analysis of all randomised
controlled trials of psychological treatment involving early
warning signs in bipolar disorder, a core component of
psychoeducation, identified only two trials of group
psychoeducation versus group support from the same centre.
These trials found that 38% of participants allocated to group
psychoeducation and 60% of those allocated to group support
experienced a bipolar episode in the subsequent 12 months.
A further meta-analysis by Bond and Anderson in 2015,
restricted to group psychoeducation versus treatment as usual
or control psychological treatment, identified ten randomised
controlled trials, but only five trials reported bipolar episodes at
12 months’ follow up. Group psychoeducation was associated
with fewer bipolar episodes at 12 months compared with
control treatment (odds ratio 2·80, 95% CI 1·63–4·82), but
effectiveness was uncertain once the two original trials were
excluded. As well as selective reporting, the trials were mostly
small, single centre, and did not control for the aim of
treatment or the natural history of bipolar disorder. We did an
update of this meta-analysis from June 1, 2013, to July 31, 2016,
with the same search terms as used by Bond and Anderson. Two
further randomised trials of group psychoeducation were
identified, each of which had a follow up of only 6 months and
through structured group psychoeducation in mental
health services.3,4
In 2003, findings from two randomised controlled
trials5,6 of people with bipolar disorder in Barcelona
showed clinically important differences in time to
all types of bipolar relapse at months 12 and 24 in
participants who underwent 21-session structured
manualised group psychoeducation compared with those
who attended attentional control support groups. Group
psychoeducation is a key recommendation of NICE and
the Canadian Network for Mood and Anxiety Treatments
for bipolar disorder and a NICE development quality
standard.1,7 The intervention is low cost, efficient, and
easy to organise and deliver because ten to 18 participants
can be treated at a time and, although therapists require
supervision, they do not need extensive training unlike
for many other psychological treatments. A meta-analysis8
of the effectiveness of psychoeducation on bipolar relapse
noted that the original Barcelona trials seemed to be
See Online for appendix outliers. The evidence was reported as weak or very weak,
with few preregistered large multicentre blinded
randomised trials done independently of the developers
of the intervention.1,8,9 Moreover, few studies controlled
for the natural history of bipolar disorder—eg, that
people with 20 or more previous bipolar episodes relapse
three times more frequently than do those with one to
seven previous episodes.10
We did the PARADES trial to independently examine
the clinical effectiveness of structured group
1030
therefore provided no additional information to previous
meta-analyses.
Added value of this study
In the present large, multicentre, 2 year randomised
controlled trial that controlled for the natural history of
bipolar disorder and the delivery and aim of treatment, group
psychoeducation was no more clinically effective than
unstructured peer support for time to the next bipolar
episode. However, group psychoeducation might have
specific benefits, especially in people early in the course of
bipolar disorder, on acceptability, time to next mania episode,
and interpersonal function. Both group psychoeducation and
peer support increase specific and individualised knowledge
about bipolar disorder.
Implications of all the available evidence
Group psychoeducation is a low-cost psychological
intervention for bipolar disorder, with a few specific
clinical benefits, especially for people early in their illness
course, if the focus of care is on improving self-management,
interpersonal function, and support, and preventing future
mania relapse. However, the intervention does not reduce
overall bipolar relapse in people with long-established bipolar
disorder, nor does it improve performance aspects of function.
psychoeducation versus unstructured peer support for
time to next bipolar episode, including the moderating
effects of number of previous bipolar episodes, in
addition to secondary outcomes in groups matched for
the duration, delivery, and aim of treatment, and
previous bipolar history. We also assessed the
experience, acceptability, and subjective value of both
treatments on the basis of systematic qualitative
enquiry and attendance at group sessions.
Methods
Study design and participants
We did this pragmatic, multicentre, parallel-group,
observer-blind, randomised controlled superiority trial
at eight community sites in two regions in England (the
North West and East Midlands). Participants were
recruited by research assistants at each site via self-
referral or referral to secondary mental health care. At
three sites recruitment was done at two separate times,
so occurred in 11 waves (appendix p 12).
The recruitment strategy was deliberately broad to
ensure that the study sample was representative of a
diversity of people with bipolar disorder. Community
mental health teams at NHS trusts at each site were
encouraged to invite potentially relevant participants.
The study was also promoted at a primary care level,
with local family doctors asked to display posters about
the trial, and through service user-run local bipolar
disorder groups, national bipolar disorder publications,
www.thelancet.com/psychiatry Vol 3 November 2016

