Sie sind auf Seite 1von 9


Ecacy of cognitive behavioural therapy for sleep

improvement in patients with persistent delusions and
hallucinations (BEST): a prospective, assessor-blind,
randomised controlled pilot trial
Daniel Freeman, Felicity Waite, Helen Startup, Elissa Myers, Rachel Lister, Josephine McInerney, Allison G Harvey, John Geddes, Zenobia Zaiwalla,
Ramon Luengo-Fernandez, Russell Foster, Lei Clifton, Ly-Mee Yu

Background Sleep disturbance occurs in most patients with delusions or hallucinations and should be treated as a Lancet Psychiatry 2015;
clinical problem in its own right. However, cognitive behavioural therapy (CBT)the best evidence-based treatment 2: 97583

for insomniahas not been tested in this patient population. We aimed to pilot procedures for a randomised trial Published Online
September 10, 2015
testing CBT for sleep problems in patients with current psychotic experiences, and to provide a preliminary
assessment of potential benet. S2215-0366(15)00314-4
See Comment page 950
Methods We did this prospective, assessor-blind, randomised controlled pilot trial (Better Sleep Trial [BEST]) at two Department of Psychiatry,
mental health centres in the UK. Patients (aged 1865 years) with persistent distressing delusions or hallucinations in Warneford Hospital
the context of insomnia and a schizophrenia spectrum diagnosis were randomly assigned (1:1), via a web-based (Prof D Freeman PhD,
randomisation system with minimisation to balance for sex, insomnia severity, and psychotic experiences, to receive F Waite DClinPsy,
H Startup DPhil, E Myers DPhil,
either eight sessions of CBT plus standard care (medication and contact with the local clinical team) or standard care R Lister BA, J McInerney MSc,
alone. Research assessors were masked to group allocation. Assessment of outcome was done at weeks 0, 12 (post- Prof J Geddes MD) and Health
treatment), and 24 (follow-up). The primary ecacy outcomes were insomnia assessed by the Insomnia Severity Economics Research Centre,
Department of Public Health
Index (ISI) and delusions and hallucinations assessed by the Psychotic Symptoms Rating Scale (PSYRATS) at week
(R Luengo-Fernandez DPhil) and
12. We did analysis by intention to treat, with an aim to provide condence interval estimation of treatment eects. Sleep and Circadian
This study is registered with ISRCTN, number 33695128. Neurosciences Institute,
Nueld Department of Clinical
Neurosciences, John Radclie
Findings Between Dec 14, 2012, and May 22, 2013, and Nov 7, 2013, and Aug 26, 2014, we randomly assigned
Hospital (Prof R Foster PhD) and
50 patients to receive CBT plus standard care (n=24) or standard care alone (n=26). The last assessments were Centre for Statistics in
completed on Feb 10, 2015. 48 (96%) patients provided follow-up data. 23 (96%) patients oered CBT took up the Medicine, Nueld Department
intervention. Compared with standard care, CBT led to reductions in insomnia in the large eect size range at week of Orthopaedics,
Rheumatology and
12 (adjusted mean dierence 61, 95% CI 3092, eect size d=19). By week 12, nine (41%) of 22 patients receiving
Musculoskeletal Sciences
CBT and one (4%) of 25 patients receiving standard care alone no longer had insomnia, with ISI scores lower than the (L Clifton PhD) and Nueld
cuto for insomnia. The treatment eect estimation for CBT covered a range from reducing but also increasing Department of Primary Care
delusions (adjusted mean dierence 03, 95% CI 20 to 26) and hallucinations (19, 65 to 27). Three patients, Health Sciences (L-M Yu MSc),
University of Oxford, Oxford,
all in the CBT group, had ve adverse events, although none were regarded as related to study treatment.
UK; Sussex Partnership NHS
Trust, Worthing, West Sussex
Interpretation Our ndings show that CBT for insomnia might be highly eective for improving sleep in patients with (H Startup); Department of
persistent delusions or hallucinations. A larger, suitably powered phase 3 study is now needed to provide a precise Psychology, University of
California, Berkeley, CA, USA
estimate of the eects of CBT for sleep problems, both on sleep and psychotic experiences. (Prof A G Harvey PhD); and
Oxford Non-Respiratory Sleep
Funding Research for Patient Benet Programme, National Institute for Health Research. Disorder Service, Oxford
University Hospitals NHS Trust,
John Radclie Hospital, Oxford,
Copyright Freeman et al. Open Access article distributed under the terms of CC BY. UK (Z Zaiwalla FRCP)
Correspondence to:
Introduction intrinsic feature of schizophrenia. These ndings extend Prof Daniel Freeman,
Sleep problems are pervasive in people with schizophrenia. into the general population. Findings from two national Department of Psychiatry,
In a study1 of patients with persecutory delusions, 54% had epidemiological studies6,7 have shown that insomnia is Warneford Hospital,
University of Oxford,
clinical insomnia, 30% had subthreshold insomnia, and strongly associated with paranoia; additionally, sleep Oxford OX3 7JX, UK
only 16% were sleeping well. In outpatients with clinically problems are associated with psychotic-like experiences in
stable schizophrenia, poor sleep is common2,3 and children.8 Yet, the treatment (or even routine assessment)
associated with an increased severity of positive symptoms.3 of sleep problems in people with schizophrenia has
Relatives of patients with schizophrenia notice sleep received scant attention. Clinical trials in populations
problems more than any other sign preceding relapse,4 and without a diagnosis of schizophrenia have shown that
a meta-analysis5 concluded that sleep disorders are an cognitive behavioural therapy (CBT) is highly eective for Vol 2 November 2015 975


