Source: Cochrane Database of Systematic Reviews
        Vol 16, #3. Art. No.: CD001027
Date:   July 16, 2008
URL:    http://www.mrw.interscience.wiley.com/cochrane/clsysrev/articles/CD001027/frame.html
        http://www.mrw.interscience.wiley.com/cochrane/clsysrev/articles/CD001027/abstract.html
Rem:    Below, you will find the review section only (pages 1-28); the report
        itself is 102 (or 98) pages.


[Intervention Review]

Cognitive behaviour therapy for chronic fatigue syndrome in adults
------------------------------------------------------------------
Jonathan R Price(1,*), Edward Mitchell(2), Elizabeth Tidy(3), Vivien Hunot(4)
1 Department of Psychiatry, University of Oxford, Oxford, UK.
2 University of Oxford Clinical School, Pembroke College, Oxford, UK.
3 University of Oxford, New College, Oxford, UK.
4 Section of Evidence Based Mental Health, Health Service & Population Research
  Department, Institute of Psychiatry, King's College London, London, UK
* Contact address: Jonathan R Price, Department of Psychiatry, University of
  Oxford, The Warneford Hospital, Headington, Oxford, OX3 7JX, UK. 
  jonathan.price@psych.ox.ac.uk. (Editorial group: Cochrane Depression, Anxiety
  and Neurosis Group.)

This version first published online: 16 July 2008 in Issue 3, 2008.

--------------------------------------------------------------------------------
ABSTRACT

Background
Chronic fatigue syndrome (CFS) is a common, debilitating and serious health
problem. Cognitive behaviour therapy (CBT) may help to alleviate the symptoms
of CFS.

Objectives
To examine the effectiveness and acceptability of CBT for CFS, alone and in
combination with other interventions, compared with usual care and other
interventions.

Search strategy
CCDANCTR-Studies and CCDANCTR-References were searched on 28/3/2008. We
conducted supplementary searches of other bibliographic databases. We
searched reference lists of retrieved articles and contacted trial authors
and experts in the field for information on ongoing/completed trials.

Selection criteria
Randomised controlled trials involving adults with a primary diagnosis of
CFS, assigned to a CBT condition compared with usual care or another
intervention, alone or in combination.

Data collection and analysis
Data on patients, interventions and outcomes were extracted by two review
authors independently, and risk of bias was assessed for each study. The
primary outcome was reduction in fatigue severity, based on a continuous
measure of symptom reduction, using the standardised mean difference (SMD),
or a dichotomous measure of clinical response, using odds ratios (OR), with
95% confidence intervals (CI).

Main results Fifteen studies (1043 CFS participants) were included in the
review. When comparing CBT with usual care (six studies, 373 participants),
the difference in fatigue mean scores at post-treatment was highly
significant in favour of CBT (SMD -0.39, 95% CI -0.60 to -0.19), with 40% of
CBT participants (four studies, 371 participants) showing clinical response
in contrast with 26% in usual care (OR 0.47, 95% CI 0.29 to 0.76). Findings
at follow-up were inconsistent. For CBT versus other psychological therapies,
comprising relaxation, counselling and education/support (four studies, 313
participants), the difference in fatigue mean scores at post-treatment
favoured CBT (SMD -0.43, 95% CI -0.65 to -0.20). Findings at follow-up were
heterogeneous and inconsistent. Only two studies compared CBT against other
interventions and one study compared CBT in combination with other
interventions against usual care.

Authors' conclusions
CBT is effective in reducing the symptoms of fatigue at post-treatment
compared with usual care, and may be more effective in reducing fatigue
symptoms compared with other psychological therapies. The evidence base at
follow-up is limited to a small group of studies with inconsistent findings.
There is a lack of evidence on the comparative effectiveness of CBT alone or
in combination with other treatments, and further studies are required to
inform the development of effective treatment programmes for people with CFS.

--------------------------------------------------------------------------------
PLAIN LANGUAGE SUMMARY

Cognitive behaviour therapy for chronic fatigue syndrome

Chronic fatigue syndrome (CFS) is a very common and disabling condition, in
which people suffer from persistent symptoms of fatigue that are unexplained.
Cognitive behaviour therapy is a psychological therapy model that is commonly
used to treat a range of psychological and chronic pain conditions. This
review aimed to find out whether CBT is effective for CBT, both as a
standalone treatment and in combination with other treatments, and whether it
is more effective than other treatments used for CFS. The review included 15
studies, with a total of 1043 CFS participants. The review showed that people
attending for CBT were more likely to have reduced fatigue symptoms at the
end of treatment than people who received usual care or were on a waiting
list for therapy, with 40% of people in the CBT group showing clinical
improvement, in contrast with 26% in usual care. At follow-up, 1-7 months
after treatment ended, people who had completed their course of CBT continued
to have lower fatigue levels, but when including people who had dropped out
of treatment, there was no difference between CBT and usual care. The review
also compared CBT against other types of psychological therapy, including
relaxation techniques, counselling and support/education, and found that
people attending for CBT was more likely to have reduced fatigue symptoms at
the end of treatment than those attending for other psychological therapies.
Physical functioning, depression, anxiety and psychological distress symptoms
were also more reduced when compared with other psychological therapies.
However at follow-up, the results were inconsistent and the studies did not
fit well together, making it difficult to draw any conclusions. Very few
studies reported on the acceptability of CBT and no studies examined side
effects. Only two studies compared the effectiveness of CBT against other
treatments, both exercise therapy, and just one study compared a combination
of CBT and other treatments with usual care. More studies should be carried
out to establish whether CBT is more helpful than other treatments for CFS,
and whether CBT in combination with other treatments is more helpful than
single treatment approaches.

--------------------------------------------------------------------------------
BACKGROUND

Description of condition

Chronic fatigue syndrome (CFS) is a condition characterised by persistent
fatigue and significant associated disability, but without evident physical
or psychological disorder to explain the problem. Sufferers endure fatigue
which may cause considerable and persisting debility. They may also
experience a lack of understanding from others, including health
professionals. This may be exacerbated by normal or equivocal findings on
physical examination or investigation despite the sufferer's experience of
illness. It is an important health care issue, presenting major problems for
both its sufferers and for health services.

Whilst fatigue is so common as to be almost a normal consequence of life
(Ridsdale 1989), many of the population experience fatigue which is much more
severe or long-lasting. About one in three of the UK population report
substantial fatigue, and one in five report fatigue having been present for
over six months (Cox 1987). Some of these sufferers have physical or
psychological disorders which cause fatigue. Nevertheless, a significant
proportion of people with disabling and chronic fatigue appear to have a
disorder which is medically unexplained. The term CFS describes this more
defined, and serious, problem, which is thought to affect up to 1% of the
general population (Wessely 1995), with a reported range of 0.006% up 3%
depending on the setting and criteria used (Afari 2003).

CFS has had many names in recent decades. These include Royal Free disease,
Iceland disease, neurasthenia, myalgic encephalomyelitis ('ME'), and
post-viral fatigue syndrome. All these is, persistent medically unexplained
fatigue which causes disability and distress. The term CFS has been adopted
and clearly defined in order to foster research on this common problem.
Opinions regarding the cause have tended to focus on either mainly physical
or mainly psychological explanations. There is increasing awareness in a
range of illnesses of the potential interaction of physical and psychological
factors in the development and maintenance of the disorder. Such illnesses
include irritable bowel syndrome, low back pain and CFS.

Many people with CFS in the community attend their general practitioners for
assessment and treatment. Some attend alternative practitioners, disappointed
with the results of orthodox medical practice. Others are referred to
out-patient clinics for assessment. Referral to a variety of specialist
clinics, including general medicine, endocrinology, infectious diseases,
neurology, and psychiatry, reflects uncertainty and, often, disagreement
among doctors and sufferers about the causes of CFS. Sufferers' and
professionals' struggle to understand the illness may lead to unsatisfactory
patient-professional relationships, and dissatisfied patients.


Description of intervention

Until recently, research studies have been unable to offer neither a
convincing explanation of the causes of CFS nor evidence for effective
treatments. Often, sufferers have been advised to limit the demands placed
upon their bodies, in order to promote recovery and prevent deterioration,
which may lead to further impairment of quality of life. Treatment strategies
for CFS include psychological, physical and pharmacological interventions.
Randomised controlled trials (RCTs) have been carried out to assess the
effectiveness of treatments as diverse as exercise therapy (Fulcher 1997,
Wearden 1998), self-help treatment (Chalder 1997), antidepressants (Behan
1995, Vercoulen 1996, Wearden 1998) and dietary supplementation with fatty
acids (Warren 1999) and folic acid (Kaslow 1989). Over the last 15 years, an
increasing number of trials for CFS have focused attention on psychological
therapies, including supportive 'non-directive' interventions and treatments
that draw from cognitive and behavioural approaches.


How the intervention might work

First developed in the 1960s, and later manualised as cognitive therapy (CT)
(Beck 1979), cognitive behavioural therapy (CBT) incorporates elements of
both behavioural therapy (BT) and cognitive therapy approaches. CBT
facilitates the identification of unhelpful, anxiety-provoking thoughts, and
challenges these negative automatic thoughts and dysfunctional underlying
assumptions through collaborative 'hypothesis-testing', using behavioural
tasks of diary-keeping and validity-testing of beliefs between sessions,
together with skills training within sessions.

For the treatment of CFS, CBT combines a rehabilitative approach of a graded
increase in activity with a psychological approach addressing thoughts and
beliefs about CFS that may impair recovery. A gradual increase in activity,
negotiated and agreed with the patient, can be used as a 'behavioural
experiment', in which core beliefs, such as about the link between increased
activity and worsened physical symptoms, can be tested. Catastrophic
interpretations of transient increases in physical symptoms can be challenged
by orthodox cognitive therapy techniques, such as the 'double column
technique', in which the original, anxious, unhelpful thought is written
alongside new, more realistic, more helpful alternatives.

Recent developments in CBT include 'third wave' approaches, including
mindfulness-based therapy (Kabat-Zinn 1991; Segal 2002), compassion-focused
therapy (Gilbert 2005) and acceptance and commitment therapy (Hayes 2003).
Originally focusing on the treatment of depression (Segal 2002), and now
being used with a number of other disorders, including CFS, mindfulness
training incorporates meditation techniques and teaches participants to
observe thoughts, emotions and sensations without trying to escape, avoid or
change them (Kabat-Zinn 1991).


Why it is important to do this review

The current body of evidence for CBT remains limited to narrative synthesis
within generic CFS reviews (NICE 2007; Chambers 2006) or to meta-analysis of
mean effect sizes (Malouff 2008). Furthermore, potential heterogeneity has
been largely based on qualitative assessment and the impact of symptom
severity and healthcare setting are uncertain moderators of effect (NICE
2007). An in-depth, up-to-date, systematic review of CBT alone and in
combination with other treatments for CFS is of key importance to inform
treatment decision by patients, clinicians and policymakers. This review is
central in a programme of Cochrane reviews for CFS, which also cover exercise
therapy (Edmonds 2004), pharmacological treatments (Rawson 2007) and
complementary approaches, including acupuncture (Zhang 2006) and traditional
Chinese herbal medicine (Adams 2007).

--------------------------------------------------------------------------------
OBJECTIVES

1. To examine the effectiveness and acceptability of CBT for CFS compared
   with usual management or waiting list control
2. To examine the comparative effectiveness and acceptability of CBT for CFS
   against other interventions (immunological, pharmacological, other
   psychological therapies, exercise alone, complementary/alternative and
   supplements)

For the purposes of the updated version of this review, a third objective has
been added:

3. To examine the effectiveness and acceptability of CBT in combination with
   other interventions against CBT alone or against other interventions alone

--------------------------------------------------------------------------------
METHODS

Criteria for considering studies for this review

Types of studies

All randomised controlled trials (RCTs), both published and unpublished, in
any language, were included in the review. Cluster RCTs, in which centres or
therapists were randomised to intervention or control groups were included,
provided that the specific aim of the study was to examine the effect of the
intervention.

Types of participants
Patient characteristics and setting Trials of male and female participants
over the age of 16 were included, irrespective of culture and setting

Diagnosis
Trials of patients who fulfilled the following diagnostic criteria for CFS
were included:
1) fatigue was the principal symptom
2) the fatigue was medically unexplained, i.e. did not appear to be
   adequately explained by abnormalities found on examination or investigation,
   by diagnosed physical disorder, or by psychiatric disorder including eating
   disorders, alcohol or drug misuse, psychosis, bipolar affective disorder 
   or severe depressive illness
3) the fatigue was of sufficient severity to significantly disable or
   distress the patient
4) the fatigue was over six months in duration* Trials involving several
   disorders were included, provided that over 90% of the patients had a 
   primary diagnosis of CFS according to the above criteria. Trials in which 
   less than 90% of participants had a primary diagnosis of CFS were included 
   in the review if data limited to CFS participants were provided
 * Although there are a number of differences between diagnostic criteria 
   used for CFS (Sharpe 1991; Lloyd 1988; Fukuda 1994;  Carruthers 2003), 
   they are all consistent in specifying a duration of at least six months 
   since onset.

Comorbidity
Studies involving participants with comorbid physical or common mental
disorders were eligible for inclusion, as long as the comorbidity was
secondary to the diagnosis of CFS. However, studies involving patients with a
comorbid psychiatric diagnosis of substance-related disorder, schizophrenia
or psychotic disorder were excluded.


Types of interventions

Intervention
Trials of experimental interventions which met the following criteria were
included:
1) The psychological treatment(s) of interest incorporated attempted
   modification of thoughts and beliefs about symptoms and illness and 
   attempted modification of behavioural responses to symptoms and illness, 
   such as rest, sleep and activity.
2) The intervention was clearly described and was supported by appropriate
   references
3) No restrictions were imposed on the length of treatment, in terms of
   duration or number of sessions, or time between sessions
4) Both individual and group treatment modalities were eligible for inclusion

In the original version of the review, the way in which modification of
thoughts, beliefs, rest, and activity was attempted was used to delineate two
'types' of CBT. 'Type A' attempted to increase activity and reduce rest time
in a systematic manner, independent of symptoms, towards 'normal' levels.
'Type B' attempted to tailor the patient's rest and activity towards levels
which were compatible with the limitations imposed by the disorder.
Therefore, Type B CBT did not explicitly attempt to increase the patient's
physical or psychological capacity beyond improving their ability to 'cope'
with their disabilities.

For the purposes of updating the review, following discussion between the
review authors, it was agreed that distinguishing between the two types of
CBT described above was not feasible. This approach to categorising studies
was therefore not adopted. Instead, interventions were categorised according
to their theoretical underpinning, as follows:
1) Traditional CBT models, including cognitive behavioural approaches drawing
   from Beckian principles (Beck 1979).
2) Recently developed CBT models, including mindfulness training (Kabat-Zinn
   1991), compassion-focused therapy (Gilbert 2005) and acceptance and
   commitment therapy (Hayes 2003).

