Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Cognitive behavioural therapies for fibromyalgia

Authors' declarations of interest

Version published: 10 September 2013 Version history

https://doi.org/10.1002/14651858.CD009796.pub2
Collapse all Expand all

Background

Fibromyalgia (FM) is a clinically well‐defined chronic condition of unknown aetiology characterized by chronic widespread pain that often co‐exists with sleep disturbances, cognitive dysfunction and fatigue. Patients often report high disability levels and negative mood. Psychotherapies focus on reducing key symptoms, improving daily functioning, mood and sense of personal control over pain.

Objectives

To assess the benefits and harms of cognitive behavioural therapies (CBTs) for treating FM at end of treatment and at long‐term (at least six months) follow‐up.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2013, Issue 8), MEDLINE (1966 to 28 August 2013), PsycINFO (1966 to 28 August 2013) and SCOPUS (1980 to 28 August 2013). We searched http://www.clinicaltrials.gov (web site of the US National Institutes of Health) and the World Health Organization Clinical Trials Registry Platform (ICTRP) (http://www.who.int/ictrp/en/) for ongoing trials (last search 28 August,2013), and the reference lists of reviewed articles.

Selection criteria

We selected randomised controlled trials of CBTs with children, adolescents and adults diagnosed with FM.

Data collection and analysis

The data of all included studies were extracted and the risks of bias of the studies were assessed independently by two review authors. Discrepancies were resolved by discussion.

Main results

Twenty‐three studies with 24 study arms with CBTs were included. A total of 2031 patients were included; 1073 patients in CBT groups and 958 patients in control groups. Only two studies were without any risk of bias. The GRADE quality of evidence of the studies was low. CBTs were superior to controls in reducing pain at end of treatment by 0.5 points on a scale of 0 to 10 (standardised mean difference (SMD) ‐ 0.29; 95% confidence interval (CI) ‐0.49 to ‐0.17) and by 0.6 points at long‐term follow‐up (median 6 months) (SMD ‐0.40; 95% CI ‐0.62 to ‐0.17); in reducing negative mood at end of treatment by 0.7 points on a scale of 0 to 10 (SMD ‐ 0.33; 95% CI ‐0.49 to ‐0.17) and by 1.3 points at long‐term follow‐up (median 6 months) (SMD ‐0.43; 95% CI ‐0.75 to ‐0.11); and in reducing disability at end of treatment by 0.7 points on a scale of 0 to 10 (SMD ‐ 0.30; 95% CI ‐0.51 to ‐0.08) and at long‐term follow‐up (median 6 months) by 1.2 points (SMD ‐0.52; 95% CI ‐0.86 to ‐0.18). There was no statistically significant difference in dropout rates for any reasons between CBTs and controls (risk ratio (RR) 0.94; 95% CI 0.65 to 1.35).

Authors' conclusions

CBTs provided a small incremental benefit over control interventions in reducing pain, negative mood and disability at the end of treatment and at long‐term follow‐up. The dropout rates due to any reason did not differ between CBTs and controls.

Plain language summary

available in

Cognitive behavioural therapies for fibromyalgia syndrome

Researchers in The Cochrane Collaboration conducted a review of research about the effects of cognitive‐behavioural therapies (CBTs) on fibromyalgia (FM). After searching for all relevant studies, they found 23 studies with up to 2031 people. Their findings are summarised below.

After about 12 weeks, children, adolescents and adults with FMS, who used CBTs compared to controls, were likely to report that CBT

‐ may reduce slightly pain, negative mood and disability at the end of the treatment;

‐ may reduce slightly pain, negative mood and disability six months after the end of treatment.

There was no difference between CBTs and controls in the number of people who withdrew from treatment.

We do not have precise information about side effects and complications of CBTs. Rare complications may include worsening of co‐existing mental disorders.

What is fibromyalgia and what are cognitive behavioural therapies?

People with FM suffer from chronic widespread pain, sleep problems and fatigue. There is no cure for FM at present, so treatments aim to relieve symptoms and to improve daily functioning.

