Behaviour Research and Therapy 42 (2004) 1377–1401
Is there evidence that cognitive behaviour therapy
is an effective treatment for schizophrenia? A cautious
or cautionary tale?
Nicholas Tarriera,?, Til Wykesb
aAcademic Division of Clinical Psychology University of Manchester, Education and Research Building
(2nd Floor), Wythenshawe Hospital, Manchester M23 9LT, UK
bDepartment of Psychology, Institute of Psychiatry, UK
Received 17 February 2004; received in revised form 4 June 2004; accepted 18 June 2004
Schizophrenia is a severe and disabling disorder with considerable psychological, social and economic
costs. Over the last 15 years there has been a significant development in the use of cognitive behaviour
therapy for psychosis (CBTp) in the treatment of schizophrenia, with 20 randomised controlled trials
having been published. The majority of this work has been with alleviating medication resistant symp-
toms in chronic patients, but preliminary work has also been carried out with speeding recovery in acute
schizophrenia and in relapse prevention and early intervention. A review of these studies indicates modest
effect sizes, with the strongest evidence available for chronic patients. There is evidence that the effect size
of the trials is significantly and negatively correlated to their methodological quality. We conclude cau-
tiously that overall there is good evidence for the efficacy and effectiveness of CBTp in the treatment of
# 2004 Elsevier Ltd. All rights reserved.
Keywords: Cognitive-behaviour therapy; Schizophrenia; Psychosis
Schizophrenia is a debilitating condition that affects 1 in a 100 of the population. It is tra-
ditionally characterised as a loss of contact with reality. Symptoms include disturbed perception
?Corresponding author. Tel.: +44-161-291-5883; fax: +44-161-291-5882.
E-mail address: firstname.lastname@example.org (N. Tarrier).
0005-7967/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
in the form of hallucinations, mainly auditory but also of the other senses, and disorders of
thought, including delusions. Schizophrenia is frequently associated with impairment of cognition
and emotion. Vocational and social functioning is often poor and is associated with consider-
able social and economic burden to the sufferer, their carers and society as a whole. The out-
come is variable, typically the disorder follows an episodic relapsing course with periods of
remission, although recovery is often incomplete and residual hallucinations and delusions are
common. Traditionally schizophrenia has been assumed to be impenetrable to psychological
therapies and treatment is typically by anti-psychotic medication in conjunction with some
type of case management. Many people are either treatment resistant to medication or con-
tinue to experience disruptive residual psychotic symptoms. In addition to the specific illness
burden depression and suicide attempts are common.
In the early 1950s treatment of schizophrenia changed dramatically from the predominantly
custodial with the introduction of chlorpromazine and later other anti-psychotic medications.
Biological thinking about schizophrenia and psychoses became the dominant model and bio-
logical treatments the established management approach. Over the subsequent three decades
there was scant interest in schizophrenia from the cognitive and behaviour therapy community,
although in 1952 Beck had published a case study of cognitive therapy in the treatment of delu-
sions (Beck, 1952) but in moving jobs he went on to establish cognitive therapy as a treatment
for depression (personal communication). Bellack in his AABT Presidential address termed
schizophrenia the ‘‘forgotten child of behaviour therapy’’ (Bellack, 1986) which reflected the
dominance within the adult CBT literature of the treatment of anxiety and depression. Perhaps
there was not much to be optimistic about. Psychotherapy, in the form of psychodynamic
psychotherapy was largely discredited as a treatment for schizophrenia (Mueser & Berenbaum,
1990) and behaviour therapy was mainly focussed on institutional care and rehabilitation,
expression of psychopathology was regarded as a behaviour to be extinguished and not of any
intrinsic interest (e.g., Liberman, Teigen, Patterson, & Baker, 1973; Meichenbaum & Cameron,
1973). However, there were developments within social psychiatry which began to challenge the
dominance of biological psychiatry in thinking about schizophrenia. In the 1980s consistent
findings that family environments, in terms of the measure of expressed emotion (EE) (Leff &
Vaughn, 1985) were strongly influential in determining the recurrence of schizophrenic symp-
toms promoted the development of family interventions that were successful in reducing relapse
(Pilling et al., 2002; Tarrier & Bobes, 2000). Many of these family interventions were strongly
behavioural (e.g., Barrowclough & Tarrier, 1992; Falloon, Boyd, & McGill, 1984) and
developed methods such as problem solving, stress management and goal setting which had
their origins in general adult mental health. Theoretical understandings of schizophrenia were
also changing with stress-vulnerability models that incorporated psychological and social ele-
ments being developed and empirically tested (e.g., Nuechterlein, 1987; Zubin & Spring, 1977).
This theoretical interest in the psychotic process itself and the more optimistic attitude that psy-
chosocial treatments could have a significant clinical benefit in the days of de-insitutionalisation
and community care set the scene for the development of CBT approaches to reduce the symp-
toms of schizophrenia. By the 1980s cognitive behaviour therapy had expanded enormously and
continued to do so. Building on successful treatment studies in affective disorder, CBT was
becoming the established treatment of choice for many non-psychotic conditions. It was inevi-
table that CBT would be tried as a possible treatment of schizophrenia.
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011378
CBT as a developing field in the treatment of non-psychotic disorders productively informed
the development of specialised CBT treatments for psychoses. Research into intrusions and
safety behaviour in anxiety disorder was transferred into theoretical models of psychosis (e.g.,
Morrison, 2001; Morrison, Haddock, & Tarrier, 1995) and broader cognitive models of
schizophrenia were also developed (e.g., Garety, Kuipers, Fowler, Freeman, & Bebbington,
Medication for the treatment of symptoms and the prevention of relapse and varied mental
health service configuration were the mainstay of the management of schizophrenia. CBT was
not developed as an alternative or stand alone treatment but as an addition to these established
management practices. CBT for schizophrenia and the psychotic disorders has now become
established, at least within the UK with the volume of interest sufficient that psychological
treatments of psychosis is a separate and one of the most popular streams at British Association
for Behavioural and Cognitive Psychotherapy (BABCP) conferences and for the past 12 years
has supported its own international conference (The Biannual International Conference of
Psychological Treatments for Schizophrenia). There has been an enormous expansion of interest
in CBT approaches to treating schizophrenia and other psychoses (which we will now refer to as
CBTp) since Bellack delivered his paper in 1986. This interest has produced an evidence base
which has been accepted as adequate for defining practice guidelines. The UK National Insti-
tute for Clinical excellence (NICE) included CBT for psychosis as a recommended treatment for
schizophrenia (NICE, 2003). However, it is pertinent for those of us in the field to reflect on the
current evidence and to identify shortcomings in the evidence base in order to improve the effec-
tiveness of treatment.
Clinical, ethical and economic considerations have encouraged clinical practice to be guided
by an evidence base which has been produced from evaluations of treatments. The evidence
develops from uncontrolled studies and small scale projects through to controlled studies and
then to large randomised controlled trials (RCTs) of efficacy and effectiveness. In spite of recent
criticisms of the appropriateness of randomised controlled trials in mental health (Richardson,
Baker, Burns, Lilford, & Muijen, 2000; Slade & Priebe, 2001), RCTs remain the gold standard
by which all treatments are judged (Doll, 1998; Pocock, 1996; Salkovskis, 2002). Once a data-
base of controlled trials has been established, meta-analysis can provide a measure of the aver-
age level of therapeutic effect across a number of different studies and also allows the effect of
the variation across studies to be investigated. The evaluations should also provide tests of the
theoretical underpinnings of the therapy in order to distinguish specific from non-specific effects.
Treatment evaluation has been a core characteristic of CBT which has given it significant
advantages over other types of psychotherapy which have not tried to establish empirical val-
idity (Tarrier, 2002). However, it is important not to accept evidence uncritically and to investi-
gate more thoroughly what conclusions can be confidently drawn. Large pragmatic multi-site
trials, common in some branches of medicine, are rare in CBT in which small single-site studies
tend to predominate.
For CBTp with schizophrenia a number of meta-analyses indicating that CBTp is effective in
treating positive psychotic symptoms in psychotic patients have been published (e.g., Gould,
Mueser, Bolton, Mays, & Goff, 2001; Pilling et al., 2002; Rector & Beck, 2001). These results
have provided the impetus for the inclusion of CBTp in regular clinical practice and the inte-
gration of such treatments within established mental health services. Despite guidelines on the
1379 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
conduct of clinical trials there can still be a large variability in methodology that can have an
impact on the obtained intervention effects. It is important to consider possible sources of
hidden bias and investigate them before trial results are combined within a meta-analysis (Gotz-
sche, 1989). The subject of this paper will be firstly, to examine the evidence for the efficacy of
CBTp for psychotic symptoms now that more trials have been completed; secondly, to see
whether it is possible to understand why CBTp is efficacious if there is evidence that it is so; and
thirdly, to investigate whether apparent treatment benefits can be attributed to any potential
source of bias and how treatment evaluation could be improved.