Control conditions
The following control interventions were included:
1) Minimal or usual medical management (also sometimes described in trials as
   standard care, treatment as usual or waiting list), to be defined here as
   being elements of clinic attendance, investigation, reassurance, and 
   simple advice, without explicit psychological therapy, and also including
   naturalistic prescribing of medication and/or other interventions
2) Immunological therapy, to include immunoglobulin, staphylococcus toxoid
   and interferon
3) Pharmacotherapy, to include hydrocortisone preparations and antidepressants
   (all classes)
4) Non-CBT psychological therapies, to include non-directive (supportive)
   therapies, behavioural relaxation training and psychodynamic therapy
5) Exercise as a stand-alone intervention
6) Complementary/alternative therapies, to include homeopathy, acupuncture,
   massage
7) Supplements, to include general supplements, essential fatty acids and
   other mineral and vitamin based preparations Combination therapy

For the purposes of the updated review, CBT in combination with other
interventions was investigated, with therapies combined with CBT categorised
as immunological, pharmacological, complementary/alternative and supplements.
Exercise in the form of graded exercise or paced activity, was assumed to be
an integral component of CBT.

Main comparisons
For the purposes of the updated review, the following main comparisons were
planned:
1. CBT versus usual care (to include treatment as usual and waiting list)
2. CBT versus immunological therapies
3. CBT versus pharmacological therapies
4. CBT versus other psychological therapies
5. CBT versus exercise
6. CBT versus complementary/alternative therapies
7. CBT versus nutritional supplements
8. CBT + other interventions (see control condition categories)
   versus usual care
9. CBT + other interventions versus CBT alone
10.CBT + other interventions versus other interventions alone


Types of outcome measures

Primary outcomes
The principal outcome used in this review was fatigue, measured as follows:
1) Reduction in fatigue severity (continuous), measured using the Chalder
   Fatigue Scale (Chalder 1993), a 14 item self-rated scale measuring 
   physical and mental fatigue, sometimes shortened to 11 items, or any other 
   validated fatigue scale.
2) Clinical recovery or improvement (dichotomous), based on diagnostic
   interview or defined cut-off on validated scales, as specified by trial
   authors.

Secondary outcomes
1) Improvement in overall physical functioning, measured using rating scales,
   either patient-rated such as the Physical Function dimension of the Short
   Form-36 (Ware 1993), or clinician-rated, such as the Karnofsky index
   (Karnofsky 1948). Other possible measures included timed walking tests and
   tests of strength or aerobic capacity
2) Depression symptoms, using measures such as Hamilton Rating Scale for
   Depression (HRSD) (Hamilton 1960), Beck Depression Inventory (BDI) (Beck
   1987) or Hospital Anxiety and Depression Scale (HADS) (Zigmond 1983)
3) Anxiety symptoms, using measures such as State Trait Anxiety Inventory
   (STAI) (Spielberger 1983) or Beck Anxiety Inventory (BAI) (Beck 1988)
4) General psychological distress, using measures such as Symptom
   Checklist-90 (SCL-90) or General Health Questionnaire
5) Quality of life, using instruments such as Short Form 36 (SF-36) (Ware
   1993)
6) Acceptability of treatment, measured through:
     a) self-rated acceptability/satisfaction scales
     b) dropout rates from trials due to adverse effects or ineffectiveness
7) Adverse effects: proportion of participants experiencing one or more
   adverse effects of treatment (adverse effects being defined as those that 
   are behavioural in basis e.g. self-harm, substance misuse, suicidal
   attempts/suicide, violence to others)
8) Overall dropout rates from trials
9) Occupational outcomes, including employment status and work absence
10)Health service resource use e.g. primary care consultation rate, secondary
   care referral rate, use of alternative practitioners Outcomes were 
   classified as post treatment, short term follow-up (I -6 months 
   post-treatment), medium term follow-up (7-12 months post-treatment) and 
   long term (longer than 12 months).


Search methods for identification of studies

A number of sources were used to identify studies for possible inclusion in
this review, including:
1) Electronic searches
     a) The Depression, Anxiety and Neurosis Cochrane Review Group Trials 
        Registers were searched during December 2006, with an updating search 
        carried out in March 2008.
        CCDANCTR-Studies was searched on 28/3/2008 using the terms
           Diagnosis = chronic fatigue syndrome
        and
           Intervention = *therap*'.
        CCDANCTR-References was searched on 28/3/2008 using the terms
           Full-text = fatigue or CFS
     b) Cochrane Central Register of Controlled Trials (CENTRAL;
        Issue 4, 2006) was searched using the search terms (chronic and fa-
        tigue and syndrome or CFS or myalgic encephalomyelitis) within
        title, abstract or keywords.
     c) Index to Theses of Great Britain and Ireland (from 1716 to 14 December
        2006) was searched at www.theses.com using advanced search (stemming 
        on, fuzziness level 4) as follows: (chronic AND fatigue AND syndrome) 
        OR (myalgic AND encephalomyelitis).
     d) We searched during January 2007 for ongoing studies at
        http://www.clinicaltrials.gov (free text search for 'chronic fatigue'
        or 'myalgic encephalomyelitis'), and http://www.controlled-trials.com
        (all registers in the meta-register of controlled trials, (chronic
        and fatigue and syndrome) or cfs, or myalgic and encephalomyelitis.
2) Citation checking
   Citation lists of potentially relevant papers obtained by these methods
   were searched for further relevant trials, as were the reference lists of 
   those further relevant trials. In addition, the citation lists of relevant 
   and recent systematic reviews were perused (NICE 2007; Chambers 2006; Cho 
   2005; Clark 2005; Rimes 2005; Carruthers 2003; CFS/ME WG 2002; RACPWG 
   2002; Mulrow 2001).
3) Personal contact
   Known specialists in the field and principal authors of those studies,
   identified by the above methods, were contacted and asked to provide 
   details of additional trials, published or unpublished, of which they were 
   aware.


Data collection and analysis

Selection of studies
Two review authors (JP and EM or VH) independently decided whether each
potential trial fulfilled inclusion criteria. Review authors were not blinded
to the names of the authors, institutions, journal of publication, and
results, when they applied the inclusion criteria. Any disagreement on the
eligibility of a study was discussed with the third review author, the
decisions documented and, where necessary, the authors of the studies were
contacted for further information.

Assessment of risk of bias in included studies
Methodological quality in the context of a systematic review of treatment
trials means the extent to which design is likely to have reduced the impact
of bias on the results. For the purposes of the updated version of the
review, methodological quality was assessed using the Cochrane Collaboration
tool for assessing risk of bias. The following domains of risk of bias were
used:
1) Sequence generation: Was the allocation sequence adequately generated?
2) Allocation concealment: Was allocation adequately concealed?
3) Blinding of participants, personnel and outcome assessors for each main
   outcome or class of outcomes: Was knowledge of the allocated intervention
   adequately prevented during the study?
4) Incomplete outcome data for each main outcome or class of outcomes: Were
   incomplete outcome data adequately addressed?
5) Selective outcome reporting: Are reports of the study free of suggestion
   of selective outcome reporting?
6) Other sources of bias: Was the study apparently free of other problems that
   could put it at a high risk of bias?

Trials were allocated to three categories within and across studies as follows:
1. Low risk of bias
2. Unclear risk of bias
3. High risk of bias.

During this procedure, review authors (EM and VH) worked independently of
each other. The review authors were not blinded to the names of the authors,
institutions, journal of publication and results of the trials. The review
authors compared their results, and any disagreements were resolved by
discussion. In the event of disagreement, a third review author (JP) was
consulted, the decisions documented, and where necessary, the authors of the
studies were contacted for further information.

Data extraction
Full data extraction, using a standardised data extraction sheet, was
performed by two review authors independently (EM and ET) on studies selected
for inclusion in the review. Information was collected regarding the
characteristics of participants, characteristics of interventions (including
'type' of CBT), characteristics of outcome measures, and results. Data on the
number with each outcome event was sought by allocated treatment group,
irrespective of compliance and whether or not the patient was deemed
ineligible or otherwise excluded from treatment or follow-up, in order to
allow an intention-to-treat analysis. If any of this information was not
available in the published trial, it was sought by correspondence with the
trial authors.

Measures of treatment effect
Continuous outcomes: where studies used the same outcome measure for a
comparison, data were pooled by calculating the weighted mean difference
(WMD). Where different measures were used to assess the same outcome for a
comparison, data were pooled by calculating the standardised mean difference
(SMD), using 95% confidence intervals.

Dichotomous outcomes: dichotomous outcomes were analysed by calculating an
odds ratio (OR) for each comparison, with the uncertainty in each result
expressed using 95% confidence intervals (CIs). For studies where the number
of participants showing clinical response were not presented in the
original articles, but means and standard deviations were reported for
continuous symptomatology scales, the number of responders was calculated and
imputed from continuous data using a validated statistical method (Furukawa
2005). When overall results were significant, the number needed to treat
(NNT) to produce one outcome was calculated by combining the overall relative
risk with an estimate of the prevalence of the event in the control group of
the trials.

Unit of analysis issues
Where studies had two or more active treatment arms to be compared against
TAU, data were managed as follows: Continuous data - means, SDs and number of
participants for each active treatment group were pooled across treatment
arms as a function of the number of participants in each arm (Law 2003) to be
compared against the control group. As an alternative strategy, the active
comparison considered to be of greatest relevance was selected (e.g. CBT was
selected in preference to CT or BT arms). Dichotomous data - active treatment
groups were collapsed into a single arm for comparison against the control
group, or the control group was split equally into two.

Dealing with missing data
Missing dichotomous data were managed through intention to treat (ITT)
analysis, in which it was assumed that patients who dropped out after
randomisation had a negative outcome. Best/worse case scenarios were also
calculated for the clinical response outcome, in which it was assumed that
dropouts in the active treatment group had positive outcomes and those in the
control group had negative outcomes (best case scenario), and that dropouts
in the active treatment group had negative outcomes and those in the control
group had positive outcomes (worst case scenario), thus providing boundaries
for the observed treatment effect.

Missing continuous data were either analysed on an endpoint basis, including
only participants with a final assessment, or analysed using last observation
carried forward to the final assessment (LOCF) if LOCF data were reported by
the trial authors. Where SDs were missing, attempts were made to obtain these
data through contacting trial authors. Where SDs were not available from
trial authors, they were calculated from t-values, confidence intervals or
standard errors, where reported in articles (Deeks 1997). If these additional
figures were not available or obtainable, the study data were not included in
the comparison of interest.

Data synthesis
A fixed effect model was used in the first instance to combine data. Where
there was evidence of statistical heterogeneity, results were recalculated
using a random effects model, in order to obtain a more conservative
estimate.

Assessment of reporting biases
Where sufficient numbers of trials allowed a meaningful presentation, funnel
plots were constructed to establish the potential influence of publication
bias.

Assessment of heterogeneity
Statistical heterogeneity was formally tested using the natural approximate
chi-square test, which provides evidence of variation in effect estimates
beyond that of chance. Since the chi-squared test has low power to assess
heterogeneity where a small number of participants or trials are included,
the p-value was conservatively set at 0.1. Heterogeneity was also tested
using the I^2 statistic, which calculates the percentage of variability due to
heterogeneity rather than chance, with I^2 values over 50% indicating strong
heterogeneity (Higgins 2003).

Subgroup analyses and investigation of heterogeneity
Clinical characteristics were examined in subgroup analyses to compare their
influence on the size of the treatment effect. Where sufficient studies were
available for inclusion, subgroup analyses were performed for:
1) Type of control condition (minimal management/standard care vs waiting
   list)
2) Modality of treatment (individual therapy vs group therapy)
3) Therapy process (face-to-face contact versus self-help)
4) Therapist experience/qualifications (delivered by trained therapists vs
   trainee therapists)
5) Increased activity introduced as component of intervention versus tailoring
   activity/rest to fit current limitations or not specified
6) Number of CBT sessions (up to and including 8 sessions vs more than 8
   sessions)
7) Severity/chronicity of CFS symptomatology at baseline
8) Setting (clinic-based versus home-based)
These subgroup analyses were also used to examine potential sources of
clinical heterogeneity.

Sensitivity analyses
Where sufficient studies were available for inclusion, sensitivity analyses
were planned to test the robustness of the findings obtained by removing
studies based on the following internal validity criteria:
1) Inadequate allocation concealment
2) Use of less stringent diagnostic inclusion criteria, such as those set out 
   by Lloyd 1988.
3) Dropout rate higher than 20%
4) Lack of formal testing of fidelity to psychological therapy manual
These sensitivity analyses were also used to examine potential sources of
methodological heterogeneity.

--------------------------------------------------------------------------------
RESULTS

Description of studies

See: Characteristics of included studies; Characteristics of excluded
studies; Characteristics of studies awaiting assessment; Characteristics of
ongoing studies.

Results of the search
An updating search of CCDANCTR-Studies and CCDANCTR-References was performed
in March 2008. Of 14 studies identified through the search of CCDANCTR-
Studies, seven were selected for inclusion in the review. Three of the
studies had been included in the original review (Deale 1996; Lloyd 1993;
Sharpe 1993) and a further four studies were included in the updated review
(King 1999; O'Dowd 2000; Prins 2001; Jason 2007). The search of
CCDANCTR-References identified an additional two studies for inclusion
(Huibers 2004; Strang 2002).

A further 20 full articles on potentially eligible studies, including
two unpublished papers and a PhD thesis, were obtained through
supplementary searches and personal communication with experts
in the field. Of those studies, six were selected for inclusion in the
review (Barrett 1992; Ridsdale 2004; Russell 2001; Strang 2002;
Whitehead 2002; Stevens 1999). Two studies (Chalder 1997;
Powell 2001) are currently awaiting assessment. Futher details are
provided below.

Included studies
In total, the combined searches identified 15 completed studies that were
eligible for inclusion in the review. Descriptive information for each study,
including an appraisal of risk of bias for methodological domains, is
presented in the Characteristics of Included Studies table.

Design
All the studies were of a randomised trial design, with randomisation at
the patient level for 14 trials. One trial (Whitehead 2002) randomised
general practices to intervention or control groups. The overall duration of
trials including follow-up ranged from 6 weeks (Strang 2002) to 14 months
(Prins 2001), with a mean duration of 7.5 months. Four studies (Barrett 1992,
Russell 2001, Strang 2002, Surawy 2005) measured final outcomes immediately
after the termination of intervention (no follow-up). The longest period of
follow-up after intervention was 8 months (Huibers 2004, O'Dowd 2000). The
overall mean duration of follow-up after intervention of 3.8 months for the
11 studies for which it could be calculated.

Seven studies reported obtaining ethical approval, and nine studies stated
that patient consent had been obtained. Two studies made no mention of
ethical approval or patient consent (King 1999; Russell 2001).

Sample sizes
The mean sample size of studies was 91 subjects, ranging from 18 subjects
in a randomised trial component of a larger series of exploratory studies
(Surawy 2005), to 278 subjects in a trial conducted at three centres (Prins
2001). Nine studies reported using power calculation to estimate the required
sample size prior to subject recruitment.

Comparison groups
Studies used between two and four arms to conduct comparisons between groups.
Not all arms of all studies were used in this review. Lloyd 1993 used a four
arm trial combining psychological therapy with active or placebo immunologic
therapy; only the two placebo arms were used in this review. Ridsdale 2004
used three arms comprising CBT, graded exercise therapy (GET), and a 'booklet
and usual care' (BUC) group. This last group was described as a prospective
cohort (not being randomised at the same time as the other groups), and was
not used for comparison in this review.

Setting
Eight studies were conducted in the UK, three in the USA, two in The
Netherlands, and two in Australia.