Research in context
Evidence before this study Added value of this study
We searched PubMed and the ISRCTN trial registry up to April Our trial is the rst randomised controlled trial to treat
10, 2015, with the terms insomnia, therapy, and insomnia in patients with current psychotic experiences, use a
schizophrenia, without date restrictions, for English-language psychological treatment for insomnia with patients with
publications of randomised controlled trials investigating the schizophrenia, and examine eects of an insomnia treatment
treatment of insomnia with patients with schizophrenia. We did for patients with schizophrenia for up to 6 months. Our
not nd any randomised trials of treatment of insomnia in ndings show that CBT for insomnia is likely to be benecial for
schizophrenia with a psychological therapy. Apart from our own reducing insomnia in patients with schizophrenia.
case report series, we did not nd any descriptions of cognitive
Implications of all the available evidence
behavioural therapy (CBT) for insomnia with patients with
Trial data are insucient for the treatment of sleep problems in
schizophrenia. Three further searches with the terms insomnia,
patients with schizophrenia. The present study, in combination
randomized, schizophrenia, and hypnotics/benzodiazepine/
with studies of CBT for insomnia in other disorders, suggests
melatonin, and reading of review papers, identied only three
that CBT could be oered as psychological treatment for
small randomised controlled trials testing the short-term eects
patients with schizophrenia. However a larger, suitably
of melatonin and eszopiclone. We found no randomised
powered phase 3 study is indicated.
controlled trials assessing treatment of insomnia in patients
selected for having current delusions and hallucinations.

treatment of insomnia,912 and CBT is considered by many hallucinations. We followed the recommendation of Lee
as the recommended treatment for this disorder.13 However, and colleagues22 that pilot studies are more about
this therapy is yet to be tested in a randomised controlled learning than conrming: they are not designed to
trial in patients with schizophrenia. formally assess evidence of benet. As such, we aimed
Treatment of sleep problems in patients with psychosis to establish recruitment and follow-up rates, indicate
might have another important benet: reductions in levels of compliance with the treatment, examine the use
delusions and hallucinations. Sleep disturbance is of sleep assessments in this population, and provide an
increasingly recognised as a potential contributory factor estimation of the potential ecacy of the intervention.
to the occurrence of a wide range of mental health The study is part of our process of improving treat-
problems.14,15 Sleep disturbance might also have a role in ments for patients with delusions and hallucinations,
the occurrence of psychotic experiences such as delusions translating advances in understanding into intervention.
and hallucinations. Findings from longitudinal studies One putative causal factor at a time is targeted and the
have shown that insomnia predicts new inceptions of eect on the psychotic experiences examined23a so-
paranoia16 and its persistence.17 Insomnia increases called interventionist-causal model approach.24 Our
negative aect, anomalous perceptions, and reasoning intention was to use clinical techniques focused on sleep
errors, which are all factors implicated in the development and then examine the subsequent eects. In this
of persecutory ideation. Studies have also linked insomnia mechanistic approach, the central need is to establish
and hallucinatory experiences.18,19 In adolescents at an change in the putative causal factor (ie, sleep) in one
ultra-high risk of psychosis, sleep diculties are a group compared with a group in which sleep patterns
predictor of positive psychotic experiences.20 Importantly, remain relatively stable. Therefore, the appropriate
research in twins has shown overlap in the genetic and design was to compare the targeting of sleep with
environmental causes of insomnia and psychotic standard care. Identication of the active ingredients of
experiences.18 If a causal link does exist, the clinical intervention was not the question that the trial was
implication is that treatment of insomnia in patients with designed to address.
schizophrenia could lessen psychotic experiences, which We had two primary outcome hypotheses: that CBT for
would provide a new treatment route for these patients. insomnia added to standard care would improve sleep in
In a case series,21 we used CBT for insomnia in patients with psychosis compared with standard care
15 patients with persistent persecutory delusions in the alone, and that CBT for insomnia added to standard care
context of non-aective psychosis. After the brief CBT would reduce delusions and hallucinations compared
intervention, we recorded large reductions in levels of with standard care alone. Our secondary hypotheses were
insomnia and paranoia. We also recorded signicant that improvements in sleep and psychotic symptoms
reductions in levels of hallucinations, anxiety, and would be maintained over at least 3 months, that
depression. A methodologically rigorous assessment is improvement in sleep would be associated with improve-
now needed. We planned the Better Sleep Trial (BEST) as ments in psychotic symptoms, and that CBT for insomnia
a randomised controlled pilot trial assessing treatment of would lead to improvements in other outcomes, such as
insomnia in patients with persistent delusions or patient wellbeing and feelings of fatigue.