Cognitive behavioural therapies (CBTs) are widely used psychological treatments for a wide range of health problems, including chronic pain. CBTs are effective in enhancing patients’ beliefs in their own abilities and developing ways to deal with health problems. The primary goals of CBTs are to change negative thoughts and feelings that individuals may have of their physical and mental problems and to change their behaviour accordingly. Patients learn skills (for example, relaxation, activity pacing) to help them to manage their pain better or develop different attitudes towards pain (for example, more acceptance), or both.

Best estimates of what happens to people with FMS when they use CBTs

Pain (higher scores mean worse or more severe pain):

‐ People who used CBTs rated their pain to be 0.5 points lower at the end of treatment (6.3% absolute improvement) and to be 0.6 points lower six months after the end of treatment on a scale of 0 to 10 (4.2% absolute improvement).

‐ People who used CBTs rates their pain to be 6.9 points on a scale of 0‐10.

‐ People who used a control treatment rated their pain to be 7.4 points on a scale of 0 to 10.

Negative mood (higher scores mean worse or more severe negative mood):

‐ People who used CBTs rated their depressed mood to be 0.7 points lower at the end of treatment (10.2% absolute improvement) and to be 1.3 points lower six months after the end of treatment on a scale of 0 to 10 (2.7% absolute improvement).

‐ People who used CBTs rated their negative mood to be 6.1 points on a scale of 0 to 10.

‐ People who had a control treatment rated their negative mood to be 6.8 points on a scale of 0 to 10.

Disability (higher scores mean more disability):

‐ People who used CBTs rated their disability to be 0.7 points lower at the end of treatment (7.2% absolute improvement) and to be 1.2 points lower six months after the end of treatment on a scale of 0 to 10 (11.7% absolute improvement).

‐ Peope who used CBTs rated their disability to be 2.0 points on a scale of 0 to 10.

‐ People who used a control treatment rated their disability to be 2.8 points on a scale of 0 to 10.

Withdrawing from treatment:

‐ The number of people who withdrew from CBTs compared to control interventions due to any reason was equal.

‐ 15 people out of 100 who used CBTs withdrew from treatment due to any reason;

‐ 15 people out of 100 who used control interventions withdrew from treatment due to any reason.

Authors' conclusions

Implications for practice

Traditional cognitive behavioural therapy and operant therapy reduce the key symptoms of FM after six months. These types of therapies should be offered face‐to face. The limited data available on internet‐ or telephone‐based CBTs do not allow a recommendation to use this type of delivery. Treatment intensity should be between five and 25 hours. Evidence was not found for benefit of self management programs as single therapy.

Even if the assessment of adverse events in most CBTs studies was insufficient and two studies reported a dropout rate of up to 5% of patients in CBTs groups due to worsening of co‐morbid mental disorders, the risk‐benefit ratio of CBTs is more favourable than that of drugs which have been approved by the US Food and Drug Administration (FDA) for FM treatment. Adverse events and dropouts due to adverse events were higher in drug than in CBTs trials (see Häuser 2010; Sommer 2012b). In addition, the positive effects of CBTs, especially of operant therapy, increase within six months after the end of therapy, whereas the positive effects of drugs disappear within two weeks after the cessation of therapy (Saxe 2012).

In some countries (for example USA), CBTs might not be easily available for patients and might be expensive and not covered by insurance (Rheumatologist Los Angeles 2012). In contrast, aerobic exercise (for example walking) is available everywhere with very low costs because it can be carried out by the patient himself without professional guidance, or after a short education. We found that CBTs were not superior to other active treatments such as aerobic exercise in the reduction of FM symptoms. However, aerobic exercise of low to moderate intensity is effective in reducing FM key symptoms (pain, fatigue, negative mood) at end of treatment and long‐term follow‐up (Busch 2007; Winkelmann 2012), to the same amount as CBTs (Nüesch 2013). Therefore, aerobic exercise and not CBTs are the first non‐pharmacologic treatment option of choice for FM (Eich 2012). This conclusion is not based on the results of head‐to‐head comparisons of CBTs versus aerobic exercise but on the higher availability and lower costs of aerobic exercise compared to CBTs.

Implications for research

The following research priorities would help to understand the effects of CBTs in FM according to EPICOT (Brown 2006), and to definitively convince patients, clinicians, researchers and payers of the benefits of CBTs in FM.