2. Is CBTp beneficial?
We have identified 20 clinical trials that investigate CBTp in patients with a diagnosis of
schizophrenia or schizophrenia spectrum disorders. These 20 trial treated a total of 739 patients
with CBTp, with a mean of 37 (SD ¼ 48, range ¼ 7 225) treated in each study. Sixteen of these
studies have been carried out within the UK with one each from Canada, The Netherlands,
Italy and the USA. Sixteen studies involve chronic outpatients, one with chronic inpatients and
three for patients hospitalised for an acute episode. Although the CBTp approach across all
these studies is similar there are differences in conceptualisation and clinical focus between dif-
ferent research and clinical centres. Furthermore, there are phase-specific aims of treatment
which will determine the nature and strategy of the intervention. For example, in chronic
patients with drug resistant symptoms, the aim will be to achieve further symptom reduction in
relatively clinically stable patients, whereas in acutely and floridly ill patients, the aim will be to
speed symptom recovery. How therapy is focussed and delivered will vary depending on these
phase-specific conditions. Studies have differed in their outcome measures, however, the vast
majority have used some measure of positive symptomatology which can be converted into a
common metric of treatment effect size thus making a comparison across studies possible. The
equation used by Gould and colleagues in their meta-analysis (Gould et al., 2001) was used to
calculate effect sizes from the different trials. This subtracted the mean of the control group
from the mean of the CBT group at post-treatment, and divided this by the standard deviation
of the control group at post-treatment. The control group was considered to be the TAU group
or a control adjunct treatment that had been hypothesised on an a priori basis to be inactive for
the main outcome. This provides an effect size based upon end point assessment at post-treat-
ment and does not take into account pre-treatment scores. This was justified because CBTp was
utilised as an adjunct to standard care so the outcome of interest was whether CBTp improved
the outcome achieved by standard care. Of the 20 studies identified, one study did not publish
data on positive symptoms and appropriate data were not available from the authors. Thus
data were only available from 19 studies. We were interested only in whether there was an effect
after treatment and not whether any effect was maintained, so we have not included any follow-
up data in these calculations. These studies have a mean effect size of 0.37(SD ¼ 0:39,
median ¼ 0:32) with a range between ?0.49 and 0.99. Using Cohen’s (1988) convention for cat-
egorising effect sizes, 14 (74%) studies achieve at least a small effect size, 6 (32%) at least a mod-
erate effect size and 3 (16%) a large effect size. Overall these studies indicate a modest effect size
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011380
in improving positive symptoms compared to standard psychiatric care (TAU), which is prob-
ably not surprising given the nature and severity of the disorder.
It is possible that publication bias could affect any conclusions on efficacy in that there is
always a bias towards the publication of positive results and studies finding no treatment effect
often remain unreported. To investigate whether a publication bias resulted in over-optimistic
conclusions Wykes, Tarrier, and Everitt (2004) applied the model for publication described by
Copas and Shi (2001) and showed that approximately 50 ‘‘file drawer’’ studies would need to
exist to reduce the estimated effect size to being not significant from zero. It is our opinion,
based on our knowledge of this research area that it is unlikely that these studies do exist. The
overall conclusion is therefore that there is a significant effect of CBTp on positive symptoms
and the analysis of this larger corpus of studies replicates the findings of previous analyses.
Although there is a general effect of CBTp, the original model for the therapy was developed
for patients who were chronic and had treatment resistant (i.e. medication resistant) symptoms.
A model of the maintenance of these symptoms on which this form of CBTp was developed
may not be appropriate either for the development of symptoms at early stages, the treatment
of acute exacerbations of symptoms that result in admission or for the prevention of these
relapses at times when the disorder is relatively stable. We have therefore divided the studies
into their target group and phases of the disorder.
2.1. Symptom management in chronic schizophrenia
In spite of maintenance medication, a considerable percentage of patients suffering from
schizophrenia continue to suffer persistent hallucinations and delusions that do not respond fur-
ther to medication. The majority of the CBTp studies, 17 out of 20, have been carried out with
patients suffering a chronic illness. These studies, for which data were available for 16, have a
mean effect size of 0.4 (SD 0.32, median 0.33) with a range from ?0.32 to 0.99.
2.2. Symptom recovery in acute schizophrenia
Three studies have investigated the use of CBTp in the treatment of acutely ill patients hospi-
talised for an acute psychotic episode. Because the participants are acutely ill and may well be
suspicious, agitated and unstable, the therapy is often implemented as a therapy envelope which
consists of a range of duration of therapy which can be delivered in a flexible manner. This
flexibility in therapy structure necessarily takes into account the mental state of the patient and
their tolerance of therapy at any point in time. These three studies have produced effect sizes of
?0.49, 0.12 and 0.93, indicating considerable variance in this small number of studies. The Soc-
rates study (Lewis et al., 2002) which is by far the largest and methodologically rigorous study,
recruited 309 early onset schizophrenic patients and produced an effect size of 0.12. Whereas the
other two studies were considerably smaller. The initial study by Drury and colleagues (Drury,
Birchwood, Cochrane, & MacMillan, 1996a, b) was innovative in that it demonstrated that
CBTp could be utilised with acutely ill and disturbed patients admitted to hospital for a florid
episode. They reported a large relative decrease in symptoms and a significantly speedier recov-
ery in the treated group. However, the methodology adopted in this study may have resulted in
bias. A replication with improved methodology was not successful (Haddock et al., 1999). The
1381 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
large multi-site Socrates study, which investigated whether CBTp implemented as soon as poss-
ible in recent onset schizophrenic patients provided modest evidence that recovery was slightly
faster in the CBTp group compared to TAU but not compared to a supportive counselling con-
trol group (Lewis et al., 2002). One of the difficulties in demonstrating an advantage of an
adjunct treatment in acutely ill patients, especially in medication naive first episode patients, is
that standard treatment (medication) is very effective at reducing florid symptoms and tends to
mask any other treatment effects. So that there was a strong and consistent treatment effect for
all patients over the first five weeks although considerable variability in outcome occurs after
this time. However, there was little evidence that CBTp protected against future relapse over the
next 18 months. Although some evidence that patients who received TAU alone had signifi-
cantly more residual symptoms (Tarrier et al., 2004).
2.3. Relapse prevention
A number of studies have investigated relapse prevention or the ability of CBT to prevent or
delay future acute episodes. Relapse is an important outcome because of the disruption and dis-
tress that symptoms exacerbation brings. CBTp was developed to reduce persistent symptoms.
Its use as a prophylactic in averting relapse was a later modification and the therapeutic empha-
sis is different. In studies of CBTp interventions where relapse prevention is just one of a series
of components achieved little success but when CBTp is focused on, and dedicated to, relapse
prevention it has had some success. Relapse rates with varying definitions are presented in
Table 1 for comparison.
2.4. Early intervention1
In addition to the effect on more established psychosis there has recently been a growing
interest in diverting the course in schizophrenia at an early stage. Schizophrenia is often appar-
ent in its first acute form during adolescence and it is thought that this early presentation affects
Relapse rates in CBT studies in up to 12 months follow-up
TAU (%) CBT (%)
Studies with relapse prevention as a component
Tarrier et al. (1999)
Durham et al. (2003)
Startup, Jackson, and Bendix (2004)
Tarrier et al. (2004)
Studies dedicated to relapse prevention
Bach and Hayes (2002)
Gumley et al. (2003)
1We have not included these early intervention results in our estimation of effect sizes as they are conceptually very
different in attempting to prevent transition to psychosis rather than treatment of established psychosis.
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011382
the developmental course of the disorder such that vital skills are never learnt and educational
and social relationship opportunities missed. Morrison and colleagues (in press) have just
reported a study of using cognitive behaviour therapy techniques in this early group to try to
avert or postpone the first acute episode of the disorder by intervening during a prodromal
period. Their technique focuses not on frank positive symptoms but on problem solving diffi-
culties that are felt to be important by their clients. The results of this first randomised control
trial look promising. The CBTp group proved more beneficial than the TAU group in prevent-
ing progression in to psychosis, preventing the prescription of anti-psychotic medication and in
reducing symptoms. Fig. 1 shows the three different outcomes in the two treatment groups.
3. What are the predictors of outcome?
3.1. Within patient variables
Patients with schizophrenia represent a heterogeneous group of whom 20–45% can be treat-
ment resistant (Kane, 1996, 1999) and 5–10% show no benefit from anti-psychotic medication
(Pantelis & Barns, 1996). Although, there is confusion between the terms treatment resistant,
incomplete recovery and treatment intolerance it is clear that treatment is incomplete in a sig-
nificant number of patients in spite of well publicised advances in anti-psychotic medication. It
is important, therefore, to understand what factors predict a good response to psychological
treatment and perhaps the converse, which patients do not derive benefit in order to tailor and
refine treatments. Factors that have been associated with poor outcome include negative symp-
toms of affective flattening and alogia (cognitive impoverishment) (Tarrier, 1996; Tarrier et al.,
Fig. 1. Percentage of patients in CBT and TAU who showed a transition into psychosis as measured by PANSS
symptoms, prescription of anti-psychotic medication and DSM IV diagnosis of psychosis (from Morrison et al., in
1383 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
1993). Factors associated with better outcome include a shorter duration of illness, less severe
symptoms at pre-treatment (Tarrier et al., 1998) and receptiveness to hypothetical contradiction
(Gartey et al., 1997).
3.2. Non-specific effects associated with any therapy
A number of studies have included non-specific control groups in addition to TAU to control
for factors assumed to be common to all individual psychological therapies. These have
included supportive counselling (Durham et al., 2003; Lewis et al., 2002; Tarrier et al., 1998),
befriending (Sensky et al., 2000) and problem solving (Tarrier et al., 1993). It has been generally
assumed that because these treatments did not include any systematic use of cognitive behav-
ioural techniques to reduce symptoms they would be therapeutically inert. That they would
have little impact on psychotic symptom outcome measures and therefore perform at the same
level as TAU. Although some differences are found between CBT and the non-specific controls
at some assessment points these are not consistent across studies (see Sensky et al., 2000; Tarrier
et al., 1998, 1999, 2000). There is some evidence that non-specific controls do more poorly on
measures of hallucinations (Lewis et al., 2002; Tarrier et al., 2001). But not one study has
shown clear and significant overall differences between CBT and the non-specific control
One difficulty in understanding the better than expected performance of non-specific control
groups is that the treatment was unstructured so it is difficult to know exactly what was work-
ing and why. Tarrier et al. (2000) has speculated that supportive counselling may have provided
a non-threatening social interaction of sufficient frequency and regularity with a skilled therapist
which improved the patient’s feelings of self-worth and encouraged them to enter the social
A non-active treatment control has often been included in order to distinguish specific effects
from those in a placebo. All psychological treatments suppose that those receiving them will
have some expectation of their efficacy. The placebo has now been investigated in some detail
and has been shown to have both psychological as well as biological effects (Di Blasi, 2003).