976 Vol 2 November 2015


Methods another assessor (which happened six times in total,

Study design and patients three times at each follow-up point). All assessments
We did this prospective, assessor-blind, randomised were therefore done blind to allocation.
controlled pilot study at two centres in the UK: Oxford
Health National Health Service (NHS) Foundation Procedures
Trusta large mental health service covering a The CBT intervention was provided one to one by
population of roughly 12 million peopleand, in the clinical psychologists (EM and FW), either in NHS
nal 4 months of recruitment, at an additional site at clinics or at the patients home. DF and HS did weekly
Northamptonshire Healthcare NHS Foundation Trust. clinical supervision. The aim was to provide the
We sought to include patients who had persistent, insomnia intervention in about eight sessions over
distressing delusions or hallucinations in the context of 12 weeks, with four sessions dened as a minimum
non-aective psychosis and insomnia. Inclusion criteria therapeutic dose, with exibility in length and number
were current delusion or hallucination; a score of at least of sessions as appropriate in this clinical group. We
2 on the distress items of the Psychotic Symptoms Rating also included telephone calls and texts between sessions
Scale (PSYRATS)25 for either a delusion or hallucination; to maintain treatment momentum. The main tech-
delusion or hallucination that had persisted for at least niques standard for CBT sleep interventions were taken
3 months; a clinical diagnosis of schizophrenia, from four main sources.2831 The intervention was
schizoaective disorder, or delusional disorder (ie, a written in a manual to guide the work, which was
diagnosis of non-aective psychosis); sleep diculties shared with the patient. Initially the sessions focused
lasting 1 month or longer and a score of 15 or more on
the Insomnia Severity Index26 (ISI; ie, clinical insomnia);
an age of 1865 years; and a medication dosage that had 192 patients referred
been stable for at least the past month. Exclusion criteria
were a primary diagnosis of sleep apnoea, alcohol or 86 excluded
substance dependency, organic syndrome or learning 83 declined to be screened
3 because of high forensic risk
disability, a command of spoken English inadequate for
engagement in therapy, and current engagement in
individual CBT. Patient enrolment was done by one full- 106 screened for eligibility
time graduate psychologist (RL).
The trial received ethics approval from the NHS
56 excluded
Research Ethics Committee South CentralOxford C 15 had an insucient score on ISI
(reference 12/SC/0138) and the protocol has been 26 had PSYRATS scores below the cuto for current
distressing delusion or hallucination
published elsewhere.27 All patients provided written 1 had insucient capacity to consent
informed consent. 9 excluded (ie, alcohol or substance dependence,
cognitive decit, organic syndrome,
or learning disability)
Randomisation and masking 1 had a command of English that was inadequate
We randomly assigned patients (1:1), via a web-based to engage in psychological therapy
2 were undergoing current individual CBT
randomisation system, to receive either CBT plus 2 declined to participate
standard care or standard care alone. The randomisation
system was designed to balance three variables with a
non-deterministic minimisation algorithm: sex (male vs 50 randomly assigned
female), severity of sleep problem (moderate [ISI score
1521] vs high [2228]), and psychotic experiences
(hallucination only vs delusion only vs hallucinations and
24 allocated to CBT plus standard care 26 allocated to standard care alone
delusions). 23 received CBT intervention 26 received standard care alone
Research assessors (RL for most of the trial, with the 1 did not receive CBT intervention
last follow-up assessments completed by a replacement
graduate psychologist [JM]) were masked to group 25 completed 12 week assessment
22 completed 12 week assessment
allocation; the trial therapists informed patients of 2 missed 12 week assessment 1 missed 12 week assessment
the randomisation outcome to maintain allocation
23 completed 24 week assessment 25 completed 24 week assessment
concealment. Precautionary strategies to prevent un- 1 missed 24 week assessment 1 missed 24 week assessment
masking included the therapist and assessor considering
room use and booking arrangements; patients being
reminded by the assessor not to talk about treatment 24 included in intention-to-treat analysis 26 included in intention-to-treat analysis
allocation; and, after the initial assessment, the assessor
not looking at the patients clinical notes. In the case of Figure: Trial prole
an allocation being revealed, we remasked by using ISI=Insomnia Severity Index. PSYRATS=Psychotic Symptoms Rating Scale. CBT=cognitive behavioural therapy. Vol 2 November 2015 977


attitudes about sleep. The intervention was deliberately

CBT group (n=24) Standard care alone
group (n=26) simplied, with the principal therapeutic techniques
being stimulus control (ie, learning to associate bed
Age (years) 396 (116) 422 (135)
with sleep) and improvement of daytime activity levels.
A detailed description of our approach to treatment of
Male 16 (67%) 18 (69%)
sleep problems in this group is available elsewhere,32
Female 8 (33%) 8 (31%)
and includes noting of adaptations needed for the
Ethnic origin
particular problems of delusions and hallucinations
White 22 (92%) 25 (96%)
interfering with sleep, attempts to sleep being overused
Black 1 (4%) 1 (4%)
by patients as an escape from voices, extensive
Chinese 1 (4%) 0 disruption of circadian rhythms, insucient daytime
Employment status activity, and fear of the bed based on past adverse
Unemployed 21 (88%) 23 (88%) experiences. Sessions were taped with patient
Part-time employed 1 (4%) 0 agreement. To assess treatment delity, six tapes,
Full-time employed 0 1 (4%) chosen at random, were rated on the Cognitive Therapy
Volunteer 1 (4%) 1 (4%) ScaleRevised33 by an independent clinical psychologist
Retired 1 (4%) 0 experienced in CBT for psychosis. All tapes were rated
Student 0 1 (4%) as providing at least satisfactory cognitive therapy (ie,
Clinical diagnosis an average score of at least three on scale items).
Schizophrenia 16 (67%) 17 (65%) Standard care was delivered according to national and
Schizoaective disorder 5 (21%) 5 (19%) local service protocols and guidelines and mainly
Delusional disorder 0 0 consisted of antipsychotic medication and contact with
Psychosis not otherwise specied 3 (13%) 4 (15%) the local clinical team. We recorded medication and
Severity of insomnia (ISI) hospital admissions from clinical notes and other
Moderate (1521) 19 (79%) 19 (73%) service provision using a modied version of the Client
High (2228) 5 (21%) 7 (27%) Service Receipt Inventory.34 In the baseline assessment,
Presence of psychotic experiences the trial patients were also assessed for insomnia using
Both delusions and hallucinations 18 (75%) 19 (73%) the Duke Structured Interview Schedule for Sleep
Delusion only 4 (17%) 3 (12%) Disorder Diagnoses.35
Hallucination only 2 (8%) 4 (15%)
Depression (BDI) Outcomes
None (013) 3 (13%) 3 (12%) As a pilot study, our main outcomes concerned the
Mild (1419) 5 (21%) 3 (12%) number of participants who were recruited, complied
Moderate (2028) 2 (8%) 6 (23%) with treatment, and followed up. The prespecied
Severe (2963) 14 (58%) 14 (54%) primary outcome measures were levels of insomnia,
assessed with the ISI,26 and levels of delusions and
Data are mean (SD) or n (%), unless otherwise indicated. CBT=cognitive behavioural therapy. ISI=Insomnia Severity hallucinations, assessed with the PSYRATS.25 Higher
Index. BDI=Beck Depression Inventory.
scores on these measures indicate greater severity.
Table 1: Baseline demographic and clinical characteristics Secondary outcome measures assessed sleep using
dierent methods: a second self-report questionnaire
measure of sleep (Pittsburgh Sleep Quality Index [PSQI]),36
on psycho-education about sleep diculties, assess- a sleep diary, and actigraphy data obtained by participants
ment of the triggering and maintenance of sleep wearing an actiwatch (CamNtech MotionWatch 8,
diculties, and goal setting. We used a checklist of CamNtech, Cambridge, UK) for at least 7 days. Additionally,
factors likely to cause sleep diculties, which was we assessed secondary outcome measures for psychiatric
generated by the team. On the basis of the assessment, symptoms: a self-report measure of paranoid thinking
the active therapeutic techniques used could have (Paranoid Thoughts Scale);37 a standard psychiatric
included stimulus control therapy (eg, setting of interviewer-rated assessment (the Positive and Negative
appropriate and regular sleep times, ensuring the bed Syndromes Scale; PANSS);38 and an adapted patient
or bedroom were used only for sleeping, not staying in reported outcome measure (CHoice of Outcome In Cbt for
bed if unable to sleep for longer than 2030 min, psychosEs),39 which assessed, for example, self-condence,
reducing sleep in the daytime), establishment of peace of mind, and a sense of being in control. Other
appropriate daytime activity and circadian rhythms (eg, secondary outcomes were fatigue (Multidimensional
obtaining natural light in the morning, regular Fatigue Inventory),40 quality of life (Euroqol 5 Dimensions 5
mealtimes, gradually shifting sleep and wake times for Levels; responses were converted into a single summary
sleep-phase problems), sleep hygiene, relaxation, and measure using UK population taris),41 and psychological
cognitive techniques addressing unhelpful beliefs and wellbeing (Warwick-Edinburgh Mental Well-being Scale).42