Evidence

No further small sample size studies with low to moderate methodology and treatment quality comparing CBTs with treatment as usual are needed. Large and high quality studies should compare CBTs with an adequate psychological placebo or best available drug therapy or aerobic exercise. Any new RCT needs to be designed and reported taking explicit account of the challenges identified and discussed in this review and in the ones of psychological therapies in chronic pain in children and adolescents (Eccleston 2012) and in adults (Williams 2012).

Population (external validity)

Studies are needed in all age groups. Future studies should include more adolescents, men and seniors. Clinical studies in FM should be conducted with patient samples representative of clinical practice, including patients with co‐morbid anxiety and depressive disorders and inflammatory rheumatic diseases. Subgroup analyses should be performed for these subpopulations. Moreover, studies should include patients with different disease durations and different severity levels all over the world.

Intervention (treatment quality)

Standardised empirical protocols should be used. Therapists providing interventions in trials should have been trained in the use of the protocol. Therapist adherence to the protocol (treatment integrity) and patient adherence to behavioural skill training (treatment uptake) should be monitored (Kashikar‐Zuck 2010). Attendance rates should be reported. Future intervention studies should be designed to assess dose–response curves for improvement of FM symptoms and evaluate follow‐up periods following completion of the active intervention period. These studies should include ongoing, intermittent assessment of CBTs interventions and FM outcome measures to understand the stability of responses and program characteristics (intensity, duration, frequency) needed to maintain gains.

Comparisons

An attention control with the same amount of time spent with the patients by therapists with the same experience in psychotherapy as in the treatment groups is the gold standard to control for unspecific effects of psychological therapies (Kashikar‐Zuck 2010; Wampold 2005). Of note, the superiority of CBTs over controls which we found in this review is based on the comparison of CBTs with treatment as usual and waiting list control, which were inadequate according to the gold standard. A recent large trial with mindfulness‐based stress reduction (MBSR) failed to demonstrate a superiority of MBSR over an adequate 'psychological placebo' (Schmidt 2011). Whether CBTs add a relevant benefit over an adequate psychological placebo still needs to be determined. The significance of CBTs compared to other established FM therapies such as aerobic exercise and drugs recommended for FM (amitriptyline, duloxetine, milnacipran, pregabalin) still needs to be determined.

Outcomes

A core set of outcome measures for psychological trials in FM is needed. The set should include the key domains of FM developed through consensus among experts and FM patients (Mease 2009) and the suggestions of the Initiative of Methods, Measurement and Pain Assessment in Clinical trials (IMMPACT) (Dworkin 2008; Dworkin 2009). Key domains include pain, sleep problems and fatigue. Responder outcomes (for example number of patients with ≥ 50% pain reduction, number of patients with ≥ 14% in the FIQ‐total score) should be reported. There is increasing recognition that psychological treatments have the potential for negative outcomes for some patients (Kashikar‐Zuck 2010). Standards, in which potential negative outcomes should be routinely monitored, by which methods, still have to be defined by the research community. In addition, a discussion of what we think is feasible as the outcome of any type of therapy in FM is appropriate. Average effect sizes across trials are relatively small, as they are across pharmacological and physical treatments in FM (Nüesch 2013; Winkelmann 2012). This raises the issue of whether FM trials should still focus on mere symptom reduction. FM trials on psychological therapies, especially acceptance‐based CBTs, could provide a better answer on how to live more satisfactorily with FM symptoms (Williams 2012).

Disease‐relevant timelines

Follow‐up of all study participants (including members of the control group) should be for at least one year (Brachaniec 2009).

Study type and methodology (methodology quality)

The following requirements should be met for the 'green' criterion of the suggested Cochrane Collaboration traffic light system (Moore 2010a). Randomization should be performed by an adequate method. Treatment allocation should be undertaken independently. Investigators and assessment staff should be blinded to prevent inadvertent bias. Analysis should be performed by an ITT analysis based on baseline observation carried forward method. Patient‐associated barriers to participate in CBTs should be investigated.