Four studies were conducted in a primary care setting (general practice
surgeries) (Huibers 2004; Ridsdale 2004; King 1999; Whitehead 2002), and
eight in secondary care (hospital outpatients or psychology department
clinics). The setting of the remaining three studies was unclear. Two studies
by doctoral researchers (Stevens 1999, Strang 2002) recruited subjects from
sources such as physicians, psychiatrists, advertisements and local CFS
support groups, but did not specify the setting in which intervention
occurred. The remaining study (Surawy 2005) did not specify sources of
subject recruitment or the setting of intervention.

Participants
The combined sample size across the 15 included studies was 1374
participants, of which 1043 participants met the inclusion criteria for this
review. In three studies, participants with a diagnosis of CFS rather than
chronic fatigue symptoms were selected for inclusion (Huibers 2004; King
1999; Ridsdale 2004). One study included additional treatment arms that were
not eligible for the purposes of this review (Lloyd 1993).

Ten studies gave reasons why not all patients eligible for inclusion were
recruited to the study, of which seven studies reported the total number of
patients eligible for inclusion. Only five studies reported the total number
of patients assessed for eligibility. Eleven studies provided extensive
demographic information on their samples such as marital and occupational
status, as well as age and gender. The remaining four studies (Lloyd 1993,
Russell 2001, Whitehead 2002; Surawy 2005) provided only minimal demographic
information comprising gender, mean age, or age range.

Age of participants
Of the ten studies which stated age criteria for subject recruitment (or the
age range of their eventual sample), four studies included participants under
the age of 18 years, as well as adult participants. The lower age range of
these studies was 17 years for one study (Lloyd 1993), 16 years for two
studies (King 1999, Ridsdale 2004), and 15 years for one study (Whitehead
2002). The oldest upper age range was 75 years (King 1999, Ridsdale 2004).
Six studies provided the mean age of their sample, and a further eight
studies provided the mean ages of comparison groups from which the weighted
overall mean age could be calculated. The overall mean age of subjects was
40.9 years in these 14 studies. One study (Surawy 2005) did not provide a
mean age of the sample or comparison groups.

Gender composition of participants
All studies reported the gender composition of their samples. 70.1% of
subjects were female.

Duration of illness
Three studies provided the mean duration of illness of their sample, and a
further nine studies provided the mean illness duration of comparison groups
from which the weighted overall mean illness duration could be calculated.
The mean duration of illness in these subjects was 56.5 months (4.7 years).

Inclusion criteria
All studies specified inclusion criteria requiring that participants had
fatigue as their main or major complaint, the minimum duration of fatigue
being six months in 12 studies, at least four months in one study (Huibers
2004), and at least three months in the two remaining studies (King 1999,
Ridsdale 2004). There was heterogeneity of other recruitment criteria between
studies. Seven studies used the 'CDC' ('Fukuda') criteria (Fukuda 1994) for
inclusion. One of these studies (Prins 2001), waived the requirement of four
of eight additional symptoms included in the CDC criteria to be present.
Three studies (Sharpe 1996, Deale 1996, Surawy 2005) used the 'Oxford'
criteria (Sharpe 1991) for participant inclusion. The sample in Deale 1996
fulfilled CDC as well as Oxford criteria. Two studies (Barrett 1992, Lloyd
1993) used the 'Australian' criteria (Lloyd 1988).

Three studies did not use standard CFS criteria. Two of these studies (King
1999, Ridsdale 2004) used a score of at least 4 on the Chalder fatigue scale
(Chalder 1993) as the basis for inclusion. The remaining study (Huibers 2004)
included subjects who scored 35 or more on the Dutch Checklist Individual
Strength; Vercoulen 1999) and who had been completely absent from work for
6-26 weeks on health grounds. These studies thus had samples suffering from
chronic fatigue, of which a sub-sample met CDC chronic fatigue syndrome
criteria (28% in King 1999, 21% in Ridsdale 2004, and 44% in Huibers 2004).
Only these subgroups were used in the current review.

Exclusion criteria
Twelve studies specified exclusionary criteria, and two (Strang 2002,
Whitehead 2002) did not. The remaining study (Russell 2001) did not stipulate
any specific exclusion criteria, but did describe an assessment process for
"readiness to engage in the rehabilitation process, group suitability, and
the presence of other complicating psychological difficulties". It is not,
however, clear whether this process resulted in exclusion of subjects. Eight
studies excluded subjects with current mental disorder such as depression or
bipolar disorder (n=3), psychosis (n=6), somatisation disorder (n=1),
substance use (n=2), or who "were at significant risk of suicide or required
urgent psychiatric treatment" (n=1). Two studies excluded those on
psychiatric medication e.g. antidepressants, unless the dose was stable. Six
studies excluded subjects whose symptoms were attributable to a comorbid
condition or a diagnosis other than CFS. Other conditions excluding
participation included pregnancy (n=2), learning difficulties precluding
completion of questionnaires (n=1), organic brain syndromes (n=1), severe
obesity (n=1), and asthma or ischaemic heart disease contraindicating a
step-test (n=1).

Other exclusionary criteria were: likely inability to attend all sessions
e.g. non ambulatory subjects (n=9), insufficient English language to allow
participation (n=4), ongoing physical investigations (n=2), fertility
treatment (n=1), previous immunologic therapy (n=1), involvement in other
previous or current CFS research (n=1).

Assessment of participants for recruitment
In eleven studies, inclusion or exclusion criteria were applied by
interviewing the subject to obtain a problem history. Some of these studies
also made use of further measures such as structured or semi-structured
psychiatric interviews e.g. SCID (n=5), questionnaires or rating scales
(n=7), physical examination of the patient (n=4), and laboratory
investigations such as full blood count, biochemistry, autoantibody
screening, thyroid function tests, and screening for infectious disease
(n=6).

Two studies (Stevens 1999, Strang 2002) used a validated questionnaire as the
sole method of determining recruitment eligibility, and two studies
(Whitehead 2002, Ridsdale 2004) recruited subjects by referral from general
practitioners who had themselves made a diagnosis of CFS using respective
study criteria.


Interventions

Comparison groups
Ten studies compared a CBT model of therapy against a wait list (WL) (n=5) or
'treatment as usual' (TUC) control condition (n=6). Three of these studies
(Barrett 1992, Prins 2001, O'Dowd 2000) had additional comparison arms
consisting of guided support/education to control for non-specific effects of
therapy. A further study (Lloyd 1993) similarly compared CBT versus a TUC
group, but both groups also attended clinic biweekly to receive intramuscular
injections of saline as placebo for dialyzable lymphocyte extract (DLE). One
study (Jason 2007) compared CBT with three comparison groups comprising
cognitive therapy, anaerobic activity therapy, and relaxation therapy. The
remaining studies compared CBT with comparison groups comprising relaxation
therapy (Deale 1996), psychodynamic counselling (King 1999) and graded
exercise therapy (Ridsdale 2004).


Manualisation

Seven studies reported using manuals or treatment protocols for the CBT
intervention (Deale 1996; Huibers 2004; Jason 2007; King 1999; Ridsdale 2004;
Strang 2002; Prins 2001). Three studies reported testing of fidelity to
manuals through audio or video-taping a random selection of CBT sessions
(Jason 2007; King 1999; Prins 2001) and two studies measured fidelity to
manuals through research team meetings (Deale 1996) or through super-
vision/checking written records (Huibers 2004).

Model of CBT
Twelve studies used a 'traditional' model of CBT. Three studies deviated from
traditional models of CBT (in terms of underlying theory as opposed to
delivery). One of these studies (Stevens 1999) used a chronobiologically-
oriented treatment protocol (Nixon 1995), focusing on "sleep-hygiene
education, biofeedback-assisted relaxation training, breathing re-training,
and cognitive therapy" (Stevens 1999). Another study (Strang 2002) used an
element of personal mastery training (Reich 1981) alongside traditional CBT,
and one study (Surawy 2005) used mindfulness-based stress reduction (MBSR;
Kabat-Zinn 1991) and mindfulness-based cognitive therapy (Segal 2002) as the
basis of its treatment model. Ten studies (Barrett 1992, Lloyd 1993, Sharpe
1993, Deale 1996, Stevens 1999, Prins 2001, Strang 2002, Whitehead 2002,
O'Dowd 2000, Jason 2007) attempted to increase activity and reduce rest time
as part of the CBT intervention. Other studies either attempted to tailor the
patients' rest and activity towards levels compatible with limitations
imposed by fatigue, or made no mention of activity/exercise in their
intervention.

Delivery of intervention
Ten studies delivered therapy on a one-to-one therapist to participant basis,
including one study (Lloyd 1993) that invited subjects to bring a close
family member with them, and a further study (Whitehead 2002) that used a
booklet and diary keeping in addition to one-to-one GP contact. Five studies
used a group therapy format, with group sizes of 8 or fewer subjects (Russell
2001), 8-10 subjects (Strang 2002), 8-12 subjects (O'Dowd 2000) or of
unspecified size (Barrett 1992, Surawy 2005).

Therapist characteristics
In six studies, therapists conducting CBT were clinical psychologists,
psychiatrists, and/or trained cognitive (-behavioural) therapists. One study
(Barrett 1992) used a doctoral level psychologist as therapist, after an
initial session with a psychiatrist. Two other studies (Stevens 1999, Strang
2002) also used doctoral level psychologists or their trained assistants to
delivery therapy. Two studies (Whitehead 2002, Huibers 2004) used GPs to
deliver therapy. One study (Jason 2007) used trained nurses with experience
of delivering psychotherapy. Three studies did not specify the nature and
training of therapists.

Supervision and protocol fidelity checks
Six studies (Deale 1996, Prins 2001, King 1999, Huibers 2004, Ridsdale 2004,
Jason 2007) specified that the therapists were supervised, and a further
study (Whitehead 2002) offered telephone support to GP therapists, but this
did not amount to formal supervision. Four studies (Prins 2001, King 1999,
Huibers 2004, Jason 2007) assessed treatment fidelity (e.g. through blind
assessment of video-taped sessions by experienced therapists), and a further
study (Deale 1996) stated that regular meetings were held to ensure
sessions adhered to the treatment protocol.

Total, frequency and length of intervention sessions
The frequency of sessions ranged from weekly in five studies (Barrett 1992,
Sharpe 1993, Deale 1996, Stevens 1999, Surawy 2005), once every two weeks
(Lloyd 1993, Prins 2001, Ridsdale 2004, O'Dowd 2000, Jason 2007), or with a
variable gap between sessions not exceeding three weeks (Deale 1996, Strang
2002, Whitehead 2002, Huibers 2004). One study (King 1999) did not specify
the frequency of sessions.

The minimum number of sessions specified was five to seven (Huibers 2004),
and the maximum was sixteen (Sharpe 1993, Prins 2001), though in one study
(Whitehead 2002) participants were asked to visit their GPs for therapy
weekly or once every two weeks over 12 months.

Session length ranged from a minimum of 30 minutes (Huibers 2004) to a
maximum of 150 minutes in a group format including a 30 minute break (Russell
2001). The modal length of session was 60 minutes. One study did not specify
the length of treatment sessions (Surawy 2005), and in a further study
(Whitehead 2002) session length was set by the treating GPs.

For the eleven studies in which total protocol treatment time could be
determined, the minimum time was 4½ hours (Ridsdale 2004), and the maximum
time 30 hours (including breaks) (Russell 2001). The mean total treatment
specified in these studies' protocols was approximately 13 hours. In the
five studies (Prins 2001, King 1999, Huibers 2004, Ridsdale 2004, Jason 2007)
reporting actual treatment adherence across conditions, the mean time spent
in CBT or control treatment was approximately eight hours.

Non-intervention elements of care programmes
Of the five studies (Deale 1996, Stevens 1999, Prins 2001, Huibers 2004,
O'Dowd 2000) reporting whether non-intervention aspects of treatment
protocols were similar, only one (Prins 2001) noted differences (forbidding
participants in the CBT condition to have physical investigations related to
chronic fatigue syndrome during the intervention, but allowing participants
in the 'guided support' and 'natural course' conditions to do so). Five
studies explicitly stated that psychiatric medication e.g. antidepressants
and anxiolytic drugs, was not forbidden by protocol. A further study (Prins
2001) noted that other treatments for CFS were forbidden by trial protocol
(but made no mention of psychiatric medication). One study (Sharpe 1993)
noted that two participants (7%) in its TUC condition received supportive
psychotherapy, and that one participant (3%) in the CBT condition received
counselling as part of vocational training. A further study (Huibers 2004)
noted that 18% of CBT and 31% of TUC participants received "psychosocial
co-interventions" during the intervention period.

Outcomes

A wide range of outcomes were used in the included studies. These include
measures of this review's primary outcome (fatigue), as well as secondary
outcomes such as physical functioning, depression, anxiety, acceptability of
treatment, quality of life, occupational outcome, treatment compliance,
dropout rate, health service resource use, illness attribution, and adverse
effects. A range of rating methods were used including self-report
questionnaires (e.g. Chalder fatigue scale; Chalder 1993), observer-rated
outcomes (e.g. Karnofsky performance status scale; Karnofsky 1948) and
physiological measures (e.g. step test, white cell count). The most common
measure of fatigue was the Chalder fatigue scale, followed by the Profile of
Mood States (POMS) fatigue scale (McNair 1992). The most common measures of
physical functioning were the Karnofsky scale and physical functioning
sub-scale of the Medical Outcomes Survey (MOS) SF-36 scale (Ware 1993).
Commonly used measures of depression included the Beck Depression Inventory
(Beck 1961), and Hospital Anxiety and Depression Scale (HADS; Zigmond 1983),
the latter also being the most common measure of anxiety along with the Beck
Anxiety Inventory (Beck 1990). Quality of life was most often measured by the
SF-36 in its entirety (including physical and mental functioning subscales).
Other outcomes had more heterogeneous measures.

Six studies (Sharpe 1993, Deale 1996, Prins 2001, Huibers 2004, Ridsdale
2004, O'Dowd 2000) used pre-determined criteria for successful outcomes on
major measures which were deemed to represent clinical improvement. These
outcomes were either of physical functioning (n=4) and/or fatigue (n=2).

Eight studies detailed dropout rates during treatment. Five studies reported
dropout rates across the duration of the trial including follow-up. One study
(Whitehead 2002) presented data as to how many patients were using activity
diaries at different points of follow-up. One study reported only the mean
dropout rate across all conditions (Jason 2007).

Excluded studies
CCDANCTR-Studies and CCDANCTR-Refs searches identified two studies (Friedberg
1994; Burgess 2001) that could not be immediately excluded by perusal of
title or abstract, but were deemed not to meet the inclusion criteria for the
review after scrutinising the full articles. Another ten studies identified
through supplementary searches or personal communication with experts in the
field were excluded after obtaining the full articles (Allen 2006,
Bleichhardt 2004, Butler 1991, Candy 2004, Donta 2003, Gielissen 2006,
Goudsmit 1996, Lyles 2003, Soderberg 2001, Taylor 2004). These are detailed
in the table entitled Characteristics of Excluded Studies, with reasons for
their exclusion from the updated version of this review.

Ongoing studies
Searches of CCDANCTR-Studies and CCDANCTR-Refs identified a total of six
ongoing studies (Wearden 2006, White 2005; Bleijenberg 2008; Carney 2008;
Gibson-Saxty 2002; Vissers 2008) that appear to meet the inclusion criteria
for this review. No findings have yet been reported or obtained from these
studies.