The requirement of a placebo condition in medication treatment trials is obvious as it is impor-
tant to know whether an active ingredient that is costly and might have potential risks does pro-
vide adequate benefits to offset any risks. However, this argument is specious in psychological
treatment as all the supposed non-active treatment controls include essential elements of
psychological treatment, for example warmth, supportive and empathic listening and some
expectation that treatment will be beneficial.
Expectancy levels as non-specific treatment elements on the efficacy of CBTp have been
manipulated in one study (Tarrier et al., 1993) but no differences were found between the two
expectancy conditions of immediate and delayed expectation of treatment benefit. However,
there was no direct measurement of the effect of the expectancy manipulation on the patient’s
own expectancies of the therapy effects. In the treatment of anxiety disorders the patient’s actual
ratings of expectation of benefit is related to improvement (e.g., Chambless, Tran, & Glass,
1997) and it seems likely that this expectancy effect may contribute to the non-specific effect of
therapy noted in many of the CBT studies. Thus, it is not clear whether expectation of benefit
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011384
had no effect or whether the experimental manipulation was too weak in Tarrier et al.’s (1993)
Di Biasi, Harkness, Georgiou, and Kleijnen (2001) makes the useful distinction of cognitive
and emotional care as constituents of the placebo response. Cognitive care aims to influence the
patient’s expectations about treatment whereas emotional care refers to the style of the therapist
interaction which aims to reduce negative feelings. So even where expectations do not bring
about an effect on outcome the emotional aspects of the interaction could bring about positive
There is evidence to show that people with schizophrenia are highly sensitive to aspects of
their emotional environment. Research over the last three decades has consistently found that
schizophrenic patients who live with relatives or carers rated as high on EE, who are critical,
hostile or emotionally over-involved (Leff & Vaughn, 1985), have significantly higher relapse
rates than those living with low EE relatives (for reviews, see Bebbington & Kuipers, 1994;
Butzlaff & Hooley, 1998; Kavanagh, 1992). Exposure to the presence of a high EE relative has
physiological consequences for the patient (Tarrier & Turpin, 1992). The measure of EE which
has been used extensively to assess the quality of the relationship with informal carers has also
been applied to professional carers. Although this line of research is much less well developed
than with families there is accumulating evidence that high EE responses in professionals is
associated with negative outcomes (Kuipers, 1998). In a prospective study of case managers,
Tattan and Tarrier (2000) found that the absence of a positive relationship between case man-
ager and patient measure three months after engagement was associated with poorer clinical
outcome nine months later independently of initial severity of illness. A positive relationship
was also significantly associated with a greater improvement in the patient’s quality of life and a
negative relationship with the least improvement. These effects occurred even when the actual
duration of contact between patient and case manager was quite low and much less than is gen-
erally found between patient and relative. Thus social interactions previously construed to be
non-specific or even inert aspects of a psychological placebo may have significant active effects,
at least with schizophrenic patients who are highly sensitive to interpersonal environments.
Because therapeutic interactions involve processes that may be termed placebo but which may
affect outcome, it was perhaps incorrect and naive to expect control conditions, such as support-
ive counselling, to be inert. If such control conditions did exert a placebo effect then this would
be demonstrated in the comparison of the control condition with TAU. This comparison would
demonstrate the size of the non-specific or placebo effect (Di Blasi, 2003; Hrobjartsson &
3.3. Effect of different models of therapy
Cognitive behaviour therapy is not a unitary model. In different disorders, different mecha-
nisms are suggested and different weights given to different elements in the model. The general
model for CBTp for psychosis is also variable and also differs in the emphasis on the cognitive
and/or the behavioural dimensions of therapy. At the extreme end of the continuum some ver-
sions of cognitive therapy focus on the historical formation of putative schema and seem close
to some forms of psychodynamic treatments, whereas others concentrate on mechanisms that
are proximal to the symptoms and behaviours that are under scrutiny. It is possible to rate or
1385 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
rank studies depending upon the emphasis that the implemented therapy gives to behavioural or
cognitive techniques. Although this is a crude analysis it may provide some idea as to what are
the effective elements of CBTp. Results obtained from an analysis when fewer trials were avail-
able reported a correlation of 0.9 between the orientation of the treatment and treatment effect
size with behavioural trials producing larger effect sizes (Tarrier & Wykes, 2004). In an attempt
to investigate the relative importance of behavioural and cognitive therapeutic procedures, the
19 CBT trials for which data were available at the time were independently ranked by a blind
rater on a dimension of behavioural to cognitive procedure, in which the pole end of the cogni-
tive dimension was an attention to historical information and schema formation. This ranking
was correlated with the study’s effect size and a trend (p ¼ 0:1) was found in favour of the more
behavioural studies (Wykes et al., 2004). This suggests that behavioural change, whether it is in
the setting of behavioural goals or in behavioural experiments, is a very important, possibly
essential, aspect of the therapeutic procedure. This is further supported by Bennett-Levy’s
(2003) report that in a non-clinical group of participants, behavioural experiments are rated as
more powerful in bringing about therapeutic change than automatic thought recording.
Although this study was not with psychotic patients it demonstrated that behavioural experi-
ments produced greater change in both beliefs and behaviour.
4. How can we explain the variability of outcome?
There is considerable heterogeneity of treatment response in the patients treated with CBTp
not just between studies but also within studies. Possible explanations include:
. variability within the therapy,
. characteristics of the patient, and
. variability within the trials.
4.1. Variability in therapy
Variability in the therapy may include the method of delivery of CBT, the model or variant
of CBT, the characteristics of the therapist and the interaction between therapist and patient,
specifically relating to therapeutic alliance.
4.1.1. Standardised vs individualised treatment protocol
There are two general ways of carrying out therapy. Protocol-driven therapy is provided
where therapy techniques are implemented from a standardised treatment manual. This stand-
ardises treatment but offers little flexibility to respond to individual differences in patients. Sec-
ondly, case formulation-driven therapy in which therapy is individualised and is based upon a
specific and idiosyncratic assessment of the patient (see Tarrier & Calam, 2002). This termin-
ology is slightly confusing since it is quite possible to have a manual or protocol that informs
on how to deliver case formulation-driven therapy. Here we take protocol-driven therapy to
imply the absence of an individualised assessment and the use of a strict modular treatment
application. It is generally assumed that the case formulation approach provides an advantage
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011386
when treating complex clinical cases, such as might be expected with psychotic patients, because
complicating and idiosyncratic factors can be accommodated within the formulation whereas
they may be ignored but become inhibitory in a strict application of a protocol. However, the
evidence for this advantage is sparse and it is difficult to envisage an adequate comparison of
the two approaches being feasible.
Although case formulation has been described as ‘‘the cornerstone of CT for psychosis’’
(Chadwick, Williams, & Mackenzie, 2003) the evidence that it enhances the therapeutic relation-
ship or brings about symptomatic improvement is equivocal at best (Chadwick et al., 2003).
However, it may be unsafe to assume that even a rigorous and inflexible protocol will result in
standard implementation of the treatment method.
4.1.2. Models and procedure in therapy
Research groups have developed different approaches to therapy which vary, both between
groups and over time. Therapeutic procedures have focused upon a range of targets including
positive symptoms in general, hallucination or delusions separately, the emotional consequences
of symptoms, underlying beliefs and putative schematic processes. These differences and varia-
tions may contribute to heterogeneity of outcomes, although there is little understanding at
present as to what particular approaches or techniques are effective or whether one model is
superior to any other. Furthermore, variability between patients could greatly exceed any varia-
bility in outcome introduced by different brands of treatment although in general the power of
more proximal or behavioural methods to change thinking styles does seem to be gaining credi-
This search for ‘‘active ingredients’’ has yet to be initiated as a research programme although
is has been marked up for future research (Pilling et al., 2002). However, faith in the productive
result of such a programme may be misplaced. If CBT is regarded as an approach rather than
just a set of techniques which may or may not be included then different conclusions can be
drawn. CBT as an approach implies there is a shared ‘‘world view’’ about the nature of therapy
and how clinical problems should be addressed which includes shared assumptions about the
origins of psychopathology and causality. This can be thought of as an adherence to a general
model and thus therapy and therapeutic actions will either be within the ‘‘spirit of CBT’’ or not.
An analogy can be drawn with motivational interviewing which is regarded very much as an
approach and not a treatment and emphasis is placed upon the ‘‘spirit’’ of MI (Miller & Roll-
nick, 2002: Chapter 4). Thus it maybe that adherence to the general ‘‘spirit of CBT’’ may be
more important than which specific individual or collection of techniques are included. The
CBT approach may encourage the patient into thinking and behaving in a way that is consistent
with the CBT model. This may have implications for training staff and dissemination as acqui-
sition of the general principles or ‘‘spirit’’ may be more difficult, requiring a more detailed
theoretical knowledge and understanding, than acquiring the use of a number of specific techni-
4.1.3. Variability in therapist competence
In cognitive therapy for major depression, even when the implementation of protocol-driven
therapy occurred in the context of rigorous training, the use of a treatment manual and adher-
ence checks, there was considerable variation in the application of treatment (Malik, Beutler,
1387 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
Alimohamed, Gallagher-Thompson, & Thompson, 2003). It is therefore of prime importance to
measure therapy delivery.