Major tasks of future research are: a) the definition of subgroups (e.g. FM patients with and without major depression or post‐traumatic stress disorder, pain persistence and pain avoidance behaviour) with development of more tailored therapies to these subgroups (Lange 2010; Loevinger 2012; Lumley 2011; van Koulil 2011), b) the identification of common (e.g. therapeutic relationship) and specific (e.g. exposure) treatment mechanisms by more frequent measures than pre‐post assessment allowing examination of lagged relationships over time (Thurns 2011) c) the development of cost‐effective telephone or internet‐assisted therapies to overcome the costs of CBTs and limited availability of providers with specialized training (Kashikar‐Zuck 2010; McBeth 2012) and d) the application of new study designs beyond the "gold standard" of RCTs, e.g. clinical effectiveness trials where clinical effectiveness is "the product of efficacy, tolerability, utility, cost, and speed" (Moore 2010c).

Summary of findings

Open in table viewer
Summary of findings for the main comparison.

Cognitive behavioural therapies compared to controls for fibromyalgia

Patient or population: Patients with fibromyalgia

Settings: In‐ and outpatient

Intervention: Cognitive behavioural therapies

Comparison: Controls (attention control, treatment as usual, waiting list, other active therapy)

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

controls

Cognitive behavioural therapies

Pain

end of treatment

(0‐10 scale)

Higher scores indicate higher pain levels

Mean pain baseline

7.37 (SD 2.10) 3

The mean pain in the intervention groups was 0.29 standard deviations lower (0.49 to 0.11 lower)

1382
(20)

⊕⊕⊝⊝
low1,2

SMD ‐0.29 (95% CI ‐0.47 to ‐0.11)

8.5% (95% CI 3.1% to 14.0%) relative improvement

6.3 % (95% 2.3% to 10.3%) CI) fewer points on the scale (absolute change)

NNTB 7 (95% CI 5 to19)

Pain
follow‐up median 6 months

(0‐10 scale)

Higher scores indicate higher pain levels

Mean pain baseline

64.72 (SD 10.44) 4

The mean pain in the intervention groups was 0.40 standard deviations lower (0.64 to 0.16 lower)

822
(14)

⊕⊕⊝⊝
low1,2

SMD ‐0.40 (95% CI ‐0.62 to ‐0.17)

6.4% (95% CI 2.7% to 9.9%) relative improvement

4.2% (95% CI 1.8% to 6.5%) fewer points on the scale (absolute change)

NNTB 10 (95% CI 6 to 24)

Negative mood

end of treatment

(0‐10 scale)

Higher scores indicate higher negative mood levels

Mean depression

baseline 6.82 (SD 3.11) 5

The mean negative mood in the intervention groups was 0.33 standard deviations lower (0.49 to 0.17 lower)

1578
(18)

⊕⊕⊝⊝
low1,2

SMD ‐0.33 (95% CI ‐0.49 to ‐0.17)

15.0% (95% CI 7.7% to 22.3%) relative improvement

10.2% (95% CI 5.2% to 15.2%) fewer points on the scale (absolute change)

NNTB 6 (95% CI 4 to12)

Negative mood
follow‐up median 6 months

(0‐50 scale)

Higher scores indicate higher negative mood levels

Mean depression

baseline 14.94 (SD 3.11) 6

The mean negative mood in the intervention groups was 0.43 standard deviations lower (0.75 to 0.11 lower)

721
(11)

⊕⊕⊝⊝
low1,2

SMD ‐0.43 (95% CI ‐0.75 to ‐0.11)

8.9% (95% CI 2.3% to 15.8%) relative improvement

2.7% (95% CI 0.1% to 4.7%) fewer points on the scale (absolute change)

NNTB 11 (95% CI 6 to 43)

Disability

end of treatment

(0‐10 scale)

Higher scores indicate disability levels

Mean physical impairment baseline 2.80 (SD 2.40) 7

The mean disability in the intervention groups was 0.30 standard deviations lower (0.51 to 0.08 lower)

1163
(15)

⊕⊕⊝⊝
low1,2

SMD ‐0.30 (95% CI ‐0.51 to ‐0.08)

25.8 % (95% CI 6.9% to 43.7% relative improvement

7.2% (95% CI 1.9% to 12.2%) fewer points on the scale (absolute change)

NNTB 7 (95% CI 4 to 26)

Disability
follow‐up median 6 months

(0‐10 scale)

Higher scores indicate disability levels

Mean physical impairment baseline 3.24 (SD 2.26) 8

The mean disability in the intervention groups was 0.52 standard deviations lower (0.86 to 0.18 lower)