Studies awaiting assessment
Two studies were identified which may fulfil review criteria but could not be
included in the present review. The first study (Chalder 1997) randomised
participants to either a self-help booklet with specific advice
(incorporating cognitive behavioural elements), or a no treatment control
group. Participants were primary care patients complaining of fatigue who had
previously presented with a possible viral illness. Inclusion criteria were
deliberately broad, but included a subgroup of participants who met 'Oxford'
criteria for CFS (12% of sample). However, the review authors have as yet
been unable to obtain this subgroup data for inclusion in the review.

The second study (Powell 2001), compared "evidence based-explanation of
symptoms that encouraged graded activity" against standardised medical care.
Delivery of intervention was in three groups (minimum, telephone, and maximum
intervention groups) with varying amounts of face-to-face contact. Whilst
each intervention included some cognitive-behavioural principles, the review
authors have been unable to establish with the study authors which, if any,
of these interventions might satisfy inclusion criteria for this review.


Risk of bias in included studies

The methodological quality of the 15 studies included in the review was
classified according to the Cochrane Collaboration's tool for assessing the
risk of bias. The risk of bias in each of the six domains of validity
(sequence generation, allocation concealment, blinding of
participants/personnel/outcome assessors, incomplete outcome data, selective
outcome reporting, and other sources of bias) was rated independently by two
review authors as 'high', 'low' or 'unclear', and differences between authors
were reconciled. The final ratings are reported in the Characteristics of
Included Studies table. A risk of bias summary graph (Figure 1) and summary
figure (Figure 2) are also presented.


1) Sequence generation

Whilst all studies specified that participants were randomly allocated to
conditions (or GP practices randomised to treatment or control conditions in
the case of Whitehead 2002), only seven studies specified the method of
generating the sequence leading to random allocation, whether by computer
(n=3) or random number table (n=4). No study specified a sequence generation
method that was quasi- or non-random, therefore these seven studies received
a 'low' risk of bias rating for sequence generation, with the remaining eight
studies receiving a rating of 'uncertain'. Eight studies reported using
blocked randomisation, and four studies stratified their samples, by presence
of major depressive disorder (Sharpe 1993), by source of referral (Deale
1996), by length of work absence (Huibers 2004), or by duration and severity
of fatigue (Ridsdale 2004).


2) Allocation concealment

Method of concealing assignment to condition from investigators and
participants was described by seven studies through the use of numbered/
opaque envelopes (variously described as being opened in front of the
participant, immediately before his/her first session, or being handled by
third parties).

Nine studies conducted preliminary univariate analyses of demographic and
historical characteristics to check that randomisation had resulted in
appropriate comparability of groups for demographic characteristics, with six
reporting differences between groups. Sharpe 1993 reported that significantly
fewer CBT participants were actively employed, and spent significantly more
days in bed than TUC subjects. Deale 1996 reported that CBT participants were
on average seven years older than the relaxation group. A greater number of
the CBT group in Strang 2002 received disability compensation than the wait
list group, while the CBT patients in Whitehead 2002 were younger and had a
shorter duration of illness than their TUC counterparts. CBT subjects in
Ridsdale 2004 had a longer median duration of fatigue than GET participants.
In O'Dowd 2000, the CBT group had twice as many males as the other two arms.

Ten studies reported baseline differences between groups on outcome
variables, either explicitly or via inference from confidence intervals in
data tables. Of these ten studies, five reported that pre-treatment
differences did indeed exist. Stevens 1999 reported their CBT group had more
morning insomnia, awoke more at night, had worse SF-36 scores, and
outperformed the WL group on step-testing. The CBT participants in Whitehead
2002 were significantly more fatigued as measured by the Chalder fatigue
scale. TUC subjects in Huibers 2004 were significantly more distressed as
measured by the SCL-90 scores but had better physical functioning as measured
by MOS SF-36 scores, whilst the CBT subjects in Ridsdale 2004 were more
fatigued on the Chalder fatigue scale. CBT and anaerobic activity therapy
groups in Jason 2007 had higher scores on a self-efficacy measure than
subjects in the relaxation therapy condition.


3) Blinding of participants, personnel and outcome assessors

One trial (O'Dowd 2000) stated that "both the participants and those
administering the assessments were unaware of which cohort the subject was
in", and indeed described its protocol as a "double-blind randomised
controlled trial". The only formal test of participant blinding to treatment
group was in the case of active or placebo DLE injection in Lloyd 1993 (which
described itself as a 'double blind, randomised, placebo controlled study'),
which showed participants estimations of whether they received active or
placebo DLE bore no significant correlation with actual treatment received.
Only one study (Huibers 2004) specifically mentioned that researchers and
treatment providers were not blinded to treatment allocation.

Of the four studies comparing CBT with other psychological/exercise therapy
interventions, only two (King 1999 and Jason 2007) assessed treatment
fidelity. Of these, one study (King 1999) used four independent assessors of
recorded sessions and found differentiation in the management approaches of
the two groups according to specified protocol. The other (Jason 2007) used a
licensed and masters level psychologist (who had acceptable interrater
reliability) to rate a random sample of sessions according to therapists' use
of techniques unique to the four treatment conditions, and indeed found the
conditions were discriminant according to corresponding treatment manuals.
A third study (Deale 1996) did not formally assess treatment fidelity, but
did hold regular meetings to ensure protocol adherence. Two other studies
(Prins 2001, Huibers 2004) also assessed treatment fidelity, not to
differentiate CBT from other psychological or graded exercise treatment, but
to check quality of CBT intervention.

The outcome variables in ten studies were assessed by self-report scales or
were based on physiological data (e.g. step-test, white cell count). The
remaining five studies included outcome measures that were based on observer
assessment. Three of these studies used the Karnofsky scale. Of these, one
study (Prins 2001) used "an independent clinical psychologist" to rate
Karnofsky score who by this description is presumed to have been blind.
Another study (Sharpe 1993) reported using a blinded researcher to rate
participants Karnofsky status, on the basis of an interview made with
participants by another researcher; however, the blinding status of the
researcher conducting the interview was not specified. The remaining study to
use the Karnofsky scale (Lloyd 1993) made no mention of blinding of assessor.

Two further studies used assessors to rate the change of participants on
clinical dimensions. One study (Jason 2007) included a clinician global
impression of change in the patient's overall functioning. No mention of
blinding to treatment group was made. One study (Deale 1996) included an
assessor rating of participants' degree of fatigue and disability through a
structured interview. The assessors for this outcome were blinded through no
test of blind was made.


4) Incomplete outcome data

Dropout rates
Whilst not all studies differentiated between dropouts during therapy and
those lost to follow-up assessment, 14 studies reported dropout rates between
baseline and end of treatment (or first posttreatment assessment) by
comparison group. The remaining study (Jason 2007) reported a mean drop-out
rate across all conditions. Studies used varying criteria to define a
dropout. For example, Ridsdale 2004 used a conservative estimate of failing
to complete all six of the sessions of therapy specified in the study
protocol, whereas Jason 2007 reports a participant was deemed to have dropped
out if they completed four or fewer sessions out of 13 specified. Some
studies (e.g. O'Dowd 2000) defined dropouts not on the basis of intervention
sessions attended, but instead on whether participants attended follow-up
assessment or not. Another study (Prins 2001) used different criteria for
dropout between groups; in the TUC group, only participants not attending
assessments were classified as dropouts. In the CBT group, stopping coming to
therapy was defined as a dropout, whereas in the guided support group,
frequent non-attendance was not classified as a dropout (a dropout was only
counted if the participant formally declared an intention to withdraw).
Whitehead 2002 reported dropouts as those unavailable for assessment at six
months (31% of subjects). However, over 70% of subjects had terminated use
of their treatment diaries or GP contact by that time.

The mean aggregate reported dropout rate of all 15 studies was 16.4%. Dropout
rates ranged from 0% (Sharpe 1993) to 40% (Whitehead 2002), with six studies
(Prins 2001, King 1999, Strang 2002, Whitehead 2002, Ridsdale 2004, Jason
2007) reporting dropout rates of over 20%. For the 14 studies reporting
dropout rates by comparison group, the mean dropout rate for the CBT
intervention was 16.8%. Of the four studies comparing CBT with another active
psychological/graded exercise therapy, only one (Jason 2007) detailed that
there were no differences between dropout rates of comparison groups (mean
25%). Dropout rates were broadly similar in the other three studies: 10% CBT
vs 13.3% relaxation (Deale 1996); 36.3% CBT vs. 31.3% counselling (King
1999); and 28.6% versus 40% graded exercise therapy (Ridsdale 2004). Dropout
rates between comparison groups also appeared broadly similar in the three
studies comparing CBT with a guided support/education arm: 6.7% CBT versus 0%
education and support (Barrett 1992); 36.6% CBT versus 30.9% guided support
(Prins 2001); and 5.8% versus 10% education and support (O'Dowd 2000).

Reasons for dropouts
Five studies (Barrett 1992, Lloyd 1993, Deale 1996, Stevens 1999, Whitehead
2002) reported reasons for dropouts during treatment by comparison group. In
none of these studies did the reasons for dropout between groups appear
dissimilar. A further study (Huibers 2004) reported reasons for dropout by
CBT group but not for any other comparison group, and two other studies (King
1999, Ridsdale 2004) gave aggregate reasons for dropout not detailed by
comparison groups. Four studies (Deale 1996, King 1999, Huibers 2004,
Ridsdale 2004) investigated differences between dropouts and completers on
demographic and baseline variables.

Method of dealing with missing data
Five studies (Lloyd 1993, Prins 2001, Russell 2001, Whitehead 2002, O'Dowd
2000) did not provide means and/or standard deviation (SD) data for outcome
variables. Baseline data for completers was also not inferable from the data
presented by King 1999. Where data was required for this review, authors were
contacted. Data were received from the authors of Prins 2001 (with the
exception of two variables which were unobtainable) and King 1999.

Eight studies reported how they dealt with missing data (from dropouts or due
to another reason such as inability/unwillingness to complete a walking
test). Seven of these studies (Sharpe 1993, Deale 1996, Prins 2001, King
1999, Huibers 2004, Ridsdale 2004, O'Dowd 2000) presented sensitivity
analyses (examining any differential effect of carrying forward data, using
multiple value imputation, or abolishing dropouts from final analysis) or
employed a full 'intention to treat' analysis carrying forward previous data
values or assuming a negative outcome for dropouts.


5) Selective outcome reporting

Only one study (O'Dowd 2000) made reference to deviation from its trial
protocol, and only one study (Barrett 1992) specified an outcome variable
which it did not report (delayed type hypersensitivity skin-test). Two other
studies indicated outcomes for which they did not then provide data (SF-36 in
King 1999; HADS in Whitehead 2002).

Whilst Lloyd 1993 collected data concerning the adverse effects of DLE
injection, data referring to adverse effects of psychological treatment was
not systematically presented by any study. One study (Sharpe 1993) noted
reasons given by subjects for any deterioration during treatment, with two of
four CBT participants who were subjectively 'worse' or 'much worse' after
treatment blaming the treatment for deterioration. Huibers 2004 noted that
'no adverse event attributable to CBT was reported'. Jason 2007 reported
numbers of participants who endorsed 'worse' or 'unchanged' to a 'Treatment
made me feel' statement, but did not differentiate between the two categories
or further investigate reasons, making inference of adverse effects
impossible.


6) Other sources of bias

Five studies (Barrett 1992, Stevens 1999, Russell 2001, Strang 2002, Surawy
2005) made use of a wait-list control condition, a design feature that may
affect the natural course of fatigue or other symptoms, either in a directive
direction through the implicit suggestion that recovery should be placed 'on
hold' until treatment, or in a positive direction, through anticipated
'rescue' by treatment. Less than half of the included studies reported the
use of manuals or treatment protocols to standardise the CBT provided, and
only three studies reported testing treatment fidelity through assessment of
a random selection of video or audio tapes of sessions (King 1999; Jason
2007; Prins 2001), thus introducing uncertainty that therapists provided
and adhered to CBT interventions in sessions.


Effects of interventions

Of 15 studies included in the review, 12 studies contributed data to primary
and secondary outcomes, and three studies contributed to the secondary
outcome of dropout only. Statistical heterogeneity was examined for each
outcome, and where indicated to be statistically significant, I^2 figures
were reported in the text. The fixed effects model was used for all outcomes
unless otherwise stated in the text. Subgroup analyses of the primary outcome
were conducted where data from at least two studies were available for each
sub-group.

COMPARISON 01: COGNITIVE BEHAVIOUR THERAPY vs USUAL CARE
Nine studies contributed to this comparison at post-treatment (Prins 2001,
Huibers 2004, O'Dowd 2000, Barrett 1992, Strang 2002, Lloyd 1993, Russell
2001, Whitehead 2002) with an additional study providing data at follow-up
(Sharpe 1993). The studies by Lloyd 1993, Russell 2001 and Whitehead 2002
provided post-treatment data for the secondary outcome of attrition only.
For the purposes of this comparison, short and medium-term follow-up outcomes
were combined.


Primary outcome

Seven studies contributed data to the primary outcome meta-analysis. Measures
used to assess fatigue symptoms and clinical response comprised the fatigue
subscale of the CIS (Huibers 2004, Prins 2001), Chalder Fatigue Scale (O'Dowd
2000, Surawy 2005), POMS fatigue scale (Barrett 1992), Fatigue Severity Scale
(Strang 2002) and a single-item fatigue Likert scale (Sharpe 1993).


1) Reduction in fatigue severity

Post treatment (Graph 01 01)
Five traditional CBT studies and one mindfulness/compassion-focused CBT
study (373 participants in total) contributed to this outcome. The difference
in fatigue mean scores between the CBT group and usual care group was highly
significant in favour of the CBT group (SMD -0.39, 95% CI -.0.60 to -0.19).
No statistical heterogeneity was indicated.

Short/medium term follow up (Graph 01 09)
Four traditional CBT studies (330 participants in total) contributed to this
outcome. The difference in fatigue severity mean scores between the CBT and
usual care group was highly significant in favour of CBT (SMD -0.47, 95%CI
-0.69 to -0.25). No statistical heterogeneity was indicated.


2) Clinical response

Post treatment (Graph 01 02)
Four studies (371 participants) contributed to this outcome. A total of 40%
of participants in the CBT group showed clinical response to treatment, in
contrast with 26% in the treatment as usual care group. The difference
between the two groups was highly significant (OR 0.47, 95% CI 0.29 to 0.76).
No statistical heterogeneity was indicated.

Short term follow up (Graph 01 10)
Three studies (353 participants) contributed to this outcome. No significant
difference in clinical response rates was indicated between the CBT and usual
care groups (OR 1.03, 95% CI 0.53 to 2.00). Some heterogeneity was indicated
(I^2=40%), and a random effects model was used.


Secondary outcomes

1) Improvement in physical functioning

Post treatment (Graph 01 03)
Three traditional CBT studies and one mindfulness/compassion-focused CBT
study (318 participants in total) contributed to this outcome. Measures used
to assess physical functioning comprised Karnofsky performance status scale
(one study) and SF-36 physical functioning subscale (three studies). The
difference in physical functioning mean scores between the CBT group and the
usual care group was not significant (SMD 0.11, 95%CI -0.32 to 0.54).
Significant heterogeneity was indicated (I^2=68%) and a random effects model
was used.

Short/medium term follow-up (Graph 01 11)
Three traditional CBT studies (275 participants) contributed to this outcome.
The difference in physical functioning mean scores between the CBT group and
the usual care group was not significant (SMD 0.15, 95%CI -0.21 to 0.51).
Significant heterogeneity was indicated (I^2=53%) and a random effects model
was used.