978 Vol 2 November 2015


We included several measures, such as the Beck Depression

Inventory (BDI),43 for the purposes of potential mediation CBT group Standard care alone group

analysis. The outcome measures were completed at weeks 0 n Mean (SD) n Mean (SD)
(baseline), 12 (post-intervention), and 24 (follow-up). Antipsychotic medication (chlorpromazine equivalent dose, mg/day)
During the trial, we recorded any adverse event that
Baseline 24 3637 (2665) 26 4958 (3581)
came to our attention. Medical notes were also checked
12 weeks 23 3598 (2893) 25 5307 (3660)
at the end of the trial for the following events prespecied
24 weeks 23 3725 (3029) 25 5313 (3732)
as adverse: all deaths, suicide attempts, serious violent
incidents, admissions to secure units, formal complaints 6 months before the trial (n)
about therapy. Psychiatric hospital admission 24 01 (03) 26 01 (03)
Psychiatrist meetings 23 12 (14) 26 25 (49)
Statistical analysis Meetings with community psychiatric nurse 23 117 (136) 26 154 (162)
The trial statisticians (LC and L-MY) prepared a fully Visits to day-care centre 23 60 (148) 26 20 (68)
detailed statistical analysis plan and the chief Meetings with general practitioner 22 28 (36) 25 30 (53)
investigator (DF) approved the plan before any analysis.
6 months during the trial (n)
No formal sample size calculation was done for this
Psychiatric hospital admission 24 01 (03) 26 01 (03)
pilot study. However, the target sample size was based
on recruitment for 15 months at an estimated rate of Psychiatrist meetings 19 11 (09) 24 23 (39)

four patients per month with one research worker (ie, Meetings with community psychiatric nurse 19 76 (82) 24 135 (139)
60 patients), which was considered adequate to obtain Visits to day-care centre 19 61 (186) 24 79 (249)
reasonably reliable sample size estimates.44 Outcomes Meetings with general practitioner 18 35 (39) 24 32 (32)
were assessed separately for assessment points at weeks
Table 2: Provision of standard care
12 and 24. We used ANCOVA to obtain estimates and
95% CIs of continuous outcomes, with adjustment for
baseline variables. The analysis plan did not include Results
reporting of p values, in accordance with recom- The gure shows the trial prole. Between Dec 14, 2012,
mendations that The analysis of a pilot study should be and May 22, 2013, and Nov 7, 2013, and Aug 26, 2014,
mainly descriptive or should focus on condence we randomly assigned 50 patients to receive CBT plus
interval estimation.45 As a sensitivity analysis, we standard care (n=24) or standard care alone (n=26),
controlled for initial overall symptom severity as with the break in enrolment due to employment of a
assessed by the PANSS and use of antipsychotic new trial therapist. 47 (94%) patients were from Oxford
medication. In a post-hoc analysis for the main Health and three (6%) were from Northamptonshire
outcomes, we constructed a mixed model to incorporate Healthcare. The last assessments were completed on
the repeated measures at the two assessments (weeks 12 Feb 10, 2015. All 50 patients completed the baseline
and 24) for each patient. We calculated eect sizes with assessment. 48 (96%) patients provided follow-up data
Cohens d by taking the estimated coecient of for the primary ecacy measures (gure). In the CBT
treatment allocation from the ANCOVA divided by the group, the mean number of sessions received was
pooled baseline SD. An eect size of 03 is considered a 73 (SD 19). On the basis of at least four CBT sessions
small eect, 05 a medium eect, and 08 or higher a constituting a minimum therapeutic dose, 23 (96%) of
large eect. We used correlation coecients to examine 24 patients had a dose of the intervention. The actual
potential associations between changes in sleep and number of treatment sessions attended in the 12 week
changes in psychotic experiences (changes in scores for period was three (n=1), four (n=1), ve (n=1), six (n=4),
each measure were calculated as week 12 score minus seven (n=5), eight (n=7), nine (n=2), ten (n=2), or
baseline score). All main analyses were done at the end 11 (n=1) sessions. The appendix shows descriptive See Online for appendix
of the last follow-up assessments (ie, there were no comments about the intervention received from ve of
interim analyses) and were based on the intention-to- the rst seven patients who had CBT.
treat population. Analyses were done with SAS Baseline and demographic characteristics were similar
(version 9.3)46 and were repeated by an independent between groups (table 1). In line with other studies of
statistician. This study is registered with ISRCTN, persistent psychotic experiences, both groups included
number 33695128. slightly more men, the average age was about 40 years,
most participants were unemployed, and the main
Role of the funding source diagnosis was schizophrenia (table 1). There were high
The funder of the study had no role in study design, data levels of depression in both groups (table 1), although
collection, data analysis, data interpretation, or writing of BDI scores were not signicantly correlated with level of
the report. DF, L-MY, and LC had full access to all the insomnia at baseline (r=007, p=0623). All but one
data in the study and had nal responsibility for the patient (in the standard care alone group) met the Duke
decision to submit for publication. Structured Interview Schedule for Sleep Disorder Vol 2 November 2015 979