There is little evidence of what constitutes essential, necessary or important elements of ther-
apy, although we can all have a plausible guess. Scales that have been developed to measure
universal and active ingredients of CBT do not weight different elements on their importance to
outcome (Cognitive Therapy Scale (CS), Blackburn et al., 2001; Young & Beck, 1980). Simi-
larly, scales developed to measure therapist’s skills in CBTp suffer from these problems, such as
the Cognitive Therapy Scale for Psychosis (CTS-Psy) (Devane et al., 1998; Haddock et al.,
2001) and the Cognitive Therapy for Psychosis: Adherence Scale (Startup, Jackson, & Pearce,
2002). In fact scales for CBTp therapist’s skills differ in the items included depending on their
deemed importance in their therapy variant.
These scales blur the measurement of fidelity, adherence to protocol, with an assessment of
the competence of the therapist. An adequate scale must include and be able to quantify at least
three subscales: essential non-specific interpersonal aspect of therapy (e.g., warmth and empa-
thy); adherence to the treatment procedure; and, individual treatment components. There is no
empirical evidence to link particular items or competence with clinical outcome, and such a
scale would be essential in investigating the importance of therapy process across studies.
It is worth noting that in most of the CBTp trials the therapists if not expert in the field have
been well qualified and experienced. It can be argued that it is important to include experienced
therapists in the delivery of any psychological therapy undergoing evaluation. The level of
expertise required for a significant therapeutic effect still requires investigation (e.g., Turkington,
Kingdon, Turner, & the Insight Group, 2002).
4.1.4. Variability in therapist–patient interaction
There is considerable evidence in the general therapeutic literature that the therapeutic
alliance has a consistent but modest association with outcome (Constantino, Castonguay, &
Schut, in press; Horvath, 1994). This appears to be a predictive association and not an artefact
of improvement (Klein et al., 2003). In the Socrates, trial therapeutic alliance was rated by both
patients and therapist. There was a significant but moderate correlation between these two
measures from different sources of r ¼ 0:42. Interestingly, patients who received CBTp rated the
therapeutic alliance as stronger than those who received supportive counselling, whereas the
therapists’ ratings of alliance showed no differences between therapies. The patients’ ratings of
stronger therapeutic alliance after four treatment sessions were also significantly associated with
reduced positive psychotic symptoms at 18 months follow-up, whereas there were no such asso-
ciations with therapists’ ratings of alliance (Bentall et al., 2003).
4.2. Variability in trial methodology
There are now established conventions concerning how clinical trials should be carried out
and analysed (Everitt & Pickles, 1999; Pocock, 1996), these have been formalised in the CON-
SORT Statement (Begg et al., 1996; Moher, Schultz, & Altman, 2001). Although, psychology
journals have been slow to adopt these conventions. Randomised controlled trials are the estab-
lished scientific method to minimise bias in evaluating treatment efficacy. For instance design
features, such as randomisation and blinding (masking) diminish bias although do not guaran-
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011388
tee freedom from it. Investigating medical trials, Moher et al. (1998) found that trials with
poorer quality masking of allocation of treatments were associated with an increased estimate of
benefit of 34% in a variety of different treatments in circulatory and digestive disease, mental
health, obstetrics and childbirth. This replicated a similar earlier finding of Schulz, Chalmers,
Hayes and Altman (1995) who also reported 30–40% exaggerated estimates of treatment efficacy
with inadequate concealment of randomisation and assessment.
4.3. Sources of variability in clinical trials
4.3.1. Sample characteristics
The method of participant selection and recruitment can introduce bias (Pocock, 1996;
Tarrier, 2001), especially with psychiatric patients who are generally difficult to engage and
treat. Due to patient heterogeneity in potential response to psychological treatment (Farhill &
Voudouris, 1996) sampling or recruitment pathways can also potentially elevate treatment
effects by including only ‘‘the best bets’’ who will be easier to engage and treat successfully.
Thus, it is important to ensure that the study sample is as representative as possible of the gen-
eral population of patients with that specific disorder and not atypical due to the process of
recruitment. For example, recruitment of volunteers by advertisement or referrals of ‘‘suitable’’
patients by clinicians will not produce as representative a sample as a geographic or epidemio-
logical cohort. Representative samples are essential to treatment generalisation and this may be
more of an issue when conducting trials in countries without universal health care (Tarrier &
Wykes, 2004; Wykes, Tarrier, & Lewis, 1998). Universal health care systems, such as the UK
National Health Service provides an ideal test bed for clinical trials because the majority of
possible recruits are likely to be present in the services for a catchment area thus allowing com-
parisons to be made between different geographical areas which have different epidemiological
Further bias may be introduced by the duration and volume of the recruitment process in
each centre. Patients who are more amenable to trial recruitment and treatment are likely to be
recruited earlier so that the longer recruitment goes on in any centre the more treatment resist-
ant patients are likely to be recruited. This is especially noticeable in studies in which a small
number of patients are recruited from a large number of sites, such as some drug trials, which
may cream off the most treatment amenable patients from a large number of centres.
In addition sample size is an important aspect of any trial. Sample sizes should be based on
adequate power calculations, although Kazdin and Bass (1989) make specific recommendations
of the number of participants in each treatment limb. Furthermore, people with schizophrenia
are not stable over time even with standard care this affects the size of the groups. This is in
comparison to CBT trials of more stable conditions such as social anxiety.
4.3.2. Allocation to treatment
The process of randomisation to treatment group has been central to clinical trials since the
late 1940s. It ensures that there is no systematic bias resulting from the allocation to treatment
process. However, ‘‘if the randomisation is not performed correctly then there is every danger
that the trial might be just as biased as the non-randomised trials...’’ (Pocock, 1996: p. 65). The
process of random allocation needs to be appropriate and clearly described (Schulz, Altman, &
1389 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
Moher, 2002). Methods such as allocation on the basis of date of birth or day of attendance are
not considered appropriate (Moher et al., 1998). The important aspect is that the mechanism of
allocation (e.g., tossing a coin, table of random numbers, and computer algorithms) prevents
selection bias and the sequence is concealed from the research team. A lack of an adequate con-
cealment of the allocation sequence that prevents investigators knowing in advance what treat-
ment the next patient will receive has been shown to be associated with larger treatment effects
(Altman et al., 2001). The effectiveness of various methods of concealment such as the use of
serially numbered, sealed, opaque envelopes or independently computer generated allocation
sequences have been debated (Balk et al., 2002; Juni, Altman, & Eggar, 2001; Schulz et al.,
2002) because even when concealment is attempted allocation schedules may be deciphered
(Juni et al., 2001; Schulz, 1995).
4.3.3. Assessment of outcome
In order for assessments of outcome to be unbiased they need to be carried out by assessors
independent of those carrying out the therapy and these assessors should be unaware of which
treatment participants have received or been allocated to (normally called blinded or masked
assessment). All the procedures used to maintain masking should be clearly described and veri-
fied at the end of the study by asking assessors to guess allocation (Basoglu, Marks, Livanou, &
Swinson, 1997), although they should be unaware that they will be required to perform this task
beforehand so as not to prime them.2
Some outcomes are easy to assess and verify because they can be obtained from administrat-
ive data (e.g., admission to hospital obtained from hospital records) but most complex treat-
ments are assessed through their direct effects on symptoms or functioning and therefore require
the use of an instrument. These instruments should be appropriate, reliable and valid. Marshall
et al. (2000) found that the use of unpublished rating scales was a significant source of bias in
schizophrenia trials and resulted in elevated treatment effects. This was especially true in trials
of psychosocial interventions where unpublished scales had an odds ratio of 2.6. That is, unpub-
lished scales resulted in an increased effect size more than twice that of studies using published
There is often a conflict between psychologists, who wish to assess outcome in a range of
domains as well as measuring process variables, mechanisms of change and predictors of out-
come, and trial methodologists, who advocate deciding on the key outcomes and keeping these
to a minimum (e.g., Johnson, 1998). Indeed, in schizophrenia research, it has often been advo-
cated that due to the multiple dysfunctions resulting from the disorder, assessment across out-
come domains is a necessity (e.g., Barrowclough & Tarrier, 1984). It is certainly essential in
testing the theoretical effects of treatment and in aiding treatment developments. The difficulty
with an excessive number of assessments is that, besides not having a clear decision on the key
and important outcome, hence risking results which are difficult to interpret, the process of
2If assessors guess correctly an alternative explanation might be that they are recalling those people who improve
rather than not being blind to allocation. This might not be a major problem in these studies as the outcome meas-
ures are continuous measures of symptoms and therefore it is less likely that improvements over baseline can be
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011390
assessment may be burdensome to the participants and result in high rates of attrition and lost
data, thus risking a differential filter and introducing bias.
There is a further issue relating to assessment over the progression of treatment which may
reduce the measured treatment effect. Many patients suffering from psychotic disorders will not
reveal their symptoms to people they do not know well or trust because they are suspicious of
others or for other reasons. This may result in early baseline assessments being quite conserva-
tive while later assessments when the patient becomes more familiar with, and trusting of, the
assessor being more expansive. This phenomenon can artificially inflate post-treatment assess-
ments of symptomatology and thus under-estimate the actual effect of treatment.