and the general media, allowing people to self-refer.
Because prevention of relapse is the main aim of the
study intervention, the target population was patients
with bipolar I or II affective disorder at increased risk of
further relapse.
We included patients if they were aged 18 years or
older, had a Structured Clinical Interview for DSM-IV
(SCID)-verified diagnosis of primary bipolar I or II
disorder,11,12 were at increased risk of relapse (defined as
occurrence of at least one episode in the past
24 months), and provided written informed consent.
We excluded participants who had experienced a manic,
hypomanic, mixed affective, or major depressive
episode currently or within the previous 4 weeks; had
current suicide plans or high suicide intent; were
unable or unwilling to provide written informed
consent; or were unable to communicate in written and
verbal English to a sufficient level to consent, complete
the measures, and take part in the groups.
Ethics approval was obtained from a national ethics
committee in Nottingham, UK. The study protocol has
been previously published.13
Randomisation and masking
Consecutive participants were individually randomly
assigned (1:1) to receive either structured group
psychoeducation or optimised unstructured group
support. Randomisation was done by a clinical trials
unit using a computer-generated stochastic allocation
sequence, with stratification by clinical site and
minimisation by number of previous bipolar episodes
within three categories (one to seven, eight to 19, or
≥20, as determined by SCID-DSM-IV criteria for past
mania, hypomania, mixed affective, or major depression
episodes). At each site, recruitment continued until a
minimum of 20 participants and a maximum of
36 participants were enrolled per wave (appendix p 12).
To ensure masked allocation, the research assistants
sent participant details to the trial coordination team
(trial administrator and trial coordinator) at a separate
site, who in turn passed the participant information to
the clinical trials unit for randomisation. The clinical
trials unit reported the randomisation allocation to the
trial coordination team, which directly informed
each participant and the lead health professional
running the treatment group to which the participant
was allocated.
To maintain concealment, research assistants were
based separately from the coordination team and all
treatment groups; treatment groups were run
identically to mask the day on which each group was
run at each site (ie, over the same time period,
frequency, duration of sessions, base, both groups
received a manual, and both were led by the same
health professional); and all follow-up data obtained by
self-complete questionnaires that could unmask the
research assistant were returned in a sealed envelope to
www.thelancet.com/psychiatry Vol 3 November 2016
Articles
the assistant doing the assessment and conveyed to the
trial office for opening and subsequent data entry by
another person.
Cases of unmasking of research assistants were
recorded and, if they occurred, all subsequent follow-up
assessments were done by another research assistant
who was masked to treatment allocation.
Procedures
At baseline, before attending either group, participants
were asked whether they had a preference for structured
group psychoeducation, optimised unstructured peer
support, or no preference. In both groups, participants
were told when they consented to the study and at the
first and subsequent sessions that the purpose of this
intervention was to share experiences to help manage
bipolar disorder by use of information given by the
group facilitators, their own experience, and the
collective experience of the group. The groups differed
only in structure, nature of delivery within the sessions,
and choice and type of content. Participants in the
psychoeducation group followed a curriculum
developed in Spain,7,8 but contextualised to English
current practice by the research team and a panel of
service users recruited for the purpose (appendix p 2).
Participants in the peer-support group set their own
agenda and chose the content of their own programme
(appendix p 3).
Both groups were run by three facilitators,
comprising two health professionals (usually one
experienced and one trainee facilitator) and a service
user facilitator with a diagnosis of bipolar disorder.
Facilitators were trained for the purpose and supervised
by a psychiatrist (RM) or clinical psychologist (FL or
SJ) experienced in delivering psychological treatment
for bipolar disorder. However, the research team had
neither devised group psychoeducation or optimised
unstructured peer support, nor gained from favouring
either intervention, allowing for a completely
independent trial. Service user facilitators were also
offered additional peer support. Both programmes
comprised 21 sessions, delivered once a week for 2 h,
spread over a maximum of 26 weeks.5,6 The group
sessions comprised a closed group starting with a
minimum of ten and a maximum of 18 participants to
capture a variety of service user experience about the
topics of every session. The sessions were held at the
same community-based site away from hospital (eg,
day centre for both groups contemporaneously on
different afternoons of the week). A manual was
produced for both psychoeducation14 (adapted by the
authors) and peer support (available from authors on
request) on the aims and conduct of each group.
Participants undergoing group psychoeducation were
encouraged to share their staying-well plan with their
health professionals (eg, community psychiatric nurse,
psychiatrist, general practitioner) and other people
1031
 Articles
who were personally important to them. Participants
were discouraged from sharing any information about
the content of the group with other service users
with bipolar disorder until their follow-up was
complete. Participants in both groups received the
study intervention in addition to their usual treatment,
usually medication; they were discouraged from
attending any other course of group, family, or
individual psychological treatment for bipolar disorder
at the same time as attending the study groups,
otherwise, treatment as usual was unconstrained and
recorded by interview and from case notes.
The psychoeducation group was run as a collaborative
workshop with a brief taught introduction of the topic
for the session, and the rest of the work taking the form
of active interaction using the collective experience of
participants.5,6 Embedded in the psychoeducation
programme is the acquisition of specific skills by each
individual, including life charting, recognition of early
warning signs, problem solving and other forms of
coping, sleep hygiene, and care planning, in addition to
general skills of actively participating and working
collaboratively in groups. Participants were encouraged
and supported to develop a plan that fitted their
personalised goals for recovery and had two elements:
actions taken every day to stay well, such as taking
medication and keeping a regular early morning
routine; and actions they would take if they started to
become unwell with mania or depression.
In the peer-support group, participants collectively
decided on an agenda for discussion at each session.
The three facilitators were present to ease discussion;
encourage participation; prevent unhelpful group
behaviour, such as bullying or scapegoating; prevent
factual misinformation; and, if directly asked, to resolve
factual uncertainty. Compared with peer support
delivered routinely by Bipolar UK, our unstructured
peer-support strategy was optimised by being closed to
new participants so that participants could get to know
each other well, and by facilitation by a peer service
user and two health professionals, enabling delivery of
expert information from health professionals if
participants requested it.
Treatment fidelity was maintained by training and
supervision, written record of the content and delivery
of the sessions completed by therapists and supervisors,
and qualitative interviews with participants. Sessions
were not audiotaped or videotaped because some
participants reported that they would find such
recording intrusive and unacceptable.
Outcomes
The primary outcome measure was time to next bipolar
episode13 recorded by research assistants at each site.
This measure was based on the SCID Longitudinal
Interval Follow-up Evaluation (LIFE),12,15 which was
done every 16 weeks for 96 weeks. We calculated time
1032
from randomisation to the first week of recurrence of
an episode of mania, hypomania, a mixed episode, or
major depression that lasted 2 consecutive weeks,
satisfying DSM-IV criteria. When a follow-up interview
was not done, a bipolar episode was recorded when
there was a description of symptoms of a manic or
depressive episode, in primary care or secondary care
mental health records, with symptoms of sufficient
duration to meet episode criteria,11,16 and when there
was a change of medication, care setting (inpatient or
crisis team), or urgency of reviewal because of
these symptoms.
Secondary outcome measures were time to next
mania-type episode (mania, hypomania, or mixed
affective episode) and time to next depressive episode;12,16
assessment of mean weekly symptoms of mania-type
symptoms and depression symptoms using the LIFE;12,16
assessment of function with the Social Adjustment
Scale (SAS)17 and Social and Occupational Functioning
Assessment Scale (SOFAS);18 observer-rated and self-
rated measures of mood (17 item GRID-Hamilton
Depression Rating Scale,19 Bech-Rafaelsen Mania
Scale,20 Hospital Anxiety and Depression Scale;21 and
self-rated overall mental and physical health (Short
Form [SF]-12 mental and physical component scores).22
At each assessment, suicide, self-neglect, and risk to
and from other people were recorded in addition to
other assessments. We recorded the type and amount
of medication rather than medication adherence as a
secondary outcome in our protocol.13 A nested
qualitative study was done to explore the experiences of
group participants and reasons for dropout from group
treatment. A maximum variance sample of participants
(number of treatment sessions attended, sex, age,
number of previous bipolar episodes, geographical
location of group) received a semi-structured digitally
recorded interview on completion of group treatment at
that site.
The SF-12 was added to the measures outlined in the
protocol to measure physical health, with approval of the
study trial steering committee. The economic analysis,
medication adherence, and theoretical psychological
measures of process will be reported separately.
Statistical analysis
On the basis of the first two randomised controlled
trials of group psychoeducation versus group support
with at least 12 months of follow up,5,6 we estimated a
differential treatment effect of 0·22 (60% recurrence in
the peer-support group and 38% in the psychoeducation
group at 12 months). Because our study involved a
group-administered intervention, we adjusted the
sample size for the clustering effect, assuming a mean
group size of 18 at randomisation and an intracluster
correlation for group therapy of 0·05. Under these
assumptions, a study with 360 participants (ten groups
per arm) would give more than 80% power, assuming a
www.thelancet.com/psychiatry Vol 3 November 2016