2) Reduction in depression symptoms

Post treatment (Graph 01 04)
Three traditional CBT studies and one mindfulness/compassion focused CBT
study (183 participants in total) contributed to this outcome. Measures used
to assess depression symptoms comprised the depression subscale of the HADS
(two studies) and BDI (two studies). The difference in depression mean scores
between the CBT group and the usual care was not significant (SMD -0.24,
95%CI -0.53 to 0.05). No statistical heterogeneity was indicated.

Short/medium term follow-up (Graph 01 12)
Two traditional CBT studies (149 participants) contributed to this outcome.
Both studies used the HADS depression subscale to measure depression
symptoms. The difference in depression mean scores between the CBT group and
the usual care group was significant (MD -1.39, 95%CI -2.60 to -0.18). No
significant heterogeneity was indicated.


3) Reduction in anxiety symptoms

Post treatment (Graph 01 05)
Three traditional CBT studies and one mindfulness/compassion focused CBT
study (183 participants in total) contributed to this outcome. Measures used
to assess anxiety symptoms comprised the anxiety subscale of the HADS (two
studies), BAI (one study) and the STAI (one study). The difference in anxiety
mean scores between the CBT group and usual care was significant (SMD -0.30,
95%CI -0.59 to -0.01). No statistical heterogeneity was indicated.

Short/medium term follow-up (Graph 01 13)
Two traditional CBT studies (149 participants in total) contributed to this
outcome. Both studies used the HADS anxiety subscale of HAD to measure
anxiety symptoms. The difference in anxiety mean scores between the CBT group
and the usual care group was not significant (MD -1.38, 95% CI -2.91 to
0.16). No statistical heterogeneity was indicated.


4) Reduction in psychological distress

Post treatment (Graph 01 06)
Three traditional CBT studies (191 participants in total) contributed to this
outcome. Measures used to assess psychological distress symptoms comprised
the SCL-90 (one study), the GHQ (one study) and a single item Likert scale of
distress (one study). The difference in mean scores between the CBT group and
usual care did not reach significance (SMD -0.27, 95%CI -0.56 to 0.01). No
significant heterogeneity was indicated.

Short term follow-up (Graph 01 14)
One traditional CBT study (61 participants) contributed to this outcome. The
difference in psychological distress mean scores between the CBT group and
the usual care group was not significant (MD -0.24, 95% CI -28.95 to 28.37).


5) Improvement in quality of life

Post treatment (Graph 01 07)
One traditional CBT study (184 participants) contributed to this outcome. The
measure used to assess improvement in quality of life comprised the Euro-Qol.
The difference in QoL improvement rates between the CBT group and the usual
care group was not significant (OR 1.19, 95%CI 0.58 to 2.46).

Short term follow-up (Graph 01 15)
One traditional CBT study (125 participants) contributed to this outcome. The
difference in QoL mean scores between the CBT group and the usual care group
was significant (MD 8.00, 95% CI 0.68 to 15.32).


6) Acceptability of treatment

No studies contributed to this outcome at post treatment or follow-up.


7) Adverse effects

No studies contributed to this outcome at post treatment or follow-up.


8) Overall dropout rates

Post treatment (Graph 01 08)
Eight traditional CBT studies and one mindfulness/compassion-focused study
(586 participants in total) contributed data to this outcome. The attrition
rate was 20.6% in the CBT group and 13.9% in the usual care group. The
difference in attrition rate between the CBT group and the usual care group
was significant in favour of usual care (OR 1.70, 95% CI 1.10 to 2.63). No
statistical heterogeneity was indicated.

Short/medium term follow-up (Graph 01 16)
Four traditional CBT studies (413 participants in total) contributed data to
this outcome. One study had no dropouts. The attrition rate was 22.7% in the
CBT group and 14.8% in the usual care group. The difference was significant
in favour of the usual care group (OR 1.46, 95% CI 0.52 to 4.10). Statistical
heterogeneity was indicated (I^2=56%) and a random effects model was used.


9) Occupational outcomes

Short/medium term follow-up (Graph 01 17)
One traditional CBT study (61 participants) contributed data to this outcome,
in which the number of days of absenteeism was measured during the 12-month
follow-up period. The difference between the CBT group and usual care group
was not significant (MD 4.35, 95% CI -50.06 to 58.76).


10) Health service resource use

No studies contributed data on health service resource use. Sub-group
analyses Subgroup analyses were conducted for the outcome of reduction in
fatigue symptoms at post treatment.

a) Individual vs group therapy (Graph 05 01) Two studies used an individual
CBT modality and four studies used group CBT. A highly significant difference
in effect was shown for individual therapy and a significant difference of
lesser magnitude was shown for group therapy when compared with usual care.

b) Treatment as usual vs waiting list (Graph 05 02) Three studies used a
treatment as usual condition and three studies used a waiting list as the
control condition. A highly significant difference in effect was shown for
the CBT group when compared with treatment as usual. In contrast a
non-significant difference in effect was shown for the CBT group when
compared with the waiting list condition.

c) Increased activity as component of CBT (Graph 05 03) Four studies
incorporated an increased activity component into the CBT intervention and
two studies tailored activity and rest to the needs of the individual or did
not specify an activity schedule. A highly significant difference in effect
was shown for the CBT group in which increased activity was incorporated when
compared with usual care. In contrast, a non-significant difference in effect
was shown for the studies in which increased activity was not part of the CBT
intervention when compared with usual care. d) Number of sessions (Graph 05
04)

Four studies offered eight or fewer sessions of CBT, and two studies offered
more than eight sessions of CBT. A highly significant difference in effect of
similar magnitude was shown for both CBT groups when compared with usual
care.


COMPARISON O2: COGNITIVE BEHAVIOURAL THERAPY vs OTHER PSYCHOLOGICAL THERAPIES

Four studies contributed to this comparison at post-treatment (Barrett 1992,
Deale 1996, Prins 2001, O'Dowd 2000), with two additional studies
contributing data at short-term (King 1999) and medium term (Jason 2007)
follow-up. The studies compared a traditional CBT approach with relaxation
(Deale 1996, Jason 2007), counselling (King 1999), guided support (Prins
2001) and education and support (O'Dowd 2000, Barrett 1992).


Primary outcome

All six studies contributed data to the primary outcome metaanalysis.
Measures used to assess fatigue symptoms and clinical response comprised the
fatigue subscale of the CIS (Prins 2001), Chalder Fatigue Scale (O'Dowd 2000,
Deale 1996, King 1999), POMS fatigue scale (Barrett 1992) and Fatigue
Severity Scale (Jason 2007).


1) Reduction in fatigue severity

Post treatment (Graph 02 01)
Four studies (313 participants in total) contributed to this outcome. The
difference in fatigue mean scores between the CBT group and other
psychological therapies group was highly significant in favour of the CBT
group (SMD -0.43, 95% CI -0.65 to -0.20). No statistical heterogeneity was
indicated.

Short term follow up (Graph 02 09)

Four studies (294 participants in total) contributed to this outcome. The
difference in fatigue severity mean scores between the CBT group and other
psychological therapies group was significant in favour of CBT (SMD -0.47,
95%CI -0.83 to -0.11). Significant heterogeneity was indicated (I^2=54%) and
a random effects model was used.


2) Clinical response

Post treatment (Graph 02 02)
Three studies (349 participants in total) contributed to this outcome. A
total of 48% of participants in the CBT group showed clinical response to
treatment, in contrast with 27% in the other psychological therapies group.
The difference between the two groups was highly significant in favour of CBT
(OR 0.35, 95% CI 0.19 to 0.62). No statistical heterogeneity was indicated.

Short/medium term follow up (Graph 02 10)
Three studies (219 participants in total) contributed to this outcome. The
difference in clinical response rates between the CBT and other psychological
therapies groups was not significant (OR 0.39, 95% CI 0.14 to 1.12).
Statistical heterogeneity was indicated (I^2=59%) and a random effects model
was used.

Long-term follow-up
One study (60 participants in original study, 53 participants in follow-up
study) contributed to this outcome. A total of 13/25 (52%) of participants in
the CBT group no longer met the UK diagnostic criteria for CFS, in comparison
with 11/28 (39%) participants in the other psychological therapies group, who
had received relaxation (Fisher's exact test p=0.42). A total of 6/25 (24%)
participants in the CBT group had made a complete recovery in comparison with
one participant (4%) in the relaxation group (Fisher's exact test p=0.02).


Secondary outcomes

1) Improvement in physical functioning

Post treatment (Graph 02 03)
Three studies (286 participants in total) contributed to this outcome.
Measures used to assess physical functioning comprised the Karnofsky
performance status scale (one study) and SF-36 physical functioning subscale
(two studies). The difference in physical functioning mean scores between the
CBT group and the other psychological therapies group was significant (SMD
0.53, 95%CI 0.10 to 0.97). Significant heterogeneity was indicated (I^2=68%) 
and a random effects model was used.

Short term follow-up (Graph 02 11)
Three studies (257 participants in total) contributed to this outcome. The
difference in physical functioning mean scores between the CBT group and the
other psychological therapies group was highly significant (SMD 0.71, 95%CI
0.27 to 1.14). Statistical heterogeneity was indicated (I^2=64%) and a
random effects model was used.


2) Reduction in depression symptoms

Post treatment (Graph 02 04)
Three studies (175 participants in total) contributed to this outcome.
Measures used to assess depression symptoms comprised the depression subscale
of the HADS (one study) and BDI (two studies). The difference in depression
mean scores between the CBT group and other psychological therapies group was
significant (SMD -0.46, 95%CI -0.76 to -0.16). No statistical heterogeneity
was indicated.

Short term follow-up (Graph 02 12)
Three studies (257 participants in total) contributed to this outcome. The
difference in depression mean scores between the CBT group and the other
psychological therapies group was non-significant (SMD -0.03, 95%CI -0.50 to
0.44). Statistical heterogeneity was indicated (I^2=57%) and a random
effects model was used.


3) Reduction in anxiety symptoms

Post treatment (Graph 02 05)
Two studies (122 participants in total) contributed to this outcome. Measures
used to assess anxiety symptoms comprised the anxiety subscale of the HADS
(one study) and the STAI (one study). The difference in anxiety mean scores
between the CBT group and other psychological therapies group was significant
(SMD -0.41, 95%CI -0.71 to -0.11). No statistical heterogeneity was
indicated.

Short term follow-up (Graph 02 13)
Two studies (124 participants in total) contributed to this outcome. The
difference in anxiety mean scores between the CBT group and the other
psychological therapies group was not significant (SMD -0.03, 95% CI -2.96 to
3.02). Statistical heterogeneity was indicated (I^2=66%) and a random
effects model was used.


4) Reduction in psychological distress

Post treatment (Graph 02 06)
Three studies (175 participants in total) contributed to this outcome.
Measures used to assess psychological distress symptoms comprised the GHQ
(two studies) and a single item Likert scale of distress (one study). The
difference in mean scores between the CBT group and other psychological
therapies was significant (SMD -0.41, 95%CI -0.71 to -0.11). No statistical
heterogeneity was indicated.

Short term follow-up (Graph 02 14)
One study (88 participants) contributed to this outcome. The difference in
psychological distress mean scores between the CBT group and the other
psychological therapies group was significant (MD -3.60, 95% CI -7.07 to
-0.13).


5) Improvement in quality of life

Post treatment (Graph 02 07)
One study (124 participants) contributed to this outcome. The measure used to
assess improvement in quality of life comprised the Euro-Qol. The difference
in QoL mean scores between the CBT group and the other psychological
therapies group was significant (MD 10.00, 95%CI 3.43 to 16.57).

Short term follow-up (Graph 02 15)
One study (116 participants) contributed to this outcome. The difference in
QoL mean scores between the CBT group and the other psychological therapies
group was highly significant (WMD 13.00, 95% CI 5.48 to 20.52).


6) Acceptability of treatment

a) Satisfaction with care/therapy (short-term follow-up) One study
contributed to this outcome (Deale 1996). A total of 21/27 participants (78%)
reported being satisfied or very satisfied with the treatment outcome, in
comparison with 13/26 (50%) in the other psychological therapies group
(relaxation) (Chi-square 2.1, p<0.05).

b) Dropout due to adverse effects No studies contributed to this outcome at
post treatment or follow-up.


7) Adverse effects

No studies contributed to this outcome at post treatment or follow-up.


8) Overall dropout rates

Post treatment (Graph 02 08)
Four studies (377 participants in total) contributed data to this outcome.
The attrition rate was 24.7% in the CBT group and 20.3% in the other
psychological therapies group. The difference in attrition rate between the
two groups was not significant (OR 1.34, 95% CI 0.81 to 2.20). No statistical
heterogeneity was indicated.

Short term follow-up (Graph 02 16)
Four studies (394 participants in total) contributed data to this outcome.
The attrition rate was 27.2% in the CBT group and 23.6% in the other
psychological therapies group. The difference in attrition rate between the
two groups was not significant (OR 1.23, 95% CI 0.76 to 1.97). No statistical
heterogeneity was indicated.


9) Occupational outcomes

No studies contributed to this outcome at post treatment or follow-up.


10) Health service resource use

No studies contributed to this outcome at post treatment or follow-up.


COMPARISON 03: COGNITIVE BEHAVIOURAL THERAPY vs EXERCISE

One study contributed data to this comparison at post-treatment (Ridsdale
2004) and one study contributed data to this comparison at follow-up (Jason
2007).


Primary outcome

Both studies contributed data to the primary outcome meta-analysis.
Measures used to assess fatigue symptoms comprised the Chalder Fatigue Scale
(Ridsdale 2004) and the Fatigue Severity Scale (Jason 2007).


1) Reduction in fatigue severity

Post treatment (Graph 03 01)
One study (36 participants) contributed to this outcome. The difference in
fatigue mean scores between the CBT group and exercise group was not
significant (MD -2.52, 95% CI -7.34 to 2.30).

Follow-up
No studies contributed data to this outcome.


2) Clinical response

Post treatment (Graph 03 02)
One study (36 participants) contributed data to this outcome. The difference
in clinical response rates between the CBT and exercise groups was not
significant (OR 0.47, 95% CI 0.10 to 2.17).

Medium term follow up (Graph 03 03)
One study (57 participants) contributed data to this outcome. The difference
in clinical response rates between the CBT and exercise groups was not
significant (OR 0.10, 95% CI 0.01 to 1.94).


Secondary outcomes

No studies contributed data to secondary outcomes at post treatment or at
follow-up.


COMPARISON O4: COGNITIVE BEHAVIOURALTHERAPY COMBINED WITH OTHER INTERVENTIONS
VS USUAL CARE

One study contributed to this comparison at post-treatment (Stevens 1999),
and compared CBT combined with biofeedback, sleep hygiene, relaxation and
aerobic exercise against usual care.


Primary outcome

One study contributed towards the primary outcome meta-analysis. The
measure used to assess fatigue symptoms comprised the POMS fatigue scale.


1) Reduction in fatigue severity

Post treatment (Graph 04 01) One study (27 participants) contributed to this
outcome. The difference in fatigue mean scores between the combination CBT
group and the usual care group was significant in favour of the combination
CBT group (WMD -7.12, 95%CI -12.83 to -1.41).

Follow-up
No studies contributed data to this outcome.