4.3.4. Control groups
The use of appropriate controls is necessary to eliminate effects other than treatment which
could be responsible for any improvements (Foulds, 1958; Pocock, 1996: p. 53). CBTp is used
as an adjunct treatment in addition to standard psychiatric care, a control treatment that
includes standard psychiatric care or treatment as usual is therefore a prerequisite (i.e. CBTp
plus TAU is more effective than TAU alone). However, TAU poses a further problem as stan-
dard care will develop over time and represents a moveable baseline as organisational manage-
ment (e.g., case management, assertive outreach, vocational programmes) and medication (e.g.,
atypical anti-psychotics) improve or are introduced.
As we have already discussed the non-specific, sometimes erroneously termed placebo effect in
psychological therapy, needs to be measured. Control treatments need to be credible and not be
subject to excessive attrition.
4.3.5. Description of treatments
Treatments should be described so that they can be independently replicated, this would be
aided by a manual or protocol. As previously discussed assessment of adherence to the treat-
ment protocol or some method of treatment quality assessment should be carried out to ensure
not only that the protocol was followed but also to ensure that there is a consistency of treat-
ment across different individuals.
The results should be analysed on an analysed-as-randomised (sometimes referred to as an
intention to treat analysis, ITT) irrespective of what the participants actually received. This
approach is recommended as it maintains the benefits of randomisation (Everitt & Pickles, 1999:
pp. 45–46). Participant attrition is an important factor and a significant problem in treating
people with schizophrenia. Data from patients lost to follow-up is unlikely to be lost at random
with the attendant risk that a differential filter will produce over-optimistic results. How missing
data, for example from withdrawals and drop outs, are handled in the analysis will affect the
validity of the inferences drawn from the statistical analysis, and large numbers of drop outs
would be considered to make any beneficial results less credible. Descriptions of statistical good
practice are available (e.g., Everitt & Pickles, 1999). Some commonly used procedures are not
considered to be appropriate, e.g., missing value substitution by ‘‘last observation carried for-
ward’’ because of its underlying assumptions and likely optimism about the precision of the
extracted effect size (Everitt & Pickles, 1999: pp. 125–128). On the other hand, where com-
pliance with treatment is poor and those who receive treatment are far fewer in number than
1391 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
those who were offered treatment, ITT analyses may mask important treatment effects, in that
the opportunity for compliance may be different between the treatment and control groups.
Dunn et al. (2003) has suggested the use of complier average causal effect (CACE) estimation as
a method of dealing with this in trial data analysis.
Where a control group for non-specific therapist effects has been included, we recommend an
analysis of the difference between the control and the TAU group as a specific measure of the
4.4. Assessing quality and variability in psychological treatment trials: the Clinical Trial
Assessment Measure (CTAM)
There are many instruments that have been designed to assess trial quality but they often suf-
fer from methodological inadequacies (Juni, Witschi, Bloch, & Eggar, 1999; Moher et al., 1995)
and are in the most part designed to assess medical or pharmacological trials and do not
adequately cover the design features important in psychological trials described earlier in this
paper. Since there appear to be no dedicated scales for assessing the quality of trials of psycho-
logical treatments in mental health, we designed the Clinical Trials Assessment Test (CTAM)
specifically for this purpose. A list of relevant features was extracted from the CONSORT
guidelines (Moher et al., 2001), a review of 25 trial assessment scales (Moher et al., 1995), and
expert opinion was then sought from psychologists, psychiatrists, statisticians and method-
ologists. These opinions provided face validity. Individual features were differentially weighted
based on previous data on methodological characteristics that can influence outcome (e.g.,
Chalmers et al., 1981; Jadad et al., 1996; Juni et al., 1999, 2001; Kazdin & Bass, 1989; Marshall
et al., 2000; Moher et al., 1998; Sterne et al., 2002). The resulting list had 15 items grouped into
six areas of trial design: sample size and recruitment method; allocation to treatment; assess-
ment of outcome; control groups; description of treatments; and analysis (see Table 2).
The CTAM showed good blind inter-rater agreement of 0.96 and adequate internal consist-
ency (Cronbach’s alpha 0.691). Concurrent validity was assessed by correlation with scores from
three other scales devised for the generic assessment quality of trials (Wykes et al., 2004).
Moher et al. (1995) recommended the scale published by Jadad et al. (1996) as being the most
robust instrument available hence this was used. In addition, we selected two scales, Brown
(1991) and Chalmers et al. (1981) which were more conservative and less conservative than the
Jadad scale based on the evaluations in Juni et al. (1999). The correlations of the scales with the
CTAM scores were as follows: CTAM and Jadad, q ¼ 0:97, p < 0:001; CTAM and Chalmers,
q ¼ 0:93, p < 0:001; CTAM and Brown, q ¼ 0:79, p < 0:001. This indicates that CTAM had
excellent concurrent validity.
5. Trial quality and effect sizes
We wished to investigate whether there was an association between the effect size of the
CBTp trials and the rating of methodological quality of the trials. We did not want to confuse
ratings of the quality of the methodology with the quality of the report (Huwiler-Muntener,
Juni, Junker, & Eggar, 2002; Moher et al., 1995) so any issues were clarified with the trial
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011392
researchers and where necessary data or information were requested that were not available
within the written report. Subsequently, we made the CTAM ratings of the individual trials
available to the trial researchers and asked for their comments.
The maximum score for the CTAM is 100 and in this sample of 19 CBTp for psychosis trials
the mean score was 63.1 (SD 18.0) with a median of 56 and a range of 39–100. The association
between CTAM scores and trial effect sizes is demonstrated in Fig. 2. There was a significant
negative correlation between effect size and CTAM score (q ¼ ?0:58, p ¼ 0:009) indicating, in
line with reports in the medical literature, that studies that employ poorer methodology over-
estimate the magnitude of positive benefits of treatment.
There was considerable variability in methodology. All, except one trial, had random allo-
cation and two studies did not have independent assessments of outcome measures (but they
The Clinical Trials Assessment Measure (CTAM)
Clinical Trials Assessment Measure (CTAM)
Sample—two questions: maximum score ¼ 10
Q1: is the sample a convenience sample (score 2) or a geographic cohort (score 5), highly selective sample, e.g., vol-
unteers (score 0)
Convenience sample—e.g., clinic attenders, referred patients or Geographic cohort—all patients eligible in a parti-
Q2: is the sample size greater than 27 participants in each treatment group (score 5) or based on described and
adequate power calculations (score 5)
Allocation—three questions: maximum score ¼ 16
Q3: is there true random allocation or minimisation allocation to treatment groups (if yes score 10)
Q4: is the process of randomisation described (score 3)
Q5: is the process of randomisation carried out independently from the trial research team (score 3)
Assessment (for the main outcome)—five questions: maximum score ¼ 32
Q6: are the assessments carried out by independent assessors and not therapists (score 10)
Q7: are standardised assessments used to measure symptoms in a standard way (score 6), idiosyncratic assessments
of symptoms (score 3)
Q8: are assessments carried out blind (masked) to treatment group allocation (score 10)
Q9: are the methods of rater blinding adequately described (score 3)
Q10: is rater blinding verified (score 3)
Control groups—one question: maximum score ¼ 16
Q11: TAU is a control group (score 6) and/or a control group that controls for non-specific effects or other estab-
lished or credible treatment (score 10)
Analysis—two questions: maximum score ¼ 15
Q12: the analysis is appropriate to the design and the type of outcome measure (score 5)
Q13: the analysis includes all those participants as randomised (sometimes referred to as an intention to treat
analysis) (score 6) and an adequate investigation and handling of drop outs from assessment if the attrition rate
exceeds 15% (score 4)
Active treatment—three questions: maximum score ¼ 11
Q14: was the treatment adequately described (score 3) and was a treatment protocol or manual used (score 3)
Q15: was adherence to the treatment protocol or treatment quality assessed (score 5)
where the criterion is not reached for any question score ¼ 0
Total score: maximum score ¼ 100
1393 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
were different studies). However, few trials reports adequately described the process of assessor
blinding or verified blinding at the end of the study. More than half of the studies did not use a
statistical method that was judged to take satisfactory account of drop outs from assessment.
In a regression analysis of the association of the six CTAM subscales with the study effect
sizes, the assessment subscale was the only scale significantly and independently associated with
study effect size (F ¼ 4:86, p ¼ 0:042) explaining 17.6% of the variance.
One of the main questions of interest was whether blinded and unblinded studies lead to dif-
ferent effect sizes, in particular whether the latter were over-optimistic about the effect of CBT.
To address this question a random effects model was applied separately to the blinded and
unblinded studies (Wykes et al., 2004). For the blinded studies, the estimated effect size was
0.208 (SE 0.077); the corresponding figure for the unblinded studies was 0.540 (SE 0.143). The
estimated 95% confidence interval for the difference is [0.12, 0.54]. It appears that unblinded
studies are likely to be optimistic about the effects of CBTp, with effect sizes of 50–100% higher
than those found in blinded studies.
It is highly probable that this is a general finding across all areas of clinical treatment and we
strongly suspect across cognitive behaviour therapy for all disorders and not just schizophrenia.
To our knowledge this type of quality assessment has not been attempted before in any area of
cognitive behaviour therapy. However, if psychological treatments are to be recognised as effec-
tive and their introduction into mental health services reliant upon an evidence base, then every
attempt must be made to improve the quality of the evaluation methods to obtain a precise and
accurate assessment of the treatment benefits. We do not wish to conclude that CBTp for posi-
tive psychotic symptoms is ineffective, far from it we are strongly encouraged that there is a
Fig. 2. Plot of effect size of CBTp trials with CBTp trials rank ordered on their CTAM scores.