Diagnoses criteria for insomnia disorder. Provision of (n=3 given CBT, n=1 given standard care alone) were
standard care was similar between groups and was fairly taking antipsychotic medication (table 2). Most
stable during the trial (table 2). All but four participants participants were also prescribed a hypnotic, anxiolytic,
or antidepressant medication (19 [79%] in the CBT group,
CBT group Standard care alone Adjusted mean Eect 21 [81%] in the standard care alone group; appendix). All
group dierence size (d)
(95% CI)
participants were outpatients.
Compared with standard care alone, CBT had a
n Mean (SD) n Mean (SD)
treatment benet on insomnia in the large eect size
Primary outcome measure range at 12 weeks (table 3). Benets were still apparent
Insomnia (ISI) at 24 weeks follow-up (table 2). At baseline, no patients
0 weeks 24 186 (32) 26 188 (33) ISI scores were lower than the cuto indicating normal
12 weeks 22 93 (55) 25 154 (54) 61 19 sleep (a score of 07). By week 12, nine (41%) of
(30 to 922) 22 patients in the CBT group and one (4%) of 25 patients
24 weeks 23 110 (56) 25 150 (57) 39 12 in the standard care alone group scored less than this
(09 to 68)
cuto. The wide condence intervals for the eects of
Delusions (PSYRATS)
CBT on delusions and hallucinations cover a range from
0 weeks 24 161 (32) 26 153 (49)
decreasing to increasing psychotic experiences (table 3).
12 weeks 22 139 (48) 25 138 (41) 03 01
(20 to 26)
Changes in the other sleep assessments were relatively
24 weeks 23 140 (47) 25 127 (57) 08 02
consistent with those for the primary outcome measure
(36 to 21) of insomnia (table 3). We recorded moderate to large
Hallucinations (PSYRATS) eect sizes in sleep quality as assessed by the PSQI
0 weeks 24 251 (121) 26 267 (92) (table 3). Data collection was less complete for the sleep
12 weeks 22 275 (92) 25 259 (81) 19 02 diaries and actigraphy than for the other secondary
(65 to 27) outcome measures, and the ensuing eect sizes were
24 weeks 23 246 (116) 25 220 (102) 34 03 small to moderate (table 3). Compared with standard
(92 to 23) care alone, patients reported reduced fatigue at week 12,
Sleep secondary outcome measures and improved quality of life and psychological wellbeing
Sleep quality (PSQI) at week 24, with small to medium eect sizes overall for
0 weeks 23 119 (32) 23 116 (29) both these categories (table 3). The condence intervals
12 weeks 21 87 (38) 23 105 (48) 19 06 for paranoia and overall psychiatric symptomatology
(04 to 41)
again span CBT potentially reducing or increasing these
24 weeks 23 69 (44) 23 96 (48) 28 09 problems (table 3). Correlations between changes in
(02 to 54)
insomnia and changes in hallucinations and paranoia
Time to sleep onset (min)*
(although not PSYRATS delusions) were mainly small
0 weeks 21 615 (514) 25 611 (327)
and positive, but the condence intervals are wide and
12 weeks 22 346 (403) 19 466 (370) 183 04
(15 to 382) include negative correlations (appendix).
24 weeks 20 333 (296) 21 579 (417) 264 06
All the notes of the trial patients were checked. Four
(25 to 503) patients were admitted to hospital during the trial (n=2
Total sleep time (min)* in each group; appendix). There were no deaths or
0 weeks 18 3937 (1278) 21 4034 (1463) complaints about therapy. Three patients, all in the CBT
12 weeks 17 4564 (1077) 18 4375 (1181) 330 02 group, had a total of ve adverse events: two suicide
(179 to 838) attempts, two serious violent incidents, and one
24 weeks 18 4659 (1171) 20 4124 (1282) 466 03 admission to a secure unit (following one of the violent
(151 to 1083) incidents). No adverse events were considered to be
Waking in night (min)* related to study treatment.
0 weeks 19 435 (402) 22 591 (633) The appendix shows results of the sensitivity analysis.
12 weeks 18 199 (222) 18 476 (521) 169 03 Results of the post-hoc analysis using a mixed model for
(47 to 385)
repeated measures analysis were similar to those
24 weeks 18 327 (357) 20 427 (466) 146 03
obtained using ANCOVA (appendix).
(111 to 403)
Total sleep time (min)
0 weeks 20 3963 (1417) 23 4700 (1287)
This is the rst randomised controlled trial testing the
12 weeks 18 4062 (1000) 19 4710 (1244) 155 01
(789 to 479) eects of CBT for insomnia in patients with current
24 weeks 18 4451 (867) 21 4498 (1411) 331 02 psychotic experiences, which comprised very long-
(270 to 933) standing insomnia, delusions, and hallucinations.
(Table 3 continues on next page) Overall our study shows the feasibility of testing the
psychological treatment adapted for this population: it