2) Clinical response

No studies contributed data to this outcome at post treatment or follow-up.


Secondary outcomes

1) Improvement in physical functioning

Post treatment (Graph 04 02)
One study (27 participants) contributed to this outcome. The measure used to
assess physical functioning comprised the SF-36 physical functioning
subscale. The difference in physical functioning mean scores between the
combination CBT group and usual care group was significant (MD 8.38, 95%CI
0.99 to 15.78).

Follow-up
No studies contributed data to this outcome


2) Overall dropout rates

Post treatment (Graph 04 03)
One study (29 participants) contributed data to this outcome. The difference
in attrition rate between the two groups was not significant (OR 0.16, 95% CI
0.01 to 3.68).

Follow-up
No studies contributed data to this outcome

Other secondary outcomes
No studies contributed data to other secondary outcomes at post treatment or
follow-up.


Consideration of publication bias

Funnel plots were produced for the Comparison 01 primary outcome of reduction
of anxiety symptoms (6 studies). Visual inspection of the funnel plot
indicated possible asymmetry, which might suggest that small trials with
negative outcomes were not included in the review. However, the small number
of studies included in the funnel plot limits further meaningful
interpretation.

--------------------------------------------------------------------------------
DISCUSSION

Summary of main results

In the original review, no meta-analyses were performed, since only three
studies were eligible for inclusion at that time, and each study used a
different control intervention. This updated version of the review includes
an additional 12 studies conducted since 1999. All 15 included studies
provided some data for metaanalyses, and four comparisons were performed. A
summary of the main findings for each comparison follows.

Cognitive behavioural therapy versus usual care
Based on a homogeneous group of six studies (373 participants), the review
provides evidence that at post-treatment, CBT was more effective than usual
care in reducing the severity of fatigue symptoms in CFS patients who had
persisted with treatment. The robustness of this finding is supported by the
outcome of clinical response (4 studies, 371 participants), with a 40%
response rate for CBT compared to 26% for usual care. However, at short to
medium term follow-up (rather than immediately post-treatment), the
evidence for the effectiveness of CBT was inconclusive. Anxiety symptoms
were lower at post treatment in patients who completed CBT sessions than
those assigned to usual care, but not at follow-up. Patients assigned to
usual care were less likely to drop out of treatment than those assigned to
CBT.

Four sub-group analyses were conducted. The first sub-group analysis found
that individual CBT interventions were highly effective in reducing fatigue
severity compared with usual care at post treatment, whereas the difference
between a group modality of CBT and usual care only just reached
significance. The second found that CBT was highly effective in reducing
fatigue severity when compared with treatment as usual control, and, in
contrast, there was no significant difference between CBT and waiting list
control. The third found that CBT in which increased activity and reduced
rest was a component of treatment was highly effective in reducing fatigue
severity compared with usual care, and, in contrast, CBT without focus on
increased activity, or in which this was not specified, was no more effective
than usual care. Finally, both 8 or fewer sessions of CBT and more than 8
sessions were highly effective in reducing fatigue severity compared with
usual care. Given the non-randomisation of studies into subgroups coupled
with the small number of studies, the significance of these findings should
be interpreted with caution.

Cognitive behavioural therapy versus other psychological therapies
Based on a homogeneous group of four studies (309 participants) the review
provides evidence that CBT was more effective than other psychological
therapies in reducing the severity of fatigue symptoms in CFS patients
persisting with treatment at post treatment. This finding is supported by the
clinical response outcome (3 studies, 259 participants), with a 48% response
rate for CBT in contrast with 28% for other psychological therapies. Patients
who completed CBT sessions achieved a greater improvement in physical
functioning, together with lower depression, anxiety and psychological
distress symptoms at post-treatment than those who completed other
psychological therapy sessions. At short term follow-up, the difference
between CBT and other psychological therapies was inconsistent, and
statistical heterogeneity was indicated.

Cognitive behavioural therapy versus exercise
The evidence is limited to two studies, both of which independently showed a
lack of difference between CBT and exercise in reducing fatigue levels or
increasing clinical response, at post treatment and at medium term follow-up.

Cognitive behavioural therapy combined with other interventions versus usual
care
The evidence is limited to a single study of 27 participants, which showed
that at post-treatment, CBT combined with biofeedback, relaxation and aerobic
exercise was more effective in reducing fatigue severity than usual care.


Overall completeness and applicability of evidence

Participants
The majority of studies included in this review were conducted in outpatient
departments or specialist clinics, and two thirds of studies excluded
patients who were 'incapable' of attending sessions or confined to
wheelchairs/bed. In limiting the population to those patients who were more
mobile, the findings from this review may not be representative of more
severely affected individuals. Nevertheless, chronicity was a key clinical
feature for participants in included studies, with mean duration of illness
estimated at 4.7 years.

The updated version of the review includes four trials in which recruitment
was wholly or partially conducted in primary care settings (Huibers 2004;
King 1999; Ridsdale 2004; Whitehead 2002). This potentially increases the
generalisability of the findings to individuals with less disabling and
severe symptoms. However, with each of these four studies, this review
restricted the analysed dataset to the subgroup of participants diagnosed
with CFS. The rationale for this was the minimisation of diagnostic
heterogeneity, although it is acknowledged that this may have reduced the
applicability of the findings to primary care settings.

Interventions
The updated review shows that the largest body of evidence comprises
traditional CBT interventions adapted for CFS populations, in which
increasing activity and reducing rest time are core features of treatment. In
line with the emergence of newer 'third wave' CBT interventions, however, a
recently conducted study of mindfulness CBT has also been included in the
review (Surawy 2005). In this intervention, participants are encouraged to
observe symptoms, thoughts and emotions without attempting to avoid or change
them. It seems likely that further trials informed by compassion focused and
mindfulness principles will be conducted and included in future updates.

Given the complex and often chronic clinical presentation of CFS, it seems
conceivable that a combination of interventions might be required for the
optimal treatment of this disorder. In updating this review, it had been
hoped to include evidence from trials of combination therapies. Surprisingly,
however, despite an exhaustive search of the literature, the evidence base
for combination therapy appears to contain just two studies (Stevens 1999;
Lloyd 1993). This precludes the ability to draw conclusions on which
interventions might be more effective in combination rather than as
stand-alone treatments.

During this review it was also planned to conduct formal comparisons between
CBT alone and other active interventions alone. However, to date, the
evidence base is limited to just two studies, both of which compared CBT to a
manualised exercise intervention. One additional group of studies (Barrett
1992; Deale 1996; O'Dowd 2000; Prins 2001; King 1999) compared CBT with
other psychological therapy interventions, including relaxation, education
and support and counselling. However, these interventions were largely
intended as attention placebo conditions, rather than as active treatments.

Outcomes
An encouraging feature of recent trials examining the effectiveness of CBT
for CFS is the use of more extended follow-up periods. The majority of
studies conducted over the last eight years incorporated follow-up time
points 6-12 months post-treatment, with one study (Deale 1996) conducting a
five year follow-up with 88% of the original sample, thus providing a more
clinically relevant measure of effectiveness and cost-effectiveness.

Effective treatments are of little value to health services unless they are
acceptable to the recipients - poor acceptability will reduce treatment
uptake and adherence. Only three trials included in this review incorporated
the outcome of satisfaction with treatment (Deale 1996; Jason 2007; Ridsdale
2004), with usable data available from one study only (Deale 1996). Given the
understandable ambivalence of many CFS patients towards a psychological
treatment for symptoms that are experienced as physical, the
under-investigation of acceptability of treatment is an important omission.
Other outcomes which are of high relevance to the individual with CFS,
including adverse effects, health-related quality of life, and occupational
outcomes, are also under-evaluated in the included studies.


Quality of the evidence

Historically, trials of psychological therapy have provided poor quality
information on the specific randomisation procedure used to assign
participants to groups. Therefore, it is encouraging that nearly half of the
studies included in the updated review described and used appropriate methods
of sequence generation and allocation concealment. Nevertheless, many studies
failed to describe their randomisation procedure, and, therefore, it is not
possible to assess the extent to which selection bias may have occurred in
this group of studies.

Blinding of therapists and participants is not feasible in trials of
psychological therapy, although it is notable that two studies in the review
did make strenuous attempt to reduce or avoid contamination between groups
(Prins 2001; O'Dowd 2000). Whilst the use of blinded assessors is feasible in
psychological therapy trials, the majority of studies used subjective
self-report measures to collect psychological outcomes. In general, little
information was provided on the data collection process, and the extent to
which response bias may have occurred is uncertain. Of the five studies that
used clinician-rated measures, only three studies reported the use of
independent or blinded assessors (Deale 1996; Prins 2001; Sharpe 1993) to
minimise detection bias.

Waiting list is frequently employed as an ethical 'no treatment' condition,
to ensure that all participants eventually receive the 'active' treatment.
This approach was used in almost half of all studies comparing CBT against
usual care. However, sub-group analyses showed that, at post-treatment,
waiting list patients had reduced fatigue symptoms equivalent to those of
patients attending for treatment. A possible explanation is that being on the
waiting list for treatment had a 'holding' positive influence, decreasing the
apparent effect of the CBT intervention.

Because psychological therapy involves the application of complex techniques
over a period of time, it is important from a methodological perspective that
each intervention is operationally defined by detailed protocols (Borkovec
2001). One of the methodological weaknesses of trials included in this review
was the apparent failure of eight of the fifteen studies to use manuals or
treatment protocols to more effectively standardise the CBT provided. Most
studies did, however, employ therapists who were experienced in the
psychological model under examination. Testing of therapists' fidelity to
treatment manuals through the systematic or random checking of audiotapes by
independent clinicians is an additional key methodological aspect of
psychological therapy studies. This ensures that any observed treatment
effect can be attributed to specific components and characteristics of the
model. However, only three of the fifteen included studies reported testing
therapists' treatment fidelity through assessing recordings of sessions
(Prins 2001; King 1999; Jason 2007). Therefore, there is no certainty in most
of the studies that therapists were adhering to the required psychological
model.

Four studies controlled for therapist attention by using low intensity
interventions such as education and support or relaxation. When compared
against these placebo interventions, CBT appeared to achieve a more
consistently positive effect across all outcomes than when compared against
usual care, a finding that appears counter-intuitive. It is possible that the
findings could be explained by the greater use of naturalistic treatments in
usual care conditions. However, another plausible explanation is the
influence of researcher allegiance to CBT, as in three of these studies the
therapist was also the primary investigator (Barrett 1992; Deale 1996; O'Dowd
2000). In addition, in the study by Prins 2001, the control condition was
group therapy with less emphasis given to regular attendance, in contrast
with the CBT intervention, which was individual in modality. Whilst it is
possible, therefore, that the effect of CBT may have been over-estimated when
compared with the low intensity conditions, it should nevertheless be
acknowledged that CBT and placebo conditions appeared to be managed equally
across studies in terms of manualisation, fidelity checking, and therapist
qualifications/experience.

Whilst the overall attrition rate across studies was reasonably low at 16.4%,
six of the fifteen studies had dropout rates of higher than 20%, and only
five studies reported reasons for attrition in full for each group.
Therefore, it is not certain to what extent there might have been systematic
differences between dropouts and completers between groups that could have
influenced treatment outcomes, leading to an underestimate or overestimate of
effect. In contrast, however, missing data was well managed in most studies,
with nine studies conducting a sensitivity analysis, analyses using intention
to treat principles, or last observation carried forward.


Potential biases in the review process

Strengths

A core strength of this review is the extensive, rigorous and highly
sensitive search strategy used to locate high quality evidence. As a result,
randomised evidence from unpublished as well as published studies were
included in the review, none of which have been included in previous high
quality systematic reviews. Furthermore, vigorous efforts made to obtain
missing data resulted in additional data from three investigators (King 1999;
Huibers 2004; Prins 2001), and responses from two further investigators who
may be able to provide access to their datasets in future updates (Chalder
1997; Russell 2001). It seems likely that all or almost all evidence that
should have been included was included and that the occurrence of
publication bias may have been minimised. The increased number of studies,
and their relative homogeneity, permitted a quantitative synthesis of some of
the data, thus improving the precision of the estimates of the effectiveness
of CBT

A further strength is the diversity of the authors' areas of expertise,
coupled with collaborative and transparent review methods, ensuring that the
potential for bias was minimised. Two independent review authors were
involved in every stage of the review process and a third author mediated
when necessary. During study selection, the authors' different perspectives
helped to inform inclusion-exclusion decisions when, for example, it was
difficult to assess the extent to which interventions were sufficiently
CBT-based to be included in the review. Of note, in the case of one study
where it was not possible to reach agreement on its inclusion, the authors of
the study concerned (Powell 2001) were invited to cast the deciding vote.


Limitations

Although every effort was made to obtain missing data from trial authors, it
was not possible in every case to obtain these data, and therefore the
included studies are not represented fully in the meta-analyses.

Given that formal blinding of participants and clinicians to treatment arm is
not inherently possible in trials of psychological therapy, the decision was
taken to assess the risk of bias for the blinding domain on the basis of
outcome assessor blinding only, whilst acknowledging that our decision to do
this could have under-estimated the extent to which clinicians' and
participants' knowledge of group assignation influenced the true effect.

Whilst the identification and inclusion of four unpublished studies (Barrett
1992; Russell 2001; Strang 2002; Stevens 1999) is regarded as a strength of
the updated review in terms of reducing publication bias, it should
nevertheless be acknowledged that these trials have not been through full
peer review and may have been especially vulnerable to bias.


Agreements and disagreements with other studies or reviews

The key high quality review of evidence in this area is that by Chambers and
colleagues (Chambers 2006), which helped to inform the UK's NICE Clinical
Guideline on the care of people with CFS-ME (NICE 2007). The Chambers 2006
review addressed a very wide range of therapeutic modalities, including not
only CBT and graded exercise, but also immunological, complementary,
pharmacological, and nutritional therapies. Like this review, it noted a
marked increase in the size and the quality of the evidence-base for
interventions for CFS. It concluded that two interventions, CBT and graded
exercise, appeared to be effective, but that there was either evidence of
lack of effectiveness or a lack of evidence for other interventions. However,
the conclusions drawn by Chambers 2006 for CBT interventions were based on a
smaller dataset than that provided in this review, included non-randomised
controlled trials, and no meta-analyses were undertaken. A recently published
systematic review and meta-analysis by Malouff 2008, using broader inclusion
criteria, and based on the calculation of a pooled effect size rather than
head to head comparisons, has drawn similiar conclusions to those of Chambers
2006.The updated findings from the current review provide robust high quality
evidence to support those reported by Chambers 2006 and Malouff 2008 for the
effectiveness of CBT interventions for CFS, and in addition, is able to
demonstrate the homogeneity of the evidence.

--------------------------------------------------------------------------------
AUTHORS' CONCLUSIONS

Implications for practice

CBT is more effective than usual care for reducing fatigue symptoms in adults
with CFS, with 40% of participants assigned to CBT showing clinical response
at post-treatment, in comparison with 26% assigned to usual care control. The
benefits of CBT in sustaining clinical response and reduction of fatigue
symptoms at short and medium term follow-up are inconclusive. The benefits of
CBT in improving physical functioning and reducing depression, anxiety and
psychological distress at post treatment and at follow-up are also uncertain.