15% loss to follow-up. During the conduct of the study,
the average group size was 14. With agreement of the
independent trial data monitoring committee, we
increased the numbers of groups per arm and
adjusted the sample size, since a smaller group size
reduces the effect of clustering. In the assumption of a
mean group size of 14, 308 participants (11 groups per
arm) would provide 82% power at a two-sided type I
error rate of 5%.
We used Kaplan–Meier curves to visualise time-to-
event data. We planned to use a Cox model with
robust standard errors to account for the effect of
therapy group in the primary analysis; however, there
was no clustering effect by group for time to next
bipolar episode, time to next mania episode, or time
to next depression episode, so we present results
from the standard Cox model. The treatment effect
(psychoeducation vs peer support) was adjusted for sex,
number of previous bipolar episodes, and recruitment
wave. The proportional hazards assumption was
checked by use of log–log plots by arm alone and
with additional covariates in the model. We examined
two prespecified treatment moderators for the
primary outcome: number of previous episodes (one to
seven, eight to 19, ≥20 episodes minimised during
randomisation) and participant treatment preference
at baseline.
For continuous longitudinal data, we used a linear
random-effects model, incorporating time as a
continuous variable to compare the two interventions.
We included random effects to account for variation
between patients in the intercept and the gradient of
the patient-specific lines. We additionally fitted models
that also included random effects for wave and arm
(nested within wave) to account for group clustering
effects. We included fixed covariates to model systematic
differences due to treatment, assessment time, and
participant characteristics. Time since randomisation
was calculated in months and was centred by
subtraction of the overall (grand) mean of assessment
times for each outcome measure. To enable all
participants with outcome data to be included in the
linear random-effects analyses, we imputed missing
baseline response values using simple (deterministic)
imputation.23 We estimated a time–treatment
interaction (ie, difference in slopes) and a main effect.
All analyses were by intention to treat. Safety results
were reviewed by an independent data monitoring
committee. The appendix (pp 4, 5) provides further
details about the statistical analysis method, and the
statistical analysis plan is available online. We did
analyses with Stata Release 13.
We analysed qualitative interviews with thematic
analysis,24 taking an inductive and emergent approach
based on participants’ experiences of the groups.
Coding was done by a multidisciplinary panel (health
and clinical psychology, psychiatry, service user)
www.thelancet.com/psychiatry Vol 3 November 2016
Articles
and subsequent interviews sought evidence to
refute emerging themes. Themes were continuously
compared against the data by use of a constant
comparative approach. Interviews were done until
themes were saturated. This trial is registered with the
International Standard Randomised Controlled Trial
registry, number ISRCTN62761948.
Role of the funding source
The funder of the study had no role in study design, data
collection, data analysis, data interpretation, or writing of
the report. The corresponding author had full access to For the statistical analysis plan
all the data in the study and had final responsibility for see https://dx.doi.org/10.6084/
m9.figshare.3205738
the decision to submit for publication.
519 people referred
160 not assessed
80 withdrawn before consent
64 with no known reason
12 because of work or family commitments
4 had already received psychoeducation
1 took too long to respond
61 were not contactable
4 were too unwell
15 were excluded or unsuitable
359 assessed for eligibility
55 excluded
18 did not meet inclusion criteria
27 withdrew consent
4 were not contactable
6 for other reasons
1 interview terminated (patient too
distressed)
1 unable to recall
4 could not make group sessions
304 randomly assigned
151 allocated to receive unstructured peer 153 allocated to receive structured group
support psychoeducation
129 participated in the intervention 133 participated in the intervention
22 did not participate in the intervention 20 did not participate in the intervention
132 completed 16 weeks of follow-up 124 completed 16 weeks of follow-up
125 completed 32 weeks of follow-up 118 completed 32 weeks of follow-up
117 completed 48 weeks of follow-up 110 completed 48 weeks of follow-up
104 completed 64 weeks of follow-up
111 completed 64 weeks of follow-up
113 completed 80 weeks of follow-up 98 completed 80 weeks of follow-up
109 completed 96 weeks of follow-up 95 completed 96 weeks of follow-up
153 case notes reviewed 151 case notes reviewed
Figure 1: Trial profile
1033
 Articles