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011394
consistency across these trials indicating a positive benefit of CBTp with what is a very difficult
disorder to treat. However, we wish to caution against exaggerated claims of the magnitude of
treatment benefit and emphasise that psychological treatments need to be evaluated with the
high degree of rigour comparable with branches of physical medicine and pharmacological
treatments. Thus, although, meta-analyses have been published on CBTp in psychosis (Gould
et al., 2001; Pilling et al., 2002; Rector & Beck, 2001), it is important to investigate potential
sources of bias within individual trials rather than simply combining their results into a meta-
analysis as this variation in quality can lead to bias and reduction in precision of the estimates
of the therapy’s effectiveness (Detsky, Naylor, O’Rourke, McGeer, & L’Abbe, 1992; Gotzsche,
One of the major issues in clinical research is whether treatments that have been validated in
clinical trials can be rolled out into clinical practice. In fact in many trials funded from public
funds it has been necessary to show that the treatment if successful could be feasibly assimilated
into standard health service practice. A number of studies have been reported on the use of
CBTp to treat psychotic patients in routine clinical practice. For instance, Turkington and
Kingdon (2000) demonstrated that CBTp could be administered by general psychiatrists work-
ing in the public health services, in spite of their heavy work load, with significant benefits to
their patients. Following specialist training, community psychiatric nurses can deliver effective
CBTp (Lancashire et al., 1997; Turkington et al., 2002).
One issue that has been raised is whether there is a trade off between quality of therapy and
accessibility of therapy. The training of less expert staff will increase the accessibility of the
treatment, however, this may be at the cost of a reduced quality of treatment and therefore,
reduced effect size. How expert or experienced a therapist needs to be to deliver effectively
potentially complicated therapeutic procedures is an empirical question.
Is there evidence that CBTp is an effective treatment for schizophrenia? Of course there is! It
is a cautionary tale for all cognitive behavioural treatments because of hidden biases introduced
by different methodological elements. We need to be cautious because there are many character-
istics of CBTp and its delivery which are likely to be important but as yet are unquantified. The
next generation of studies of CBTp need to concentrate on the identification and measurement
of these characteristics in large trials where their relative effects on outcome can be measured. It
is only through the continued closer examination of treatment results that the effectiveness of
CBTp can be increased and the lives of those who suffer from schizophrenia be improved.
1395 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
Altman, D. G., Schulz, K. F., Moher, D., Eggar, M., Davidoff, F., Elbourne, D., Gotzsche, P. C., Lang, T., & The
CONSORT Group. (2001). The revised CONSORT statement for reporting randomised trials: explanation and
elaboration. Annals of Internal Medicine, 134, 663–694.
Bach, P., & Hayes, S. C. (2002). The use of acceptance and commitment therapy to prevent the rehospitalisation of
psychotic patients: a randomised controlled trial. Journal of Consulting and Clinical Psychology, 70, 1129–1139.
Balk, E. M., Boni, P. A., Moskowitz, H., Schmid, C. H., Ioannidis, J. P. A., Wang, C., & Lau, J. (2002). Correlation
of quality measures with estimates of treatment effects in meta-analysis of randomised controlled trials. Journal of
the American Medical Association, 287, 2973–2982.
Barrowclough, C., & Tarrier, N. (1984). Psychosocial interventions with families and their effects on the course of
schizophrenia: a review. Psychological Medicine, 14, 629–642.
Barrowclough, C., & Tarrier, N. (1992). Families of schizophrenic patients: A cognitive-behavioural intervention.
London: Chapman & Hall.
Basoglu, M., Marks, I., Livanou, M., & Swinson, R. (1997). Double-blindness procedure, rater blindness, and ratings
of outcome. Archives of General Psychiatry, 54, 744–748.
Bebbington, P., & Kuipers, L. (1994). The predictive utility of expressed emotion in schizophrenia: an aggregate
analysis. Psychological Medicine, 24, 707–718.
Beck, A. T. (1952). Successful out-patient psychotherapy of a chronic schizophrenic with a delusion based on bor-
rowed guilt. Psychiatry, 15, 305–312.
Begg, C., Cho, M., Eastwood, S., Horton, R., Moher, D., Dolkin, I., Pitkin, R., Rennie, D., Schulz, K. F., Simelk,
D., & Stroup, D. F. (1996). Improving the quality of reporting of randomized controlled trials: the CONSORT
statement. Journal of the American Medical Association, 276, 637–639.
Bellack, A. S. (1986). Schizophrenia: behavior therapy’s forgotten child. Behavior Therapy, 17, 199–214.
Bennett-Levy, J. (2003). Mechanisms of change in cognitive therapy: the case of automatic thought records and
behavioural experiments. Behavioural and Cognitive Psychotherapy, 31, 261–278.
Bentall, R. P., Lewis, S., Tarrier, N., Haddock, G., Drake, R., & Day, J. (2003). Relationships matter: the impact of
the therapeutic alliance on outcome in schizophrenia. Schizophrenia Research, 60(Suppl. 1), 319.
Blackburn, I.-M., James, I. A., Milne, D. L., Baker, C., Standart, S., Garland, A., & Reichelt, K. F. (2001). The
revised cognitive therapy scale (CTS-R) psychometric properties. Behavioural and Cognitive Psychotherapy, 29,
Brown, S. A. (1991). Measurement of quality of primary studies for meta-analysis. Nursing Research, 40, 352–355.
Butzlaff, R. L., & Hooley, J. M. (1998). Expressed emotion and psychiatric relapse: a meta-analysis. Archives of Gen-
eral Psychiatry, 55, 547–552.
Chadwich, P., Williams, C., & Mackenzie, J. (2003). Impact of case formulation in cognitive behaviour therapy for
psychosis. Behaviour Research and Therapy, 41, 671–680.
Chalmers, T. C., Smith, H., Blackburn, B., Silverman, B., Schroeder, B., Reitman, D., & Ambroz, A. (1981). A
method for assessing the quality of a randomised control trial. Controlled Clinical Trials, 2, 31–49.
Chambless, D. L., Tran, G. Q., & Glass, C. R. (1997). Predictors of response to cognitive-behavioral group therapy
for social phobia. Journal of Anxiety Disorders, 11, 221–240.
Cohen, J. (1988). Statistical power analysis for the behavioural sciences. New York: Academic Press (revised edition).
Constantino, M. J., Castonguay, L. G., & Schut, A. J. (in press). The working alliance: a flagship for the scientist-
practitioner model in psychotherapy. In G. Tryon (Ed.), Counseling based on research. New York: Allyn & Bacon.
Copas, J. B., & Shi, J. Q. (2001). A sensitivity analysis for publication bias in systematic reviews. Statistical Methods
in Medical Research, 10, 251–265.
Detsky, A. S., Naylor, C. D., O’Rourke, K., McGeer, A. J., & L’Abbe, K. A. (1992). Incorporating variations in the
trial quality of individual randomised trial into meta-analysis. Journal of Clinical Epidemiology, 45, 255–265.
Devane, S., Haddock, G., Lancashire, S., Baguley, I., Butterworth, A., Tarrier, N., James, A., & Molyneux, P.
(1998). The clinical skills of community psychiatric nurses working with patients who have severe and enduring
mental health problems: an empirical analysis. Journal of Advanced Nursing, 27, 253–260.
Di Blasi, Z. (2003). The crack in the biomedical box. The Psychologist, 16, 72–76.
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011396
Di Blasi, Z., Harkness, E., Georgiou, A., & Kleijnen, J. (2001). Influence of context on health outcomes: a systematic
review. The Lancet, 357, 757–762.
Doll, R. (1998). Controlled trials: the 1948 watershed. British Medical Journal, 317, 1217–1220.
Drury, V., Birchwood, M., Cochrane, R., & MacMillan, F. (1996a). Cognitive therapy and recovery from acute psy-
chosis: a controlled trial. I. Impact on psychotic symptoms. British Journal of Psychiatry, 169, 593–601.
Drury, V., Birchwood, M., Cochrane, R., & MacMillan, F. (1996b). Cognitive therapy and recovery from acute psy-
chosis: a controlled trial. II. Impact on recovery time. British Journal of Psychiatry, 169, 602–607.
Dunn, G., Maracy, M., Dowrick, C., Ayuso-Mateos, J. L., Dalgard, O. S., Page, H., Lehtinen, V., Casey, P.,
Wilkinson, C., Vazquez-Barquero, J. L., Wilkinson, G., & The ODIN Group. (2003). Estimating psychological
treatment effects from a randomised controlled trial with both non-compliance and loss to follow-up. British Jour-
nal of Psychiatry, 183, 323–331.
Durham, R. C., Guthrie, M., Morton, R. V., Reid, D. A., Treliving, L. R., Fowler, D., & MacDonald, R. R. (2003).
Tayside-Fife clinical trial of cognitive-behaviour therapy for medication-resistant psychotic symptoms: results to
three month follow-up. British Journal of Psychiatry, 182, 303–311.
Everitt, B. S., & Pickles, A. (1999). Statistical aspects of the design and analysis of clinical trials. London: Imperial
Falloon, I. R. H., Boyd, J. L., & McGill, C. W. (1984). Family care of schizophrenia. New York: Guilford Press.
Farhill, J., & Voudouris, N. (1996). Persisting auditory hallucinations: prospects for non-medication interventions in
a hospital population. Behaviour Change, 13, 112–123.
Foulds, G. A. (1958). Clinical research in psychiatry. Journal of Mental Science, 104, 259–265.