CBT may be more effective in reducing fatigue symptoms at post treatment than
psychological therapies using relaxation techniques or a support and
education approach, and may also be more effective in improving physical
functioning and reducing depression, anxiety and psychological distress at
post-treatment. However these findings are based on a small body of evidence
in which other therapies were designed as attention placebo controls, which
limits confidence in the findings obtained. The body of evidence for CBT
compared with relaxation or support and education at short and medium term
follow-up is very small, heterogeneous and inconclusive.

CBT may be an acceptable form of treatment for CFS, although this evidence is
currently based on a small number of individual studies. Whilst the overall
attrition rate from CBT interventions is reasonably low, at 16%, reasons for
dropout were under-reported in studies, and may not only have been due to low
acceptability or ineffectiveness of CBT.

Currently there is a lack of available evidence on the effectiveness of CBT
as a stand-alone intervention or in combination with other interventions
compared with usual care or other types of treatment (including immunological
therapies, pharmacological therapies, exercise, complementary/alternative
therapies and nutritional supplements) for CFS.


Implications for research

The body of evidence for the effectiveness of CBT for CFS has grown
considerably in quantity and methodological quality over the past eight
years. The studies included in this review demonstrated some methodological
strengths, including the increasing use of appropriate randomisation
procedures, attempts to control for therapist attention and time, and
follow-up assessments conducted 6-12 months after the completion of
treatment. Further specific recommendations to enhance the internal validity
of future studies include the recruitment of larger appropriately powered
samples, the use of treatment as usual control conditions rather than waiting
list, systematic assessment of psychological therapy fidelity to
manuals/protocols, and further use of blinded outcome assessors. To increase
the applicability of the findings to CFS individuals, clinicians and policy
makers, future studies should include adverse effects, acceptablity and
quality of life measures, together with cost-effectiveness, health service
resource use and occupational outcomes.

Whilst the evidence base for CBT as a stand-alone treatment for CFS continues
to expand, there is a surprising lack of high quality evidence on the
effectiveness of CBT alone or in combination with other treatments to inform
the development of clinical management programmes for people with CFS. Given
the complex and chronic clinical presentation of CFS, coupled with the
evidence that CBT alone is effective for no more than 40-48% of patients at
the end of treatment, trials examining CBT in combination with other
interventions to increase treatment effectiveness should be prioritised.

To date trials have been largely conducted in secondary or primary care
settings, recruiting patients who are sufficiently mobile to attend
out-patient clinics for CBT sessions. Studies based in domicilary or
in-patient settings should be considered in order to establish the extent to
which CBT is effective for those who are severely disabled with CFS.

Whilst the review provides some very preliminary findings for the
effectiveness of CBT using an individual modality and using increased
activity, further study of these aspects of the CBT interventions are
required in order to be able to draw valid conclusions on their superior
benefit.

There are a number of trials currently in progress (Bleijenberg 2008; Carney
2008; Gibson-Saxty 2002; Vissers 2008; Wearden 2006; White 2005), which will
help to strengthen the evidence base on CBT interventions for CFS. The
findings of these studies are awaited with interest.

--------------------------------------------------------------------------------
ACKNOWLEDGEMENTS

We would like to aknowledge Dr J Couper, St Vincent's Hospital, University of
Melbourne, Australia, for his contribution on the original version of the
review.

We are most grateful to the following authors for kindly providing us with
subgroup data from their trials:

Dr Marcus Huibers, Department of Epidemiology, Maastricht University, The
Netherlands

Dr Judith Prins, University Medical Centre Nijmegen, The Netherlands

Dr Leone Ridsdale, Academic Neuroscience Centre, Institute of Psychiatry,
London, UK


We also acknowledge the following authors for responding to our request for
further information:

Professor Trudie Chalder, Academic Department of Psychological Medicine,
Institute of Psychiatry, London, UK

Dr Vanessa Russell, Adult Psychological Services (East Riding), Willerby, UK

Professor Andrew Lloyd, University of New South Wales, Sydney, Australia

Mr Paul Seed, Divison of Reproduction and Endocrinology, King's College
London

Professor Peter Campion, Postgraduate Medical Institute, University of Hull

--------------------------------------------------------------------------------
FIGURE CAPTIONS

Figure 1. Risk of bias graph: review authors' judgement about each risk of
          bias item presented as percentages across all included studies
Figure 2. Risk of bias summary: review authors' judgements about each risk of
          bias item for each included study

--------------------------------------------------------------------------------
REFERENCES

References to studies included in this review

Barrett 1992 {unpublished data only}
   Barrett E. Cognitive-behaviour therapy in the management of
   chronic fatigue syndrome. M Psych University of Sydney 1992.

Deale 1996 {published data only}
   Deale A, Chalder T, Marks I, Wessely S. Cognitive behavior therapy
   for chronic fatigue syndrome: A randomized controlled trial. Ameri-
   can Journal of Psychiatry 1997;154:408-14.

 * Deale A, Chalder T, Marks I, Wessely S. Cognitive behaviour ther-
   apy for chronic fatigue syndrom: A randomised controlled trial. South 
   Thames Research and Development, Open Day Feb 28. 1996.

   Deale A, Chalder T, Wessely S. Illness beliefs and treatment outcome
   in chronic fatigue syndrome. Journal of Psychosomatic Research 1998;
   45:77-83.

   Deale A, Husain K, Chalder T, Wessely S. Long-term outcome of
   cognitive behavior therapy versus relaxation therapy for chronic fa-
   tigue syndrome: a 5-year follow-up study. American Journal of Psy-
   chiatry 2001;158(12):2038-42.

   Peakman M, Deale A, Field R, Mahalingam M, Wessely S. Clini-
   cal improvement in chronic fatigue syndrome is not associated with
   lymphocyte subsets of function or activation. Clinical Immunology
   and Immunopathology 1997;82(1):83-91.

   Quarmby L, Rimes KA, Deale A, Wessely S, Chalder T. Cognitive-
   behaviour therapy for chronic fatigue syndrome: Comparison of out-
   comes within and outside the confines of a randomised cotrolled trial.
   Behaviour Research and Therapy 2007;45(6):1085-94.

   Wessely S. Cognitive behaviour therapy versus relaxation for Chronic
   Fatigue Syndrome: OUtcome at five year follow-up. National Re-
   search Register 1999.

Huibers 2004 {published data only}
   Huibers MJ, Beurskens AJ, Van Schayck CP, Baselmans E, Metsemak-
   ers JF, Knottnerus JA, et al.Efficacy of cognitive-behavioural therapy
   by general practitioners for unexplained fatigue among employees:
   Randomised controlled trial. British Journal of Psychiatry 2004;184:
   240-6.

Jason 2007 {published data only (unpublished sought but not used)}
   Jason LA, Torres-Harding S, Friedberg F, Corradi K, Njoku MG,
   Donalek J, et al.Non-pharmacologic interventions for CFS: A ran-
   domized trial. Journal of Clinical Psychology in Medical Settings 2007;
   14:275-96.

King 1999 {published data only}
   Chalder T, Godfrey E, Ridsdale L, King M, Wessely S. Predictors
   of outcome in a fatigued population in primary care following a
   randomized controlled trial. Psychological Medicine 2003;33(2):283-7.

   Chisholm D, Godfrey E, Ridsdale L, Chalder T, King M, Seed P,
   Wallace P, Wessely S. Chronic fatigue in general practice: Economic
   evaluation of counselling versus cognitive behaviour therapy. British
   Journal of General Practice 2001;51(462):15-8.

   Godfrey E. Aspects of the therapeutic alliance associated with a good
   outcome: An analysis of the process of therapy taken from a ran-
   domised controlled trial of CBT versus counselling for patients with
   chronic fatigue in primary care. 28th Annual Conference of the
   British Association for Behavioural and Cognitive Psychotherapies.
   London, 2000; Vol. July 19-22.

   Godfrey E, Chalder T, Ridsdale L, Seed P, Ogden J. Does the ther-
   apeutic alliance affect outcome in the context of an RCT of CBT
   versus counselling for chronic fatigue in primary care?. 34th Annual
   Conference of the British Association for Behavioural and Cognitive
   Psychotherapies. Warwick, 2006; Vol. July 19­21:81-2.

   Godfrey E, Chalder T, Ridsdale L, Seed P, Ogden J. Investigating
   the active ingredients of cognitive behaviour therapy and counselling
   for patients with chronic fatigue in primary care: developing a new
   process measure to assess treatment fidelity and predict outcome.
   British Journal of Clinical Psychology 2007;46(3):253-72.

 * King MB. Randomised controlled trial of counselling or CBT for
   patients with chronic fatigue in general practitioners. National Re-
   search Register 1999.

   Ridsdale L, Godfrey E, Chalder T, Seed P, King M, Wallace P, Wes-
   sely S. Chronic fatigue in general practice: Is counselling as good as
   cognitive behaviour therapy?. British Journal of General Practice 2001;
   51(462):19-24.

   Ridsdale L, Godfrey E, Chalder T, Seed P, King M, Wallace P, Wes-
   sely S and Fatigue Trialists' Group. Chronic fatigue in general prac-
   tice: is counselling as good as cognitive behaviour therapy? A UK
   randomised trial. British Journal of General Practice 2001;51:19-24.

Lloyd 1993 {published data only}
   Lloyd AR, Hickie I, Brockman A, Hickie C, Wilson A, Dwyer J, et
   al.Immunologic and psychologic therapy for patients with chronic
   fatigue syndrome: a double-blind, placebo-controlled trial. American
   Journal of Medicine 1993;94:197-203.

O'Dowd 2000 {published data only}
 * O'Dowd H. Cognitive behavioural therapy in chronic fatigue syn-
   drome (CFS): A randomised controlled trial of an outpatient group
   programme. Current Controlled Trials 2000.

   O'Dowd H, Gladwell P, Rogers C, Hollinghurst S, Gregory A. Cogni-
   tive behavioural therapy in chronic fatigue syndrome: A randomised
   controlled trial of an outpatient group programme. 33rd Annual
   Conference of the British Association for Behavioural and Cognitive
   Psychotherapies. Canterbury, 2005; Vol. July 21­23:78-9.

   O'Dowd H, Gladwell P, Rogers CA, Hollinghurst S, Gregory A.
   Cognitive behavioural therapy in chronic fatigue syndrome: a ran-
   domised controlled trial of an outpatient group programme. Health
   Technology Assessment 2006;10(37):iii-iv, ix-x, 1-121.

Prins 2001 {published data only}
   Bazelmans E, Prins JB, Hoogveld S, Bleijenberg G. Manual-based
   cognitive behaviour therapy for chronic fatigue syndrome: therapists'
   adherence and perceptions. Cognitive Behaviour Therapy 2004;33(3):
   143-50.

   Knoop JA, Prins JB, Bleijenberg G. Is CBT an effective treatment for
   pain symptoms in chronic fatigue syndrome?. 32nd Congress of the
   British Association for Behavioural and Cognitive Psychotherapies
   (jointly with European Association of Behavioural and Cognitive
   Therapies). Manchester, 2004 September 7­11:212-3.

   Prins J, Bleijenberg G, Royweler EK, van der Meer J. Effect of psy-
   chiatric disorders on outcome of cognitive-behavioural therapy for
   chronic fatigue syndrome. British Journal of Psychiatry 2005;187:
   184-5.

   Prins JB, Bleijenberg G, Bazelmans E, Elving LD, de Boo TM, Sev-
   erens JL, et al.Cognitive behaviour therapy for chronic fatigue syn-
   drome: a multicentre randomised controlled trial. Lancet 2001;357:
   841-7.

   Prins JB, Bos E, Huibers MJ, Servaes P, van der Werff SP, van der
   Meer JW, et al.Social support and the persistence of complaints in
   chronic fatigue syndrome. Psychotherapy and Psychosomatics 2004;73
   (3):174-82.

   Severens JL, Prins JB, van der Wilt GJ, van der Meer JW, Bleijenberg
   G. Cost-effectiveness of cognitive behaviour therapy for patients with
   chronic fatigue syndrome. QJM 2004;97(3):153-6.

Ridsdale 2004 {published and unpublished data}
   Ridsdale L, Darbishire L, Seed PT. Is graded exercise better than
   cognitive behaviour therapy for fatigue? A UK randomized trial in
   primary care. Psychological Medicine 2004;34:37-49.

Russell 2001 {unpublished data only}
   Russell V, Gaston AM, Lewin RJP, Atkinson CM, Campion PD.
   Group rehabilitation for adult chronic fatigue syndrome. unpub-
   lished article.

Sharpe 1993 {published and unpublished data}
 * Sharpe M. Non-pharmacological approaches to treatment. Chronic
   Fatigue Syndrome: Ciba Foundation Symposium. Vol. 173, 298-308,
   Chichester: John Wiley & Sons, 1993.

   Sharpe M, Hawton K, Peto T. Cognitive behaviour therapy for the
   chronic fatigue syndrome. BMJ 1998;312:1098.

   Sharpe M, Hawton K, Simkin S, Surawy C, Hackmann A, Klimes I,
   Peto T, Warrell D, Seagroatt S. Cognitive behavior therapy for chronic
   fatigue syndrome: a randomized controlled study. Verhaltenstherapie
   1998;8(2):118-24.

   Sharpe M, Hawton K, Simkin S, Surawy C, Hackmann A, Klimes I,
   et al.Cognitive behaviour therapy for the chronic fatigue syndrome: a
   randomized controlled trial. British Medical Journal 1996;312:22-6.

Stevens 1999 {unpublished data only}
   Stevens M, Gevirtz R, Verity L, Wiederhold M. Chronic fatigue syn-
   drome: a chronobiologically oriented controlled treatment outcome
   study. unpublished article 1999.

Strang 2002 {unpublished data only}
   Strang JM. Treatment of chronic fatigue syndrom: a cognitive-behav-
   ioral approach to enhance personal mastery. Dissertation: Arizona
   State University August 2002.

Surawy 2005 {published data only}
   Surawy C, Roberts J, Silver A. The effect of mindfulnes training on
   mood and measures of fatigue, activity, and quality of life in patients
   with chronic fatigue syndrome on a hospital waiting list: a series of
   exploratory studies. Behavioural and Cognitive Psychotherapy 2005;
   33:103-9.

Whitehead 2002 {published data only}
   Whitehead L, Campion P. Can general practitioners manage chronic
   fatigue syndrome. Journal of Chronic Fatigue Syndrome 2002;10(1):
   55-64.


References to studies excluded from this review

Allen 2006 {published data only}
   Allen L, Woolfolk R, Escobar J, Gara M, Hamer R. Cognitive-behav-
   ioral therapy for somatization disorder. Archives of Internal Medicine 
   2006;166:1512-8.

Bleichhardt 2004 {published data only}
   Bleichhardt G, Timmer B, Rief W. Cognitive-behavioural therapy for
   patients with multiple somatoform symptoms--a randomised con-
   trolled trial in tertiary care. Journal of Psychosomatic Research 2004;
   56:449-54.

Burgess 2001 {published data only}
   Burgess M, Chalder T. Telephone cognitive behaviour therapy for
   chronic fatigue syndrome in secondary care. Behavioural and Cogni-
   tive Psychotherapy 2001;29(4):447-55.

Butler 1991 {published data only}
   Bonner D, Ron M, Chalder T, Butler S, Wessely S. Chronic fatigue
   syndrome: a follow up study. Journal of Neurology, Neurosurgery and
   Psychiatry 1994;57:617-21.