Results
Psychoeducation Peer-support Between Sept 28, 2009, and Jan 9, 2012, we randomly
group (n=153) group (n=151) assigned 304 participants to receive structured group
Age (years) 44·2 (11·1) 46·5 (11·4) psychoeducation (n=153) or optimised unstructured
Sex peer support (n=151; figure 1). When asked about
Female 92 (60%) 85 (56%) treatment preference before randomisation, all
Male 61 (40%) 66 (44%)
participants indicated that they were prepared to attend
Number of previous bipolar
either group; however, there was a slight preference for
episodes psychoeducation over peer support if participants were
1–7 21 (14%) 18 (12%) given a choice rather than randomisation (table 1). Data
8–19 50 (33%) 45 (30%) for all bipolar, mania, and depression relapses were
≥20 82 (54%) 88 (58%) completed with primary care and secondary care
Bipolar I 114 (75%) 129 (85%) records. Ten unmaskings were reported (n=3 during
Marital status psychoeducation and n=7 during peer support) whereby
Married or living as married 63 (41%) 46 (30%) the participant divulged their group allocation to their
Never married 46 (30%) 53 (35%) research assistant; subsequent follow-up was done by a
Divorced, separated, or 44 (29%) 52 (34%)
different assistant masked to allocation. Six raters made
widowed 292 inter-rater reliability assessments of 17 (6%)
One or more children 90 (59%) 93 (62%) participants across all baseline and follow-up
Employment timepoints. The intracluster correlation was 0·71
Full-time or part-time work 45 (29%) 37 (24%) (95% CI 0·64–0·85) for depression scores and 0·57
Unemployed, sickness leave, or 108 (71%) 114 (76%) (0·40–0·83) for mania scores, showing a high level of
retired inter-rater reliability for depression symptoms and a
White British ethnic origin 140 (92%) 138 (91%) moderate level of agreement for mania symptoms, with
Region no evidence of changes in inter-rater reliability over
North West 86 (56%) 86 (57%) time. Therapist and supervision records and qualitative
East Midlands 67 (44%) 65 (43%) interviews confirmed that psychoeducation sessions
Group preference followed the manual and were therapist led, whereas
Psychoeducation 55 (36%) 54 (36%) peer-support sessions followed topics decided by the
No preference 66 (43%) 57 (38%) participants and were peer led.
Peer support 29 (19%) 35 (23%) Baseline characteristics were similar between groups
Missing 3 (2%) 5 (3%) (table 1). More than half the participants had 20 or more
Psychosis, lifetime 97 (63%) 114 (75%) previous bipolar episodes and most were taking mood
Any anxiety disorder stabilising medication and either antipsychotic drugs
Lifetime 87 (57%) 75 (50%) or antidepressants (table 1). Participants scored below
Past month 68 (44%) 57 (38%)
the clinical threshold for significant depression, mania,
Any alcohol abuse or dependence
or anxiety symptoms, and showed mild to moderate
Lifetime 56 (37%) 61 (40%)
impairment in social adjustment at baseline (appendix
Past month 4 (3%) 12 (8%)
pp 4–6). Demographic and clinical characteristics did
not differ significantly between participants who self-
Any drug abuse or dependence
referred or were referred by secondary care, nor
Lifetime 10 (7%) 21 (14%)
between participants from different sites.
Past month 2 (1%) 2 (1%)
Table 2 shows attendance at treatment groups.
Borderline/antisocial personality 12 (8%) 13 (9%)
disorder Overall, attendance at psychoeducation groups was
Medication
Mood stabiliser 108 (71%) 118 (78%) Psychoeducation group Peer-support group
Antipsychotic 83 (54%) 84 (56%) (n=153) (n=151)