Gartey, P. A., Fowler, D., Kuipers, E., Freeman, D., Dunn, G., Bebbington, P., Hadley, C., & Jones, S. (1997).
London-East Anglia randomised controlled trial of cognitive-behavioural therapy for psychosis. II. Predictors of
outcome. British Journal of Psychiatry, 171, 420–426.
Garety, P. A., Kuipers, E., Fowler, D., Freeman, D., & Bebbington, P. E. (2001). A cognitive model of the positive
symptoms of psychosis. Psychological Medicine, 31, 189–195.
Gotzsche, P. C. (1989). Methodology and overt and hidden bias in reports of 196 double-blind trials of nonsteroidal
anti-inflammatory drugs in rheumatoid arthritis. Controlled Clinical Trials, 10, 31–56.
Gould, R. A., Mueser, K. T., Bolton, E., Mays, V., & Goff, D. (2001). Cognitive therapy for psychosis in schizo-
phrenia: an effect size analysis. Schizophrenia Research, 48, 335–342.
Gumley, A., O’Grady, M., McNay, L., Reilly, J., Power, K., Karatzias, A., Tait, A., Chouliara, Z., & White, C.
(2003). A randomised trial of targeted cognitive behavioural therapy in schizophrenia: effects on relapse at 12-
months. Psychological Medicine, 33, 419–431.
Haddock, G., Devane, S., Bradshaw, T., McGovern, J., Tarrier, N., Kinderman, P., Baguley, I., Lancashire, S., &
Harris, N. (2001). An investigation into the psychometric properties of the cognitive therapy scale for psychosis
(CTS-Psy). Behavioral and Cognitive Psychotherapy, 29, 93–106.
Haddock, G., Tarrier, N., Morrison, A. P., Hopkins, R., Drake, R., & Lewis, S. (1999). A pilot study evaluating the
effectiveness of individual inpatient cognitive-behavioural therapy in early psychosis. Social Psychiatry and Psychi-
atric Epidemiology, 34, 254–258.
Horvath, A. O. (1994). Research on alliance. In A. O. Horvath, & L. S. Greenberg (Eds.), The working alliance:
Theory, research and practice (pp. 259–286). New York: Wiley.
Hrobjartsson, A., & Gotzsche, P. C. (2001). Is the placebo powerless? An analysis of clinical trials comparing placebo
with no treatment. New England Journal of Medicine, 344, 1594–1602.
Huwiler-Muntener, K., Juni, P., Junker, C., & Eggar, M. (2002). Quality of reporting of randomised trials as a meas-
ure of methodologic quality. Journal of the American Medical Association, 287, 2801–2804.
Jadad, A. R., Moore, R. A., Carroll, D., Jenkinson, C., Reynolds, D. J. M., Gavaghan, D. J., & McQuay, H. J.
(1996). Assessing the quality of reports of randomized clinical trials: is blinding necessary? Controlled Clinical
Trials, 17, 1–12.
Johnson, T. (1998). Clinical trials in psychiatry: background and statistical perspective. Statistical Methods in Medi-
cal Research, 7, 209–234.
Juni, P., Altman, G., & Eggar, M. (2001). Assessing the quality of controlled clinical trials. British Medical Journal,
1397 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
Juni, P., Witschi, A., Bloch, R., & Eggar, M. (1999). The hazards of scoring the quality of clinical trials for meta-
analysis. Journal of the American Medical Association, 282, 1054–1060.
Kane, J. M. (1996). Treatment resistant schizophrenic patients. Journal of Clinical Psychiatry, 57(Suppl. 9), 35–40.
Kane, J. M. (1999). Management strategies for the treatment of schizophrenia. Journal of Clinical Psychiatry,
60(Suppl. 12), 13–17.
Kavanagh, D. J. (1992). Recent developments in expressed emotion and schizophrenia. British Journal of Psychiatry,
Kazdin, A. E., & Bass, P. (1989). Power to detect differences between alternative treatments in comparative psycho-
therapy outcome research. Journal of Consulting and Clinical Psychology, 57, 138–147.
Klein, D. N., Schwartz, J. E., Santiago, N. J., Vivian, D., Vocisano, C., Castonguay, L. G., Blalock, J. A., Markowitz,
J. C., Rothbaum, B., Thase, M. E., Arnow, B., Manber, R., Riso, L. P., McCullough, J. P., & Borian, F.
E. (2003). Therapeutic alliance in depression treatment: controlling for prior change and patient character-
istics. Journal of Consulting and Clinical Psychology, 71, 997–1006.
Kuipers, E. (1998). Working with carers: interventions for relatives and staff carers of those who have psychosis.
In T. Wykes, N. Tarrier, & S. Lewis (Eds.), Outcome and innovation in psychological treatment of schizophrenia
(pp. 201–214). Chichester: Wiley.
Lancashire, S., Haddock, G., Tarrier, N., Baguley, I., Butterworth, C. A., & Brooker, C. (1997). The impact of train-
ing community psychiatric nurses to use psychosocial interventions with people who have serious mental health
problems: The Thorn Nurse Training Project. Psychiatric Services, 48, 39–41.
Leff, J. P., & Vaughn, C. E. (1985). Expressed emotion in families. New York: Guilford Press.
Lewis, S. W., Tarrier, N., Haddock, G., Bentall, R., Kinderman, P., Kingdon, D., Siddle, R., Drake, R., Everitt, J.,
Leadley, K., Benn, A., Glazebrook, K., Haley, C., Akhtar, S., Davies, L., Palmer, S., Faragher, B., & Dunn, G.
(2002). Randomised controlled trial of cognitive-behaviour therapy in early schizophrenia: acute phase outcomes.
British Journal of Psychiatry, 181(Suppl. 43), 91–97.
Liberman, R. P., Teigen, J., Patterson, R., & Baker, V. (1973). Reducing delusional speech in chronic paranoid schiz-
ophrenics. Journal of Applied Behavior Analysis, 6, 57–64.
Malik, M. L., Beutler, L. E., Alimohamed, S., Gallagher-Thompson, D., & Thompson, L. (2003). Are all cognitive
therapies alike? A comparison of cognitive and noncognitive therapy processes and implications for the appli-
cation of empirically supported treatments. Journal of Clinical and Consulting Psychology, 71, 150–158.
Marshall, M., Lockwood, A., Bradley, C., Adams, C., Joy, C., & Fenton, M. (2000). Unpublished rating scales; a
major source of bias in randomised controlled trials of schizophrenia. British Journal of Psychiatry, 176, 249–252.
Meichenbaum, D., & Cameron, R. (1973). Training schizophrenics to talk to themselves: a means of developing
attentional control. Behavior Therapy, 4, 515–534.
Miller, W. R., & Rollnick, S. (2002). Motivational interviewing: preparing people for change (2nd ed.) (pp. 33–42).
New York: Guilford Press (Chapter 4 What is motivational interviewing?).
Moher, D., Jadad, A. R., Nichol, G., Penman, M., Tugwell, P., & Walsh, S. (1995). Assessing the quality of rando-
mised controlled trials: an annotated bibliography of scales and checklists. Controlled Clinical Trials, 16, 62–73.
Moher, D., Pham, B., Jones, A., Cook, D. J., Jadad, A. R., Moher, M., Tugwell, P., & Klassen, T. P. (1998). Does
quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses? Lancet,
Moher, D., Schultz, K., & Altman, D. (2001). The CONSORT statement: revised recommendations for improving
the quality of reports of parallel group randomisation. Journal of the American Medical Association, 285,
Morrison, A. P. (2001). The interpretation of intrusions in psychosis: an integrative cognitive approach to hallucina-
tions and delusions. Behavioural and Cognitive Psychotherapy, 29, 257–276.
Morrison, A. P., French, P., Walford, L., Lewis, S., Kilcommons, A., Green, J., Lomax, S. & Bentall, R. B. (in
press). A randomised controlled trial of cognitive therapy for the prevention of psychosis in people at ultra-high
risk British Journal of Psychiatry (in press).
Morrison, A. P., Haddock, G., & Tarrier, N. (1995). Intrusive thoughts and auditory hallucinations: a cognitive
approach. Behavioural and Cognitive Psychotherapy, 23, 265–280.
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011398
Mueser, K. T., & Berenbaum, H. (1990). Psychodynamic treatment of schizophrenia: is there a future?. Psychological
Medicine, 20, 253–262.
National Institute for Clinical Excellence. (2003). Schizophrenia: core interventions in the treatment and management
of schizophrenia in primary and secondary care. UK: National Institute for Clinical Excellence.
Nuechterlein, K. H. (1987). Vulnerability models of schizophrenia: state of the art. In H. Hafner, W. F. Gattaz, &
Jangarik (Eds.), Searches for the causes of schizophrenia. Berlin: Springer Press.
Pantelis, C., & Barns, T. R. E. (1996). Drug strategies in treatment resistant schizophrenia. Australian and New
Zealand Journal of Psychiatry, 30, 20–37.
Pilling, S., Bebbington, P., Kuipers, E., Garety, P., Geddes, J., Orbach, G., & Morgan, C. (2002). Psychological
treatments in schizophrenia: I. Meta-analysis of family interventions and cognitive behaviour therapy. Psychologi-
cal Medicine, 32, 763–782.
Pocock, S. J. (1996). Clinical trials: A practical approach. Chichester: Wiley.
Rector, N. A., & Beck, A. T. (2001). Cognitive behavioral therapy for schizophrenia: an empirical review. Journal of
Nervous and Mental Disease, 189, 278–287.