980 Vol 2 November 2015


was possible to recruit patients to the trial, to randomise

CBT group Standard care alone Adjusted mean Eect
them, to keep assessors masked to allocation, to retain group dierence size (d)
patients in the trial, to assess a wide range of measures, (95% CI)
and to implement the treatment. Indeed, there was a n Mean (SD) n Mean (SD)
very low dropout rate for the trial assessments and a
(Continued from previous page)
very high uptake of CBT. Our ndings also show that
Psychiatric secondary measures
CBT for insomnia, a relatively brief intervention, is
Paranoia (GPTS)
likely to have substantial benets in improving sleep
0 weeks 24 908 (287) 26 905 (298)
for patients with current delusions and hallucinations.
12 weeks 22 896 (368) 24 962 (373) 67 02
Benets in sleep were sustained up to the nal follow- (102 to 235)
up assessment. 24 weeks 20 783 (348) 25 881 (350) 98 03
Findings from a qualitative study47 of a subsample of (75 to 272)
the trial participants are consistent with those of the Total symptoms (PANSS)
present quantitative analysis. Our results compare well 0 weeks 24 836 (162) 26 797 (141)
with the only randomised controlled trial that we are 12 weeks 22 775 (121) 24 793 (146) 34 02
aware of that targeted insomnia in patients with (23 to 90)
schizophrenia with one of the commonly prescribed 24 weeks 21 748 (147) 24 758 (118) 33 02
sedative hypnotics (so-called Z-drugs). In a randomised (29 to 95)
double-blind trial48 done over 8 weeks in 39 clinically Fatigue (MFI)
stable outpatients with schizophrenia who had insomnia 0 weeks 23 438 (164) 26 476 (153)
but were not recruited for current psychotic experiences, 12 weeks 22 291 (190) 24 454 (196) 105 07
(21 to 189)
eszoplicone was compared with placebo. The medication
led to a decrease in ISI scores of 378 (95% CI 0275) 24 weeks 21 259 (214) 25 384 (181) 90 06
(11 to 191)
versus placebo. The reduction in ISI score in the present
Patient outcomes (CHOICE)
study is very similar to that reported in a meta-analysis
0 weeks 23 522 (211) 26 550 (144)
of CBT for comorbid insomnia for patients with physical
12 weeks 22 580 (227) 24 499 (183) 85 05
or psychiatric disorders (ISI mean change 64 (14 to 185)
[SE 127]).11 Our trial did not prove informative about 24 weeks 21 600 (228) 23 575 (218) 38 02
the potential eect of the sleep treatment on the (73 to 149)
psychotic experiences. The wide condence intervals Quality of life (EQ-5D-5L)
include the possibility that the treatment can either 0 weeks 24 055 (023) 26 060 (022)
reduce or increase delusions and hallucinations. A 12 weeks 22 063 (025) 24 055 (022) 011 05
larger trial will provide a better estimate of the eects on (002 to 023)
psychotic experiences. 24 weeks 22 063 (027) 25 058 (020) 008 04
The clear caution when discussing potential clinical (005 to 021)
eects is that this study was a pilot study. Our trial is Wellbeing (WEMWBS)
insuciently powered to detect anything but the largest 0 weeks 24 353 (93) 26 370 (78)
eect sizes. There was no formal power calculation 12 weeks 22 361 (107) 24 340 (89) 26 03
(21 to 74)
because we did not primarily intend to detect treatment
24 weeks 21 394 (99) 25 347 (79) 48 06
eects. We did not achieve the target sample size of
(05 to 93)
60 patients, but this information will form part of the
recruitment strategy plan for the future phase 3 trial. CBT=cognitive behavioural therapy. ISI=Insomnia Severity Index. PSYRATS=Psychotic Symptoms Rating Scale.
PSQI=Pittsburgh Sleep Quality Index. GPTS=Green et al Paranoid Thought Scales. PANSS=Positive and Negative Syndromes
Our main purpose was to assess the feasibility of the
Scale. MFI=Multidimensional Fatigue Inventory. CHOICE=CHoice of Outcome In Cbt for psychoses. EQ5D=Euroqol 5
assessment and to rene procedures. A key observation Dimensions 5 Levels. WEMWBS=Warwick-Edinburgh Mental Well-being Scale. *Sleep diary. Actigraphy data.
was the wide variety of sleep disturbance within the
patient group, including patients going to bed early in Table 3: Scores for the primary and secondary outcome measures

the evening and simply lying awake for hours, patients

not having a bed, patients having an association with outcomes. Such presentations are consistent with
their bed as a place of trauma, patients awake and pacing evidence of varied circadian rhythm dysfunction in
all night, patients sleep being disturbed through the use patients with schizophrenia49 and, indeed, the complexity
of hypnotics, patients being drowsy and oversleeping at of sleep disturbance beyond insomnia in other disorders
various times of the day, and patients only going to bed such as bipolar disorder.50 Notably, our trial did not
in the early hours (delayed-phase sleep), whereas others include polysomnography. Obvious practical diculties
had obviously irregular sleepwake patterns. Our exist in implementation of such a measurement, and
treatment manual will become more specic to each of subjective reports of problems must be the key clinical
these types of presentation and it would be of interest to priority. Instead of polysomnography we used actigraphy
establish whether they are associated with dierent to assess sleepwake timing and fragmentation. This Vol 2 November 2015 981