Antidepressant 71 (46%) 71 (47%) n (%) Cumulative n (%) Cumulative

Hypnotic or anti-anxiety 21 (14%) 19 (13%) proportion (%) proportion (%)
Medication equivalents (mg/day) 0 20 (13%) ·· 22 (15%) ··
Chlorpromazine 188·2 182·8 1–5 25 (16%) 87% 41 (27%) 85%
Imipramine 83·6 80·4 6–10 12 (8%) 71% 17 (11%) 58%
Diazepam 2·25 2·43
11–15 29 (19%) 63% 25 (17%) 47%
Data are mean (SD) or n (%), unless otherwise specified. 16–21 67 (44%) 44% 46 (30%) 30%

Table 1: Baseline characteristics Table 2: Number of psychoeducation and peer-support sessions attended

1034 www.thelancet.com/psychiatry Vol 3 November 2016

 Articles

greater than at peer-support groups (median 14 sessions 100 Psychoeducation

[IQR three to 18] vs nine sessions [two to 17]; Peer support
Mann–Whitney test Z score 2·23; p=0·026).

Estimated proportion (%)

80
By 96 weeks, 89 (58%) participants in the
psychoeducation group had experienced a next bipolar 60
episode versus 98 (65%) participants in the peer-
40
support group (figure 2). The median time from
baseline to next bipolar episode was 67 weeks (95% CI 20
37–91) in the psychoeducation group versus 48 weeks
HR 0·83, 95% CI 0·62–1·11; p=0·217
(31–66) in the peer-support group. There was no 0
evidence of an intervention effect after adjustment 0 20 40 60 80 100
Time (weeks)
for prespecified covariates in a Cox proportional Number at risk
hazards model (hazard ratio [HR] 0·83, 95% CI Psychoeducation 153 111 87 79 68 0
Peer support 151 102 81 65 56 0
0·62–1·11; p=0·217; figure 2). Planned moderator
analysis based on an interaction between treatment Figure 2: Kaplan–Meier estimates of time to next depressive or manic bipolar
group and number of previous bipolar episodes showed episode
HR=hazard ratio.
that psychoeducation delayed time to next bipolar
episode compared with peer support in participants
100 Psychoeducation
with one to seven previous episodes (χ² 6·80; HR 0·28, Peer support
95% CI 0·12 to 0·68; p=0·034; appendix p 13). There 80
Estimated proportion (%)

was no difference between groups for participants with

eight to 19 previous bipolar episodes (HR 0·86, 95% CI 60
0·50–1·49) or 20 or more episodes (1·01, 0·70–1·46).
Participant treatment preference before randomisation 40
had no moderating effect on time to next bipolar
20
episode (χ² 1·95; p=0·38).
The proportion of participants with a next mania- HR 0·66, 95% CI 0·44–1·00; p=0·049
0
type episode was lower in the psychoeducation group 0 20 40 60 80 100
(25% [n=39]) than in the peer-support group Time (weeks)
Number at risk
(35% [n=53]). In both treatment groups, insufficient Psychoeducation 153 141 131 122 113 0
numbers of participants relapsed to report the Peer support 151 130 119 107 99 0
median time to relapse with 95% CIs. The 25th
percentile of the time to episode was longer in the Figure 3: Kaplan–Meier estimates of time to next manic bipolar episode
HR=hazard ratio.
psychoeducation group (90 weeks) than in the peer-
support group (53 weeks), but the upper 95% CI could
100 Psychoeducation
not be defined. After adjustment for prespecified Peer support
covariates in a Cox proportional hazards model, there
Estimated proportion (%)