664
(9)

⊕⊕⊝⊝
low1,2

SMD ‐0.52 (95% CI ‐0.86 to ‐0.18)

36.4% (95% CI 1.3% to 60.2%) relative improvement

11.7% (95% CI 4.1% to 19.4%) fewer points on the scale (absolute change)

NNTB 4 (95% CI 3 to12)

Acceptability

end of treatment (dropouts from study due to any reasons)

136 (94 to 195) per 1000

127 (88 to 182)

RR 0.94 (0.65 to 1.35)

1914 (21)

⊕⊕⊝⊝
low1

Absolute risk difference
0% (95% CI ‐1 to 0)

Relative per cent change
6% (95% CI 15%
improvement to 35%
worsening)

Not statistically significant

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Some studies with lack of reported allocation concealment, without intention‐to‐treat analysis and with selective outcome reporting

2 High heterogeneity of treatment effect

3 Luciano 2011: N=216 patients; Pain VAS 0‐10 scale

4 Alda 2011: N=113 patients; Pain VAS 0‐100 scale

5 Luciano 2011: N=216 patients; Depression VAS 0‐10 scale

6 Alda 2011; N=113 patients; Hamilton Rating Scale for Depression (0‐50)

7 Luciano 2011: N=216 patients; Physical impairment VAS 0‐10 scale

8 Alda 2011; N=113 patients; Physical impairment VAS 0‐10 scale

Open in table viewer
Summary of findings 2. Cognitive behavioural therapies versus controls for fibromyalgia

Cognitive behavioural therapies versus controls for fibromyalgia

Patient or population: Patients with fibromyalgia

Settings: In‐ and outpatients

Intervention: Cognitive behavioural therapies

Comparison: Controls (attention control, treatment as usual, waiting list, other active therapy)

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Control

Cognitive behavioural therapies versus controls final treatment

Fatigue

end of treatment

(0‐10 scale)

Higher scores indicate higher fatigue levels

Mean fatigue score 8.13 (SD 1.89) 3

The mean fatigue in the intervention groups was
0.25 standard deviations lower
(0.49 to 0.02 lower)

910
(11 studies)

⊕⊕⊝⊝
low1

SMD ‐0.25 (95% CI ‐0.49 to ‐0.02)

5.8% (95% CI 0.05% to 11.3%) relative improvement

4.7% (95% CI 0.4% to 9.3%) fewer points on the scale (absolute change)

NNTB 9 (95% CI 5 to109)

Sleep problems

end of treatment

(0‐50 scale)

Higher scores indicate more sleep problems

Mean sleep problems score 27.9 (SD 8.8) 4

The mean sleep problems in the intervention groups was
0.4 standard deviations lower
(0.85 lower to 0.05 higher)

422
(8 studies)

⊕⊕⊝⊝
low2

SMD ‐0.40 (95% CI ‐0.85 to 0.05)

0.3% (95% CI ‐0.03% to 1.7%) relative improvement

7.0% (95% CI ‐0.90% to 15.0%) fewer points on the scale (absolute change)

NNTB 5 (95% CI ‐45 to 3)

Health‐related

quality of life

end of treatment

(0‐80 scale)

Higher scores indicate lower health‐related quality of life

Mean health‐related quality of life score 55.97 (SD 15.95) 5

The mean health‐related quality of life in the intervention groups was
0.23 standard deviations lower
(0.38 to 0.08 lower)

1238
(13 studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.23 (95% CI ‐0.38 to ‐0.08)

0.08% (95% CI 0.03% to 0.13%) relative improvement

4.6% (95% CI 1.6% to 7.6%) fewer points on the scale (absolute change)

NNTB 9 (95% CI 6 to 27)

Fatigue

Follow‐up median 6 months

(0‐10 scale)

Higher scores indicate higher fatigue levels

Mean fatigue score

Mean 8.32 (SD 2.17)6

The mean fatigue in the intervention groups was
0.46 standard deviations lower
(0.77 to 0.15 lower)

429
(6 studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.46 (95% CI ‐0.77 to ‐0.15)

1.2% (95% CI 0.4% to 2.0%) relative improvement

10.0% (95% CI 3.2% to 16.7%) fewer points on the scale (absolute change)