form of assessment (wearing a watch-like device) was Declaration of interests

not agreed to by all participants. Furthermore, we have We declare no competing interests.
concerns that the movement recordings are not able to Acknowledgments
dierentiate between actual sleep and drowsy but awake The trial was funded by a grant from the National Health Service (NHS)
National Institute for Health Research (NIHR) Research for Patient
inactivity, which is common in this patient group. Such Benet Programme (reference PB-PG-0211-10007). DF is supported by a
recordings might be better used to assess changes in Medical Research Council senior clinical fellowship. Research support to
overall activity levels, which is an important outcome in RGF, DF, and RL is provided by a Wellcome Trust Strategic Award
itself. Sleep improvement might also lead to physical (098461/Z/12/Z) for the Oxford Sleep and Circadian Neurosciences
Institute. JG is an NHS NIHR senior investigator. We thank Annabel
health benets.51 Ivins for facilitating recruitment in Northamptonshire. The views and
The key ecacy question addressed in the design of opinions expressed herein are those of the authors and do not
the present study was whether CBT can improve sleep necessarily reect those of the Research for Patient Benet Programme,
in patients with current psychotic experiences and NIHR, NHS, or the Department of Health.
insomnia. We assessed this question in the context of References
1 Freeman D, Pugh K, Vorontsova N, Southgate L. Insomnia and
our translational research that aims to improve paranoia. Schizophr Res 2009; 108: 28084.
treatments for people with psychosis.23 The causal 2 Poulin J, Chouinard S, Pampoulova T, Lecomte Y, Stip E, Godbut R.
framework that we use leads to the initial aim of Sleep habits in middle-aged, non-hospitalized men and women
showing change in the targeted mechanism (eg, sleep with schizophrenia. Psychiatry Res 2010; 30: 27478.
3 Xiang Y-T, Weng Y-Z, Leung C-M, Tang W-K, Lai K, Ungvari G.
disruption), hence the comparison of a sleep treatment Prevalence and correlates of insomnia and its impact on quality of
(whereby the expectation is that sleep will improve) life in Chinese schizophrenia patients. Sleep 2009; 32: 10509.
with continued standard care (whereby sleep patterns 4 Birchwood M, Smith J, Macmillan F, et al. Predicting relapse in
schizophrenia. Psychol Med 1989; 19: 64956.
are likely to remain stable). The chosen design also
5 Chouinar S, Poulin J, Stip E, Godbut R. Sleep in untreated patients
addresses the important clinical question of the with schizophrenia: a meta-analysis. Schizophr Bull 2004; 30: 95767.
potential overall benet provided when the intervention 6 Freeman D, Brugha T, Meltzer H, Jenkins R, Stahl D, Bebbington P.
is added to standard care. From some perspectives, this Persecutory ideation and insomnia: ndings from the second
British National Survey of Psychiatric Morbidity. J Psychiatr Res
question is key: The primary question facing clinicians 2010; 44: 102126.
and policy makers is whether adding a particular form 7 Freeman D, McManus S, Brugha T, Meltzer H, Jenkins R,
of treatment will signicantly improve symptoms or Bebbington P. Concomitants of paranoia in the general population.
Psychol Med 2011; 41: 92336.
functioning compared to the often-constrained services 8 Jeppesen P, Clemmensen L, Munkholm A, et al. Psychotic experiences
as usual.52 However, this question would only establish co-occur with sleep problems, negative aect and mental disorders in
whether the extra treatment has benets, not which preadolescence. J Child Psychol Psychiatry 2015; 56: 55865.
elements of the package are actually necessary or 9 Harvey AG, Blanger L, Talbot L, et al. Comparative ecacy of
behavior therapy, cognitive therapy, and cognitive behavior therapy
whether there are better forms of treatment. These for chronic insomnia: a randomized controlled trial.
latter questions require dierent choices of control J Consult Clin Psychol 2014; 82: 67083.
group in order to answer them. Nonetheless, our 10 Espie CA, MacMahoon KM, Kelly HL, et al. Randomized clinical
eectiveness trial of nurse-administered small group cognitive
clinical impression is that, for example, the extra behaviour therapy for persistent insomnia in general practice.
contact time alone provided by the CBT is an Sleep 2007; 30: 57484.
insucient explanation for such large improvements 11 Geiger-Brown J, Rogers VE, Liu W, Ludeman EM, Downton K,
Diaz-Abad M. Cognitive behavioral therapy in persons with
in sleep in a patient group that typically has long- comorbid insomnia. Sleep Med Rev 2015; 23: 5467.
standing and complex problems. Indeed, CBT has 12 Okajima I, Komada Y, Inque Y. A meta-analysis on the treatment
already been shown to have benet compared with eectiveness of cognitive behavioral therapy for primary insomnia.
Sleep Biol Rhythm 2011; 9: 2434.
attention-control conditions in the treatment of 13 Morin CM, Benca R. Chronic insomnia. The Lancet 2012; 379: 112941.
primary insomnia.53 14 Harvey AG, Murray G, Chandler RA, Soehner A. Sleep disturbance
Sleep diculties are a problem that patients with as transdiagnostic: consideration of neurobiological mechanisms.
psychosis nd distressing and very much wish to receive Clin Psychol Rev 2011; 31: 22535.
15 Foster RG, Peirson SN, Wul K, Winnebeck E, Vetter C,
eective treatment for from mental health services. A Roenneberg T. Sleep and circadian rhythm disruption in social
large, multicentre trial is now needed that can establish jetlag and mental illness. Prog Mol Biol Transl Sci 2013; 119: 32546.
the eects of CBT for insomnia, delivered by front-line 16 Freeman D, Stahl D, McManus S, et al. Insomnia, worry, anxiety
and depression as predictors of the occurrence and the persistence
mental health sta, both on sleep and psychotic of persecutory ideation. Soc Psychiatry Psychiatr Epidemiol 2012;
experiences. 47: 1195203.
Contributors 17 Freeman D, Thompson C, Vorontsova N, et al. Paranoia and post-
traumatic stress disorder in the months after a physical assault.
DF, L-MY, HS, and EM designed the study. DF was the principal
Psychological Med 2013; 43: 267384.
investigator and had the main responsibility for drafting the study
18 Taylor M, Gregory A, Freeman D, Ronald A. Do sleep disturbances
protocol and report. FW and EM were the trial therapists. DF, HS, and
and psychotic experiences in adolescence share genetic and
AGH provided the training and supervision for the trial therapists. RL environmental inuences? J Abnorm Psychol 2015; published online
and JM were the research assessors. L-MY and LC were responsible for May 4.
the trial outcome analyses. DF, L-MY, and LC take responsibility for the 19 Sheaves B, Porcheret K, Tsanas A, et al. Insomnia, nightmares and
integrity of the data and the accuracy of the data analysis. All authors chronotype as markers of risk for severe mental illness: results
contributed to, read, and approved the nal report. from a student population. Sleep (in press).