80
was weak evidence that time to mania-type episode
was longer in participants in the psychoeducation 60
group than in those in the peer-support group
(figure 3). The difference between groups in time to 40

mania episode appeared at around session 15 when

20
early warning signs of mania were being discussed
HR 0·96, 95% CI 0·70–1·31; p=0·784
during psychoeducation. 0
The proportion of participants with a depressive 0 20 40 60 80 100
episode was 52% (n=80) in the psychoeducation group Time (weeks)
Number at risk
compared with 54% (n=82) in the peer-support Psychoeducation 153 118 93 85 76 0
group. Again, insufficient participants relapsed to Peer support 151 113 96 78 70 0
report the median time to relapse with 95% CI. The Figure 4: Kaplan–Meier estimates of time to next depressive bipolar episode
25th percentile for time from baseline to next HR=hazard ratio.
depressive episode was 22 weeks (95% CI 15–29) in the
psychoeducation group and 19 weeks (15–28) in the psychoeducation had faster improvement in the SAS
peer-support group. Time to next depressive episode interpersonal domain than did those undergoing peer
did not differ significantly between groups (figure 4). support (difference in gradient –0·017, 95% CI
There were no significant intervention effects on –0·030 to –0·004; p=0·012; appendix p 8). There was no
symptoms or self-rated mental or physical health treatment by time interaction for the remaining SOFAS
outcomes (appendix pp 6, 7). Participants undergoing and SAS (overall, friction, and dependency) functional

www.thelancet.com/psychiatry Vol 3 November 2016 1035

 Articles
outcomes (appendix p 7). There were no clinically
important differences in medication use between
groups (data not shown).
Two themes emerged from qualitative study of the
value of both groups: “increased knowledge”, in general
about bipolar disorder and specifically applied to the
participant as an individual, and “people like me”
tackling isolation and stigma and sharing similar
experiences of having bipolar disorder (appendix
pp 9–11). Some participants attributed dropping out of
treatment to the lack of structure of the peer-support
groups (appendix p 10).
Four (1%) participants (one undergoing psycho-
education and three undergoing peer support) died
during follow-up; three from natural causes and one
from open verdict. The independent trial steering and
data monitoring and ethics committees deemed these
deaths unrelated to the interventions or procedures in
the trial.
Discussion
Our findings show that the primary clinical outcome of
time to next bipolar episode did not differ significantly
between patients undergoing group psychoeducation
for bipolar disorder and those undergoing peer support
given adjunctively to medication. Both groups provided
general information about bipolar disorder tailored
to the participant, and provided emotional support
beyond general written information.1,2 However, we
noted evidence of some superiority of psychoeducation
over peer support, but not for peer support over
psychoeducation. People with seven or fewer previous
bipolar episodes had a greater delay in time to next
bipolar episode than those with eight or more previous
episodes, whereas time to mania was delayed and
interpersonal function was improved in all participants.
Attendance at psychoeducation groups was better
than at peer-support groups, and the lack of structure
of peer-support groups was seen as a reason for
participant dropout.
The strengths of our study are its large size, multi-
centre design, independence of the research team,
pre-registration, allocation concealment, complete data
for the primary outcome at 2 years, careful matching
of therapists and duration of treatment, minimisation
of patients across treatment arms for the natural history
of the disorder, few breaks in masking, training and
supervision to ensure fidelity to treatment, and
full reporting of clinical outcomes identified by NICE
as being relevant to patients and clinicians.1 Broad
eligibility criteria and 22 groups at eight centres
increase the generalisability of our findings, but,
compared with the Barcelona trials,5,6 we required only
4 weeks of euthymia compared with 6 months, our
population had more comorbidity, and we used less
experienced therapists, all of which might have reduced
the effect size of the psychoeducation strategy.
1036
Limitations of the study are the low rate of completion
of self-rated symptomatic and functional outcomes, the
moderate reliability of the assessment of mania
symptoms, no formal rating of blinding, and the
absence of recorded treatment sessions to ensure
fidelity to treatment. We do not report overall time
spent in relapse, but this will be separately reported in
an economic analysis. The trial design did not include a
treatment-as-usual control group with minimal
information giving; however, such practice is not
supported by NICE.1,2 Because the peer-support
intervention included some approaches that might be
components of effective psychological treatments
for bipolar disorder, such as problem solving,1,25
psychoeducation might have been more effective than
treatment as usual.
The present results confirm those from a meta-analysis
of randomised controlled trials of psychoeducation,8
which suggested that the first randomised controlled
trials of group psychoeducation versus group support5,6
might be outliers in showing a much greater treatment
effect versus control treatment than subsequent
randomised trials. When methodological issues from
previous trials are addressed, there is less evidence of
clinical superiority of structured group psychoeducation
versus unstructured group support. Our findings
showing that psychoeducation might have been more
effective than peer support against time to next
depression episode and time to next bipolar episode in
people with few previous bipolar episodes confirm
findings from two post-hoc analyses of randomised
trials16,25 and two trials of psychological treatment
confined to people with early onset bipolar disorder.26,27
Previous trials5,25 of group psychoeducation versus group
support might have shown a greater effect size on time to
the next bipolar episode than our study because more
partici-pants were recruited early in the course of their
illness. Psychoeducation might be more effective if early
warning-sign interventions are delivered early in the
course of the group intervention, because half the
participants had dropped out by week 14 when
these effective techniques28,29 were discussed and
differences in time to mania relapse between the two
groups start to emerge.
Group psychoeducation provides both general and
tailored information and support from both health
professional and service user perspectives at a relatively
low cost; as such, this approach might have merit
in clinical practice, but not to the exclusion of other
approaches that could deliver the same quality
of information and support alongside medication.1
However, the optimised peer support provided in this
study is unavailable routinely in the UK in the
precise form delivered in our study. Furthermore the
effects of psychoeducation on mania and interpersonal
function might provide additional benefits to service
users in work and family life, and to clinical
www.thelancet.com/psychiatry Vol 3 November 2016