   Butler S, Chalder T, Ron M, Wessely S. Cognitive behaviour therapy
   in chronic fatigue syndrome. Journal of Neurology, Neurosurgery and
   Psychiatry 1991;54:153-8.

Candy 2004 {published data only}
   Candy, B. Chalder, T. Cleare, A. Wessely, S. Hotopf, M. A ran-
   domised controlled trial of a psycho-educational intervention to aid
   recovery in infectious mononucleosis.. Journal of Psychosomatic Re-
   search 2004;57:89-94.

Donta 2003 {published data only}
   Donta ST, Clauw DJ, Engel CC Jr,. Guarino P. Peduzzi P, Williams
   DA, et al.Cognitive behavioral therapy and aerobic exercise for Gulf
   War veterans' illnesses: a randomized controlled trial. JAMA 2003;
   289(11):1396-404.

Friedberg 1994 {published data only}
   Friedberg F, Krupp LB. A comparison of cognitive behavioral treat-
   ment for chronic fatigue syndrome and primary depression. Clinical
   Infectious Diseases 1994;18(Suppl. Clinical Infectious Diseases 1994;
   18(Suppl 1):105-10.

Gielissen 2006 {published data only}
   Gielissen MF, Verhagen S, Witjes F, Bleijenberg G. Effects of cogni-
   tive behavior therapy in severely fatigued disease-free cancer patients
   compared with patients waiting for cognitive behavior therapy: a ran-
   domized controlled trial. J Clin Oncol 2006;24(30):4882-7.

Goudsmit 1996 {unpublished data only}
   Goudsmit EM. The psychological aspects and management of
   chronic fatigue syndrome. Unpublished PhD Thesis, Brunel Univer-
   sity 1996.

Hatcher 1998 {published data only}
   Hatcher S. A randomised double-blind placebo controlled trial of
   dothiepin and graded activity in the treatment of chronic fatigue
   syndrome. personal communication 1998.

Lyles 2003 {published data only}
   Lyles JS, Hodges A, Collins C, Lein C, Given CW, Given B, et
   al.Using nurse practitioners to implement an intervention in primary
   care for high-utilizing patients with medically unexplained symp-
   toms. General Hospital Psychiatry 2003;25(2):67-73.

Soderberg 2001 {published data only}
   Soderberg S, Evengard, B. Short-term group therapy for patients with
   chronic fatigue syndrome. Psychotherapy and Psychosomatics 2001;70:
   108-11.

Taylor 2004 {published data only}
   Taylor R. Quality of life and symptom severity for individuals with
   chronic fatigue syndrome: Findings from a randomized clinical trial.
   The American Journal of Occupational Therapy 2004;58:35-43.

Zhang 1989 {published data only}
   Zhang X. Short-term psycotherapy of 107 neurasthenia cases. Acta
   Academiae Medicinae Hebie 1989;10(4):349-51.


References to studies awaiting assessment

Chalder 1997 {published and unpublished data}
   Chalder T, Wallace P, Wessely S. Self-help treatment of chronic fa-
   tigue in the community: A randomized controlled trial. British Jour-
   nal of Health Psychology 1997;2:189-97.

Powell 2001 {published data only}
   Powell P, Bentall RP, Nye FJ, Edwards HT. Radomised controlled
   trial of patient education to encourage graded exercise in chronic
   fatigue syndrome. BMJ 2001;322:1-6.


References to ongoing studies

Bleijenberg 2008 {unpublished data only}
 * Bleijenberg G. The effectiveness of cognitive behavioural therapy
   in groups for patients with chronic fatigue syndrome (CFS): a ran-
   domised controlled study. controlled-trials.com 2008.

Carney 2008 {unpublished data only}
 * Carney C, Jones-Alexander J. Behavioral insomnia therapy with
   chronic fatigue syndrome. clinicaltrials.gov 2008.

Gibson-Saxty 2002 {unpublished data only}
 * Gibson-Saxty J. Group therapy for chronic fatigue syndrome. Na-
   tional Research Register 2002.

Vissers 2008 {unpublished data only}
 * Vissers H. The efficacy and predicting variables of a multidisci-
   plinary disability resolution (MDR) program for CFS patients re-
   ceiving long term disability benefits from income protection insurers.
   controlled-trials.com 2008.

Wearden 2006 {unpublished data only}
   Wearden AJ, Riste L, Dowrick C, Chew-Graham C, Bentall RP, Mor-
   riss RK, Peters S, Dunn G, Richardson G, Lovell K, Powell P. Fatigue
   Intervention by Nurses Evaluation - The FINE Trial. A randomised
   controlled trial of nurse led self-help treatment for patients in primary
   care with chronic fatigue syndrome: study protocol. BMC Medicine
   2006;4:9.

White 2005 {unpublished data only}
 * White PD, Sharpe MC, Chalder T, DeCesare JC, Walwyn R. Pro-
   tocol for the PACE trial: A randomised controlled trial of adaptive
   pacing, cognitive behaviour therapy and graded exercise as supple-
   ments to standardised specialist medical care versus standardised spe-
   cialist medical care alone for patients with the chronic fatigue syn-
   drome/myalgic encephalomyelitis or encephalopathy. BMC Neurol-
   ogy 2007;7:6.


Additional references

Adams 2007
   Adams D, Wu T, Tai S, Wiebe N, Vohra S. Traditional Chinese
   medicinal herbs for idiopathic chronic fatigue and chronic fatigue
   syndrome. Cochrane Database of Systematic Reviews 2007, Issue 2.
                                                                             
Afari 2003
   Afari N, Buchwald D. Chronic fatigue syndrome: a review. American
   Journal of Psychiatry 160;2:221-36.

Beck 1961
   Beck AT, Ward CH, Mendelson M, Erbaugh J. An inventory for
   measuring depression. Archives of General Psychiatry 1961;4:561-71.
                                                                             
Beck 1979
   Beck AT, Rush AJ, Shaw BF, Emery G. Cognitive Therapy of Depres-
   sion. New York, NY: Guildford Press, 1979.

Beck 1987
   Beck AT, Steer R. Beck depression inventory: manual. San Antonio,
   TX: Psychological Corporation, 1987.

Beck 1988
   Beck AT, Epstein N, Brown G, Steer RA. An inventory for measuring
   clinical anxiety: Psychometric properties. Journal of Consulting and
   Clinical Psychology 1988;56:893-7.

Behan 1995
   Behan PO, Hannifah H. 5-HT reuptake inhibitors in CFS. EOS
   Rivista di Immunologia ed Immunofarmacologia 1995;15 (1-2):66-9.

Borkovec 2001
   Borkovec TD, Ruscio AM. Psychotherapy for generalized anxiety
   disorder. Journal of Clinical Psychiatry 2001;62(Suppl 11):15-19.

Carruthers 2003
   Carruthers BM, Jain AK, De Meirleir KL, Peterson DL, Klimas NG,
   et al.Myalgic encephalomyelitis/chronic fatigue syndrome: clinical
   working case definition, diagnostic and treatment protocols. Journal
   of Chronic Fatigue Syndrome 2003;11(1):7-97.

CFS/ME WG 2002
   CFS/ME Working Group. Report of the CFS/ME Working Group to
   the Chief Medical Officer. London: Department of Health, 2002.

Chalder 1993
   Chalder T. Development of a fatigue scale. Journal of Psychosomatic
   Research 1993;37:147-53.

Chambers 2006
   Chambers D, Bagnall AM, Hempel S, Forbes C. Interventions for the
   treatment, management and rehabilitation of patients with chronic
   fatigue syndrome/myalgic encephalomyelitis: an updated systematic
   review. Journal of the Royal Society of Medicine 2006;99:506-520.

Cho 2005
   Cho, JH, Hotopf M, Wessely S. The placebo response in the treat-
   ment of chronic fatigue syndrom: a systematic review and meta-anal-
   ysis. Psychosomatic Medicine 2005;87:301-313.

Clark 2005
   Clark LV, White PD. The role of deconditioning and therapeutic
   exercise in chronic fatigue syndrome (CFS). Journal of Mental Health
   2005;14(3):237-52.

Cox 1987
   Cox B, Blaxter M, Buckle A, et al.The health and lifestyle survey.
   London: Health Promotion Research Trust, 1987.

Deeks 1997
   Deeks J. Are you sure that's a standard deviation? (part 1). Cochrane
   News 1997;10:11-2.

Edmonds 2004
   Edmonds M, McGuire H, Price J. Exercise therapy for chronic fatigue
   syndrome. Cochrane Database of Systematic Reviews 2004, Issue 3.

Fukuda 1994
   Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff
   A. The chronic fatigue syndrome: A comprehensive approach to its
   definition and study. Annals of Internal Medicine 1994;121(12):953-9.

Fulcher 1997
   Fulcher KY, White PD. Randomised controlled trial of graded exer-
   cise in patients with the chronic fatigue syndrome. BMJ 1997;314
   (7095):1647-52.

Furukawa 2005
   Furukawa TA, Cipriani A, Barbui C, Brambilla P, Watanabe N. Im-
   puting response rates from means and standard deviations in meta-
   analyses. International Clinical Psychopharmacology 2005;1:49-52.

Gilbert 2005
   Gilbert P (Ed). Compassion: Conceptualisations, Research and Use in
   Psychotherapy. Routledge, 2005.

Hamilton 1960
   Hamilton M. A rating scale for depression. Journal of Neurology Neu-
   rosurgery and Psychiatry 1960;23:56-62.

Hayes 2003
   Hayes, Steven C, Kirk D. Strosahl, Kelly G. Acceptance and Com-
   mitment Therapy : An Experiential Approach to Behavior Change. The
   Guilford Press, 2003.

Higgins 2003
   Higgins JT, Thompson SG, Deeks JJ, Altman DG. Measuring in-
   consistency in meta-analyses. BMJ 2003;327(6):557-60.

Kabat-Zinn 1991
   Kabat-Zinn J. Full catastrophe living: The programme of the stress re-
   duction clinic at the Unviersity of Massachusetts Medical Centre. New
   York: Delta, 1991.

Karnofsky 1948
   Karnofsky D, Abelmann W, Craver L, Burchenal J. The use of nitro-
   gen mustards in the palliative treatment of carcinoma. Cancer 1948;
   1:634-56.

Kaslow 1989
   Kaslow JE, Rucker L, Onishi R. Liver extract-folic acid-cyanocobal-
   amin vs placebo for chronic fatigue syndrome. Archives of Internal
   Medicine 1989;149(11):2501-3.

Law 2003
   Law J, Garrett Z, Nye C. Speech and language therapy interven-
   tions for children with primary speech and language delay or disor-
   der. Cochrane Database of Systematic Reviews 2003, Issue 3.

Lloyd 1988
   Lloyd A, Wakefield D, Boughton C, Dwyer J. What is myalgic en-
   cephalomyelitis?. Lancet 1988;1:1286-7.

Malouff 2008
   Malouff JM, Thorsteinsson EB, Rooke SE, Bhullar N, Schutte NS.
   Efficacy of cognitive behavioral therapy for chronic fatigue syndrome:
   A meta-analysis. Clinical Psychology Review 2008;28:736-745.

McNair 1992
   McNair DM, Lorr M, Dropppleman LF. Profile of Mood States Man-
   ual. San Diego: Educational and Industrial Testing Service, 1992.

Mulrow 2001
   Mulrow CD, Ramirez G, Cornell JE, Allsup K. Defining and manag-
   ing chronic fatigue syndrome. Evidence Report/Technology Assessment
   No 42. San Antonio Evidence-based Practice Center, University of
   Texas, 2001.

NICE 2007
   Chronic fatigue syndrome/myalgic encephatomyelitis (or encephalopathy):
   diagnosis and management of CFS/ME in adults and children. National
   Institute of Health and Clinical Excellence (2007) Clinical Guideline 53.

Nixon 1995
   Nixon PGF. Hyperventilation and Chronic Fatigue Syndrome.
   Queen's Journal of Medicine 1995;88:73-74.

RACPWG 2002
   Royal Australasian College of Physicians Working Group. Chronic
   fatigue syndrome: Clinical practice guidelines. Medical Journal of
   Australia 2002;176:S17-S55.

Rawson 2007
   Rawson KM, Rickards H, Haque S, Ward C. Pharmacological treat-
   ments for chronic fatigue syndrome. Cochrane Database of Systematic
   Reviews 2007, Issue 4.

Reich 1981
   Reich JW, Zautra AJ. Life events and persona causation: some re-
   lationships with satisfaction and distress. Journal of Personality and
   Social Psychology 1981;41(5):1002-12.

Ridsdale 1989
   Ridsdale L. Chronic fatigue in family practice. Journal of Family Prac-
   tice 1989;29:486-8.

Rimes 2005
   Rimes K, Chalder T. Treatments for chronic fatigue syndrome. Oc-
   cupational Medicine 2005;55:32-9.

Segal 2002
   Segal Z, Williams JMG, Teasdale JD. Mindfulness based cognitive
   therapy for depression: A new approach for preventing relapse. New York:
   Guildford Press, 2002.

Sharpe 1991
   Sharpe M, Archard L, Banatvala J, et al.Chronic fatigue syndrome:
   guidelines for research. Journal of the Royal Society of Medicine 1991;
   84:118-21.

Spielberger 1983
   Spielberger CD, Gorsuch RL, Lushene R, Vagg PR, Jacobs GA. Man-
   ual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting
   Psychologists Press, 1983.

Vercoulen 1996
   Vercoulen JH, Swanink CM, Zitman FG, Vreden SG, Hoofs MP,
   Fennis JF. Randomised, double-blind, placebo-controlled study of
   fluoxetine in chronic fatigue syndrome. Lancet 1996;347(9005):
   858-61.

Vercoulen 1999
   Vercoulen JH, Alberts M, Bleijenberg G. The Checklist Individual
   Strength (CIS). Gedragstherapie 1999;32:31-6.

Ware 1993
   Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey
   Manual and Interpretation Guide. Boston, MA: New England Med-
   ical Centre, 1993.

Warren 1999
   Warren G, McKendrick M, Peet M. The role of essential fatty acids in
   chronic fatigue syndrome. A case-controlled study of red-cell mem-
   brane essential fatty acids (efa) and a placebo-controlled treatment
   study with high dose of efa. Acta Neurologica Scandinavica 1999;99
   (2):112-6.

Wearden 1998
   Wearden AJ, Morriss RK, Mullis R, Strickland PL, Pearson DJ, Ap-
   pleby L. Randomised, double-blind, placebo-controlled trial of flu-
   oxetine and graded exercise for chronic fatigue syndrome. British
   Journal of Psychiatry 1998;172:485-90.

Wessely 1995
   Wessely S. The epidemiology of chronic fatigue syndrome. Epidemi-
   ologic Reviews 1995;17:139-51.

Zhang 2006
   Zhang W, Liu ZS, Wu Taixiang, Peng WN. Acupuncture for chronic
   fatigue syndrome. Cochrane Database of Systematic Reviews 2006, Is-
   sue Issue 2.

Zigmond 1983
   Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale.
   Acta Psychiatrica Scandinavica 1983;67:361-70.

* Indicates the major publication for the study

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Next Chapters - not included here
   CHARACTERISTICS OF STUDIES
   DATA AND ANALYSES
   WHAT'S NEW
   HISTORY
   CONTRIBUTIONS OF AUTHORS
   DECLARATIONS OF INTEREST

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
(c) 2008 The Cochrane Collaboration
(c) 2008 John Wiley & Sons, Ltd.