982 Vol 2 November 2015


20 Lunsford-Avery JR, LeBourgeois MK, Gupta T, Mittal VA. Actigraphic- 37 Green C, Freeman D, Kuipers E, et al. Measuring ideas of
measured sleep disturbance predicts increased positive symptoms in persecution and reference: the Green et al Paranoid Thought Scales
adolescents at ultra high-risk for psychosis. Schizophr Res 2015; (GPTS). Psychol Med 2008; 38: 10111.
published online March 26. DOI:10.1016/j.schres.2015.03.013. 38 Kay SR. Positive and negative syndromes in schizophrenia.
21 Myers E, Startup H, Freeman D. Cognitive behavioural treatment of New York, NY: Brunner, 1991.
insomnia in individuals with persistent persecutory delusions. 39 Greenwood K, Sweeney A, Williams S, et al. CHoice of Outcome In
J Behav Ther Exp Psychiatry 2011; 42: 33036. Cbt for psychosEs (CHOICE): the development of a new service
22 Lee E, Whitehead A, Jacques R, Julious S. The statistical user-led outcome measure of CBT for psychosis.
interpretation of pilot trials: should signicance thresholds be Schizophr Bull 2010; 36: 12635.
reconsidered? BMC Med Res Methodol 2014; 14: 41. 40 Smets EM, Garssen B, Bonke B, De Haes JC. The Multidimensional
23 Freeman D, Dunn G, Startup H, et al. Eects of cognitive behaviour Fatigue Inventory (MFI) psychometric qualities of an instrument to
therapy for worry on persecutory delusions in patients with assess fatigue. J Psychosom Res 1995; 39: 31525.
psychosis (WIT): a parallel, single-blind, randomised controlled trial 41 Brooks R, Rabin R, de Charro F. The measurement and valuation of
with a mediation analysis. Lancet Psychiatry 2015; 2: 30513. health status using EQ-5D: a European perspective. Dordrecht:
24 Kendler KS, Campbell J. Interventionist causal models in Kluwer Academic Publishers, 2003.
psychiatry. Psychological Med 2009; 39: 88187. 42 Tennant R, Hiller L, Fishwick R, et al. The Warwick-Edinburgh
25 Haddock G, McCarron J, Tarrier N, Faragher FB. Scales to measure Mental Well-being Scale (WEMWBS): development and UK
dimensions of hallucinations and delusions: the psychotic symptom validation. Health Qual Life Outcomes 2007; 5: 63.
rating scales (PSYRATS). Psychol Med 1999; 29: 87989. 43 Beck AT, Steer RA, Brown GK. BDI-II Manual. San Antonio, TX:
26 Bastien CH, Vallires A, Morin CM. Validation of the Insomnia The Psychological Corporation, 1996.
Severity Index as an outcome measure for insomnia research. 44 Browne RH. On the use of a pilot sample for sample size
Sleep Med 2011; 2: 297307. determination. Stat Med 1995; 14: 193340.
27 Freeman D, Startup H, Myers E, et al. The eects of using cognitive 45 Lancaster G, Dodd S, Williamson P. Design and analysis of pilot
behavioural therapy to improve sleep for patients with delusions studies: recommendations for good practice. J Eval Clin Pract 2004;
and hallucinations (the BEST study): study protocol. Trials 2013; 10: 30712.
14: 214. 46 SAS Institute. Base SAS 9.3 procedures guide. Cary, NC: SAS
28 Espie CA. Overcoming insomnia and sleep problems: a self help Institute, 2011.
guide using cognitive behavioural techniques. London: Constable 47 Waite F, Evans N, Myers E, et al. The patient experience of sleep
and Robinson, 2006. problems and their treatment in the context of current delusions
29 Freeman D, Freeman J. Know your mind: the complete family and hallucinations. Psychol Psychother 2015; published online
reference guide to emotional health. New York, NY: Sterling Aug 18. DOI:10.1111/papt.12073.
Publishing, 2010. 48 Tek C, Palmese L, Krystal A, et al. The impact of eszopiclone on
30 Meir H, Kryger MD. A womans guide to sleep disorders. New York, sleep and cognition in patients with schizophrenia and insomnia.
NY: McGraw-Hill, 2004. Schizophr Res 2014; 160: 18085.
31 Harvey AG, Sharpley AL, Ree MJ, Stinson K, Clark DM. An open 49 Wul K, Dijk D-J, Middleton B, Foster RG, Joyce EM. Sleep and
trial of cognitive therapy for chronic insomnia. Behav Res Ther 2007; circadian rhythm disruption in schizophrenia. Br J Psychiatry 2012;
45: 2491501. 200: 30816.
32 Waite F, Myers E, Harvey AG, et al. Treating sleep problems in 50 Harvey AG, Soehner AM, Kaplan KA, et al. Treating insomnia
patients with schizophrenia. Behav Cogn Psychother 2015; published improves mood state, sleep, and functioning in bipolar disorder.
online July 30. DOI: J Consult Clin Psych 2010; 83: 56477.
33 Blackburn I, James I, Milne D, et al. (2001). The Revised Cognitive 51 Pritchett D, Wul K, Oliver P, et al. Evaluating the links between
Therapy Scale (CTS-R). Behav Cogn Psychother 2001; 29: 43146. schizophrenia and sleep and circadian rhythm disruption.
34 Beecham J, Knapp M. (1992). Costing psychiatric interventions. In: J Neural Transm 2012; 119: 106175.
Thornicroft G, Brewin CR, Wing JK, eds. Measuring mental health 52 Mueser K, Glynn S. Have the potential benets of CBT for severe
needs. London: Gaskell, 1992: 16384. mental disorders been undersold? World Psychiatry 2014; 13: 25356.
35 Edinger JD, Wyatt JK, Olsen MK, et al. Reliability and validity of the 53 Edinger J, Wohlgemuth W, Radtke R, Marsh G, Quillian R.
DUKE structured interview for sleep disorders for insomnia Cognitive behavioral therapy for treatment of chronic primary
screening. Sleep 2009; 32: 0810. insomnia. JAMA 2001; 285: 185664.
36 Buysse DJ, Ancoli-Israel S, Edinger JD, Lichstein KL, Morin CM.
Recommendations for a standard research assessment of insomnia.
Sleep 2006; 29: 115573. Vol 2 November 2015 983