Richardson, A., Baker, M., Burns, T., Lilford, R. J., & Muijen, M. (2000). Reflections on methodological issues in
mental health research. Journal of Mental Health, 9, 463–470.
Salkovskis, P. M. (2002). Empirically grounded clinical interventions. Cognitive behaviour therapy progresses
through a multi-dimensional approach to clinical science. Behavioural and Cognitive Psychotherapy, 30, 3–10.
Schulz, K. F. (1995). Subverting randomisation in controlled trials. Journal of the American Medical Association, 274,
Schulz, K. F., Altman, D. G., & Moher, D. (2002). Allocation concealment in clinical trials. Journal of the American
Medical Association, 288, 2406–2407.
Schulz, K. F., Chalmers, I., Hayes, R. J., & Altman, D. G. (1995). Empirical evidence of bias: dimensions of meth-
odological quality associated with estimates of treatment effect in controlled trials. Journal of the American Medi-
cal Association, 280, 178–180.
Sensky, R., Turkington, D., Kingdon, D., Scott, J. L., Scott, J., Siddle, R., O’Carrol, M., & Barnes, T. R. (2000). A
randomised controlled trial of cognitive-behavioural therapy for persistent symptoms in schizophrenia resistant to
medication. Archives of General Psychiatry, 57, 165–172.
Slade, M., & Priebe, S. (2001). Are randomised controlled trials the only gold that glitters? British Journal of Psy-
chiatry, 179, 286–287.
Startup, M., Jackson, M., & Bendix, S. (2004). North Wales randomised controlled trial of cognitive behaviour ther-
apy for acute schizophrenia spectrum disorders: outcome at six and twelve months. Psychological Medicine, 34,
Startup, M., Jackson, M., & Pearce, E. (2002). Assessing therapist adherence to cognitive-behavioural therapy for
psychosis. Behavioural and Cognitive Psychotherapy, 30, 329–339.
Sterne, J. A. C., Juni, P., Schulz, K. F., Altman, D. G., Barlet, C., & Eggar, M. (2002). Statistical methods for
assessing the influence of study characteristics on treatment effects in meta-epidemiological research. Statistics in
Medicine, 21, 1513–1524.
Tarrier, N. (1996). A psychological approach to the management of schizophrenia. In M. Moscarelli, & N. Sartorius
(Eds.), The economics of schizophrenia (pp. 271–286). Chichester: Wiley.
Tarrier, N. (2001). What can be learnt from clinical trials? A reply to Devilly and Foa. Journal of Consulting and
Clinical Psychology, 69, 117–118.
Tarrier, N. (2002). Yes, cognitive behaviour therapy may well be all you need. British Medical Journal, 324, 291–292.
Tarrier, N., Beckett, R., Harwood, S., Baker, A., Yusupoff, L., & Ugarteburu, I. (1993). A controlled trial of two
cognitive behavioural methods of treating drug-resistant residual psychotic symptoms in schizophrenic patients: I.
Outcome. British Journal of Psychiatry, 162, 524–532.
Tarrier, N., & Bobes, J. (2000). The importance of psychosocial interventions and patient involvement in the treat-
ment of schizophrenia. International Journal of Psychiatry in Clinical Practice, 4(Suppl. 1), 35–52.
Tarrier, N., & Calam, R. (2002). New developments in cognitive-behavioural case formulation. Epidemiological, sys-
temic and social context: an integrative approach. Cognitive and Behavioural Psychotherapy, 30, 311–328.
1399 N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401
Tarrier, N., Kinney, C., McCarthy, E., Humphreys, L., Wittkowski, A., & Morris, J. (2000). Two year follow-up of
cognitive-behaviour therapy and supportive counselling in the treatment of persistent positive symptoms in
chronic schizophrenia. Journal of Consulting and Clinical Psychology, 68, 917–922.
Tarrier, N., Kinney, C., McCarthy, E., Wittkowski, A., Yusupoff, L., Gledhill, A., & Morris, J. (2001). Are some
types of psychotic symptoms more responsive to CBT? Behavioural and Cognitive Psychotherapy, 29, 45–55.
Tarrier, N., Lewis, S. W., Haddock, G., Bentall, R., Drake, R., Dunn, G., Kinderman, P., Kingdon, D., Siddle, R.,
Everitt, J., Leadley, K., Benn, A., Glazebrook, K., Haley, C., Akhtar, S., Davies, L., & Palmer, S. (2004). 18
month follow-up of a randomised, controlled clinical trial of cognitive-behaviour therapy in first episode and early
schizophrenia (The SoCRATES Trial). British Journal of Psychiatry, 184, 231–239.
Tarrier, N., & Turpin, G. (1992). Psychosocial factors, arousal and schizophrenic relapse: a review of the psychophy-
siological data. British Journal of Psychiatry, 161, 3–11.
Tarrier, N., Wittkowski, A., Kinney, C., McCarthy, E., Morris, J., & Humphreys, L. (1999). The durability of the
effects of cognitive behaviour therapy in the treatment of chronic schizophrenia: twelve months follow-up. British
Journal of Psychiatry, 174, 500–504.
Tarrier, N., & Wykes, T. (2004). Cognitive-behavioural treatments of psychosis: clinical trials and methodological
issues in clinical psychology. In S. Day, S. Green, & D. Machin (Eds.), Textbook of clinical trials. Chichester:
Tarrier, N., Yusupoff, L., Kinney, C., McCarthy, E., Gledhill, A., Haddock, G., & Morris, J. (1998). A randomised
controlled trial of intensive cognitive behaviour therapy for chronic schizophrenia. British Medical Journal, 317,
Tattan, T., & Tarrier, N. (2000). The expressed emotion of case managers of the seriously mentally ill: the influence
of EE and the quality of the relationship on clinical outcomes. Psychological Medicine, 30, 195–204.
Turkington, D., & Kindon, D. (2000). Cognitive behaviour therapy techniques for general psychiatrists in the man-
agement of patients with psychosis. British Journal of Psychiatry, 177, 101–106.
Turkington, D., Kingdon, D., Turner, T., & The Insight Group. (2002). Effectiveness of a brief cognitive behavioural
therapy intervention in the treatment of schizophrenia. British Journal of Psychiatry, 180, 523–528.
Wykes, T., Tarrier, N., & Everitt B. (submitted). Cognitive behaviour therapy (CBT) for schizophrenia: effect sizes,
clinical models and methodological rigour (submitted for publication).
Wykes, T., Tarrier, N., & Lewis, T. (1998). Introduction. Outcome and innovation in the psychological treatment of
schizophrenia (pp. 1–15). Chichester: Wiley.
Young, J. & Beck, A. T. (1980). The cognitive therapy scale. Philadelphia Centre for Cognitive Therapy. Unpub-
Zubin, J., & Spring, B. (1977). Vulnerability: a new view of schizophrenia. Journal of Abnormal Psychology, 86,
Further reading: reports of clinical trials
Barrowclough, C., Haddock, G., Tarrier, N., Lewis, S., Moring, J., O’Brien, R., Schofield, N., & McGovern, J.
(2001). Randomized controlled trial of motivational interviewing, cognitive behaviour therapy, and family inter-
vention of patients with co-morbid schizophrenia and substance use disorder. American Journal of Psychiatry, 158,
Drake, R. E., Osher, F. C., & Wallach, M. A. (1989). Alcohol use and abuse in schizophrenia: a prospective com-
munity study. Journal of Nervous and Mental Disease, 177, 408–414.
Garety, P. A., Kuipers, E., Fowler, D., Chamberlain, F., & Dunn, G. (1994). Cognitive behavioural therapy for drug
resistant psychosis. British Journal of Psychiatry, 67, 259–271.
Haddock, G., Slade, P. D., Bentall, R. P., Reid, D., & Faragher, E. B. (1998). A comparison of the long-term effec-
tiveness of distraction and focussing in the treatment of auditory hallucinations. British Journal of Medical Psy-
chology, 71, 339–349.
Hall, P., & Tarrier, N. (2003). The cognitive-behavioural treatment of low self-esteem in psychotic patients: a pilot
study. Behaviour Research and Therapy, 41, 317–332.
N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–14011400
Kuipers, E., Garety, P., Fowler, D., Dunn, G., Bebbington, P., Freeman, D., & Hadley, C. (1997). London-East Download full-text
Anglia randomised controlled trial of cognitive-behavioural therapy for psychosis. British Journal of Psychiatry,
Milton, F., Patwa, V. K., & Hafner, R. J. (1978). Confrontation vs belief modification in persistently deluded
patients. British Journal of Medical Psychology, 51, 127–130.
Pinto, A., La Pia, S., Mennella, R., Giorgio, D., & De Simone, L. (1999). Cognitive-behavioural therapy and cloza-
pine for clients with treatment-refractory schizophrenia. Psychiatric Services, 50, 901–904.
Rector, N., Seeman, M. V., & Segal, Z. V. (2003). Cognitive therapy for schizophrenia: a preliminary randomised
trial. Schizophrenia Research, 63, 1–11.
Valmaggia, L. (2002). A randomised controlled trial of cognitive behaviour therapy with treatment refractory posi-
tive symptoms of schizophrenia. Thesis submitted for PhD at University of Manchester.
Valmaggia, L., van der Gaag, M., Tarrier, N., Pijnenborg, G. H. M., & Sloof, C. J. (in press). A randomized con-
trolled trial of cognitive behavior therapy with treatment refractory positive symptoms of schizophrenia British
Journal of Psychiatry (in press).
1401N. Tarrier, T. Wykes / Behaviour Research and Therapy 42 (2004) 1377–1401