NNTB 5 (95% CI 3 to 14)

Sleep problems
Follow‐up median 6 months

(0‐50 scale)

Higher scores indicate more sleep problems

Mean sleep problems score 27.9 (SD 8.8) 4

The mean sleep problems in the intervention groups was
0.64 standard deviations lower
(1.31 lower to 0.03 higher)

378
(7 studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.64 (95% CI ‐1.31 to 0.03)

0.4% (95% CI ‐0.02% to 0.8%) relative improvement

11.2% (95% CI ‐0.53% to 23.1%) fewer points on the scale (absolute change)

NNTB 4 (95% CI ‐74 to 2)

Health‐related

quality of life

Follow‐up median 6 months

(0‐80 scale)

Higher scores indicate lower health‐related quality of life

Mean health‐related quality if life score 64.48

(SD 10.50) 7

The mean health‐related quality of life in the intervention groups was
0.19 standard deviations lower
(0.58 lower to 0.21 higher)

425
(6 studies)

⊕⊕⊝⊝
low1

SMD ‐0.19 (95% CI ‐0.58 to 0.21)

0.03% (95% CI ‐0.03% to 0.15%) relative improvement

2.0% (95% CI ‐2.2% to 6.1%) fewer points on the scale (absolute change)

NNTB 12 (95% CI ‐17 to 6)

Acceptability
Follow‐up: median 6 months

See comment

See comment

Not estimable

‐

See comment

Not assessed

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Some studies with lack of reported allocation concealment, without intention‐to‐treat analysis and with selective reporting
2 High heterogeneity of treatment effect

3 Luciano 2011: N=216 patients; VAS 0‐10 scale

4 Castel 2012; N=60 patients; NRS 0‐50 scale

5 Luciano 2011: N=216 patients; VAS 0‐80 scale

6 Alda 2010; N= 113; VAS 0‐10 scale

7 Alda 2010; N= 113; VAS 0‐100 scale

Background

Description of the condition

The key symptoms of fibromyalgia (FM) are chronic widespread pain associated with cognitive dysfunction, physical fatigue and sleep disturbances (Häuser 2008; Wolfe 2010). Patients often report high disability levels and poor quality of life along with extensive use of medical care (Winkelmann 2011; Wolfe 1997). Lacking a specific laboratory test, diagnosis is established by a history of the key symptoms and the exclusion of somatic diseases sufficiently explaining the key symptoms (Häuser 2008; Wolfe 2010). For a clinical diagnosis the 1990 and 2010 American College of Rheumatology (ACR) criteria (Wolfe 1990; Wolfe 2010) and the Association of the Medical Scientific Societies in Germany (AWMF) diagnostic criteria (Häuser 2010) can be used. In the past other standardised criteria have been used to diagnose FM (Smythe 1981; Yunus 1981).

FM is estimated to affect 1% to 2% of people in the United States (Lawrence 2008) and 2.1% to 2.9% in Europe (Branco 2010; Wolfe 2013).

The definite aetiology (causes) of this syndrome remains unknown. A model of interacting biological and psychosocial variables in the predisposition, triggering and development of the chronicity of FM has been suggested (Sommer 2012a). Depression (Forseth 1999), genes (e.g. 5‐hydroxytryptamine 2A receptor 102T/C polymorphism) (Lee 2012), obesity combined with physical inactivity (Mork 2010), physical and sexual abuse in childhood (Häuser 2011), sleep problems (Mork 2012) and smoking (Choi 2010) predict future development of FM. Depression and post‐traumatic stress disorder worsen FM symptoms (Dell' Osso 2012; Lange 2010).

Several factors are associated with the pathophysiology (functional changes associated with or resulting from disease) of FM, but the relationship is unclear. The functional changes include alteration of pain processing in the brain, reduced reactivity of the hypothalamus‐pituitary‐adrenal axis to stress, increased pro‐inflammatory and reduced anti‐inflammatory cytokine profiles (produced by cells involved in inflammation), and disturbances in neurotransmitters such as dopamine and serotonin (Sommer 2012a). Prolonged exposure to stress, as outlined above, may contribute to these functional changes (Bradley 2009).

Current treatments for FM are not curative. Drugs (Häuser 2013; Moore 2011; Tort 2012) and exercise therapies (Busch 2007) aim to relieve symptoms and improve quality of life and functional abilities.