services through reduced involvement of inpatient and
crisis teams.1,29
Although meta-analysis of previous randomised
controlled trials suggests that group psychoeducation
might increase time to relapse compared with treatment
as usual, these trials were of low or very low quality, so
some doubt remains that requires further research.
Our results and the results of other psychological
interventions16,26,27,30 suggest that the optimum provision
of structured psychological interventions such as
psychoeducation early in the course of bipolar disorder
could have important benefits on the course of illness,
and merits further research.
Contributors
RM, FL, LD, SP, CR, and SJ obtained funding for the study. RM led
the study in the East Midlands and FL led the study in the North West.
LR was the trial coordinator. RL was the service user lead and SP was
the qualitative analysis lead. CR led and supervised the statistical
analysis. GL and HR led the data collection for the study. FH did the
statistical analysis. RM, FL, and SJ supervised the interventions, and
supervised data collection with LR and SP. RM led the writing of the
manuscript. All authors designed, interpreted data, and wrote and
approved the final version of the manuscript.
Declaration of interests
We declare no competing interests.
Acknowledgments
This paper presents independent research funded by the National
Institute for Health Research (NIHR) under its Programme Grants for
Applied Research Programme (Reference Number RP-PG-0407-10389).
RM was partly funded by the NIHR Collaboration for Leadership in
Applied Health Research and Care East Midlands. We received further
support from primary care trusts, mental health trusts, the Mental
Health Research Network, and Comprehensive Local Research
Networks in the East Midlands and North West England. The views
expressed by the authors do not necessarily reflect those of the NIHR,
the National Health Service (NHS), or the Department of Health in
England. We thank Bipolar UK, Mind, and Mood Swings for helping to
publicise the study; Manchester Health and Social Care NHS Trust for
hosting the PARADES programme; the University of Nottingham for
providing sponsorship and clinical trials unit support; and the
Spectrum Centre, University of Lancaster, and University of Manchester
for additional support. In particular we acknowledge the contributions
of the NIHR PARADES Psychoeducation Study Group, without whom
the study would not have been possible: Rebecca Anderson,
Marcus Barker, Lucy Bateman, Alison Beck, Nancy Black, Lisa Brown,
Phillip Byrne, Elizabeth Camacho, Jane Fisher, Lynda Fretwell,
Lorraine Getten (now Ward), Kay Hampshire (now Glint),
Paul Hammersley, Claire Hilton, Jelena Jovanoska, Steven Kendall,
Dawn Knowles, Rachel Lambourne, Brian Langshaw, Susan Lomas,
Natasha Lyon, Deborah Mayes, Elly McGrath, Sian Newman,
Kirsten Nokling, Dionysios Ntais, Puru Pathy, Kathryn Reeveley,
Kirsty Stevenson, Katherine Taylor, Karthik Thangavelu, Elizabeth Tyler,
and Emma Weymouth. We also thank the participants in the study.
References
1 National Collaborating Centre for Mental Health. Bipolar
disorder: the NICE guideline on the assessment and
management of bipolar disorder in adults, children and young
people in primary and secondary care. Updated edition. CG185.
Leicester and London: British Psychological Society and
Gaskell, 2014.
2 National Institute for Health and Care Excellence. Service user
experience in adult mental health: improving the experience of
care for people using adult NHS mental health services.
(Clinical Guideline 136). NICE: London, 2011.
3 van Gent EM, Vida SL, Zwart FM. Group therapy in addition to
lithium therapy in patients with bipolar disorders. Acta Psychiatr Belg
1988; 88: 405–18.
www.thelancet.com/psychiatry Vol 3 November 2016
Articles
4 Peet M, Harvey NS. Lithium maintenance: 1. A standard education
programme for patients. Br J Psychiatry 1991; 158: 197–200.
5 Colom F, Vieta E, Martinez-Aran A, et al. A randomized trial on the
efficacy of group psychoeducation in the prophylaxis of recurrences
in bipolar patients whose disease is in remission.
Arch Gen Psychiatry 2003; 60: 402–07.
6 Colom F, Vieta E, Reinares M, et al. Psychoeducation efficacy in
bipolar disorders: beyond compliance enhancement.
J Clin Psychiatry 2003; 64: 1101–05.
7 Yatham LN, Kennedy SH, Parikh SV, et al. Canadian Network for
Mood and Anxiety Treatments (CANMAT) and International Society
for Bipolar Disorders (ISBD) collaborative update of CANMAT
guidelines for the management of patients with bipolar disorder:
update 2013. Bipolar Disord 2013; 15: 1–44.
8 Bond K, Anderson IM. Psychoeducation for relapse prevention in
bipolar disorder: a systematic review of efficacy in randomized
controlled trials. Bipolar Disord 2015; 17: 349–62.
9 Oud M, Mayo-Wilson E, Braidwood R, et al.
Psychological interventions for adults with bipolar disorder:
systematic review and meta-analysis. Br J Psychiatry 2016;
208: 213–22.
10 Lobban F, Taylor L, Chandler C, et al. Enhanced relapse prevention
for bipolar disorder by community mental health teams: cluster
feasibility randomised trial. Br J Psychiatry 2010; 196: 59–63.
11 American Psychiatric Association. Diagnostic and Statistical Manual
of Mental Disorders, 4th edn. Washington, DC: American
Psychiatric Association, 1994.
12 First MB, Spitzer RL, Gibbon M, Endicott J. Structured Clinical
Interview for DSM-IV Axis 1 Disorders (SCID-I). Washington, DC:
American Psychiatric Press, 1997.
13 Morriss RK, Lobban F, Riste L, et al. Pragmatic randomised
controlled trial of group psychoeducation versus group support
in the maintenance of bipolar disorder. BMC Psychiatry 2011;
11: 214.
14 Colom F, Vieta E. Psychoeducation manual for bipolar disorder.
Cambridge: Cambridge University Press, 2016.
15 Paykel ES, Abbott R, Morriss R, et al. Sub-syndromal and syndromal
symptoms in the longitudinal course of bipolar disorder.
Br J Psychiatry 2006; 189: 118–23.
16 Scott J, Paykel E, Morriss R, et al. Cognitive-behavioural therapy for
severe and recurrent bipolar disorders: randomised controlled trial.
Br J Psychiatry 2006; 188: 313–20.
17 Morriss R, Scott J, Paykel E, et al. Social adjustment based on
reported behaviour in bipolar affective disorder. Bipolar Disord 2007;
9: 53–62.
18 Goldman HH, Skodol AE, Lave TR. Revising Axis V for DSM-IV:
a review of measures of social functioning. Am J Psychiatry 1992;
49: 1148–56.
19 Williams JBW, Kobak KA, Bech P, et al. The GRID-HAMD:
standardization of the Hamilton Depression Rating Scale.
Int Clin Psychopharmacol 2008; 23: 120–29.
20 Licht RW, Jensen J. Validation of the Bech-Rafaelsen Mania Scale
using latent structure analysis. Acta Psychiatr Scand 1997;
96: 367–72.
21 Zigmond AS, Snaith RP. The hospital anxiety and depression scale.
Acta Psychiatr Scand 1983; 67: 361–70.
22 Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health
Survey: construction of scales and preliminary tests of reliability and
validity. Med Care 1996; 34: 220–33.
23 White IR, Thompson SG. Adjusting for partially missing
baseline measurements in randomized trials. Stat Med 2005;
24: 993–1007.
24 Braun V, Clarke V. Using thematic analysis in psychology.
Qual Res Psychol 2006; 3: 77–101.
25 Miklowitz DJ, Goodwin GM, Bauer MS, Geddes JR. Common and
specific elements of psychosocial treatments for bipolar disorder:
a survey of clinicians participating in randomized trials.
J Psychiatr Pract 2008; 14: 77–85.
26 Kessing LV, Hansen HV, Hvenegaard A, et al; Early Intervention
Affective Disorders (EIA) Trial Group. Treatment in a specialised
out-patient mood disorder clinic v. standard out-patient treatment
in the early course of bipolar disorder: randomised clinical trial.
Br J Psychiatry 2013; 202: 212–19.
1037
 Articles

27 Jones SH, Smith G, Mulligan LD, et al. Recovery-focused 30 Colom F, Reinares M, Pacchiarotti I, et al. Has number of
cognitive-behavioural therapy for recent-onset bipolar disorder: previous episodes any effect on response to group
randomised controlled pilot trial. Br J Psychiatry 2015; 206: 58–66. psychoeducation in bipolar patients? A 5-year follow-up post hoc
28 Perry A, Tarrier N, Morriss R, et al. Randomised controlled trial of analysis. Acta Neuropsychiatr 2010; 22: 50–53.
teaching bipolar disorder patients to identify early symptoms of
relapse and obtain early treatment. BMJ 1999; 318: 149–53.
29 Morriss RK, Faizal MA, Jones AP, et al. Interventions for helping
people recognise early signs of recurrence in bipolar disorder.
Cochrane Database Syst Rev 2007; 1: CD004854.

1038 www.thelancet.com/psychiatry Vol 3 November 2016