Description of the intervention

Behavioural and cognitive behavioural psychological therapies are the dominant contemporary psychological treatments for a wide range of health problems, including chronic pain (Morley 2011). Behavioural and cognitive behavioural psychological therapies are used to manage chronic pain by attempting to change negative thoughts about pain, and introduce behaviour modification, including self‐management techniques, to improve function and cope with pain. However, there is no universally accepted definition of which techniques constitute behavioural and cognitive behavioural psychological therapies (Morley 2011). Due to the broad variety of behavioural and cognitive behavioural psychological therapy techniques, we use in the following context the term 'cognitive behavioural therapies' (CBTs). For the purposes of this review we will consider the following techniques (Jensen 2011).

  1. Operant therapy, which requires techniques to increase activity, the inclusion of significant others to reduce reinforcement of pain behaviours, and the reduction of pain‐contingent medication (Fordyce 1976; Thieme 2003).

  2. Traditional cognitive behavioural therapy (CBT), which requires monitoring of one's own thoughts, feelings and behaviours with respect to the target symptom (e.g. by a symptom diary) and the promotion of alternative ways of coping with the target symptom (also labelled as problem‐solving techniques, self management, coping skills), through methods such as activity participation and skill‐building or practice opportunities (Bennett 2006).

  3. Self management education programmes, which require information on the clinical picture of FMS, cognitive and behavioural skills mastery to manage pain and limitations of daily activities, and modelling as supplied by the facilitators to target cognitive, behavioral and emotional change (Burckhardt 2005b; Warsi 2003).

  4. Acceptance‐based CBTs, which include acceptance and commitment therapy, or contextual CBT or mindfulness‐based cognitive therapy. All these therapies use acceptance techniques (e.g. mindfulness meditation training) to facilitate a separation between 'self' and one’s thoughts, feelings and pain experience, and encourage patients to base their actions on their most important values as opposed to their immediate feelings, thoughts and pain (Veehof 2011).

How the intervention might work

CBTs include interventions that are based on the premise that chronic pain and other symptoms of FM are maintained and influenced by emotional and cognitive (conscious intellectual activities such as thinking, reasoning or remembering) as well as behavioural factors. A typical treatment protocol for traditional CBT will involve methods aimed directly at assessing the thoughts associated with pain, the extent of avoidance of unpleasant thoughts and of painful experiences, and the consequences of these. A common focus is on strongly held beliefs about pain and their relationship with behaviour, which typically worsens the situation in the shorter or longer term. Behavioural methods focus on the identification of behaviour that is contingent on pain, or upon events which provide pain relief or comfort, and the development of behaviour that is contingent instead on goal achievement related to the values of the individual with pain (Bennett 2006; Williams 2012). Most CBTs include education (information on the etiology of the disease including importance of psychological factors; treatment options; working mechanisms of psychological and drug therapies).

Why it is important to do this review

The significance of CBTs in the management of FM still needs to be determined. Systematic narrative and quantitative reviews on CBTs in FM have had divergent results. Koulil (Koulil 2007) concluded from six randomized controlled trials (RCTs) that the effects on pain, disability and mood were limited, and that it was mostly CBTs within a multi‐component approach that yielded improvements. Bennett concluded from six RCTs that CBT as a single treatment modality did not offer any distinct advantage over well‐planned group programmes of education or exercise, or both (Bennett 2006). Thieme and coworkers concluded from 14 studies that CBTs were superior to controls in most key domains of FMS post‐treatment and at follow‐up (Thieme 2009). A recent Cochrane review on the efficacy of psychological therapies in chronic pain syndromes included only six studies with FM patients and did not present a subgroup analysis of FM patients (Williams 2012). Another recent review on psychological therapies in FM concluded that CBTs were effective in relieving FM symptoms and superior to other psychological therapies. However, this review included a combination of CBTs with aerobic exercise (multi‐component therapies) and did not compare the results of CBTs with control groups (Glombiewski 2010). A meta‐analysis on CBTs in FM found that CBTs were superior to controls in reducing depressed mood post‐treatment but not superior in reducing pain, fatigue, sleep and limitations in quality of life post‐treatment and at follow‐up. This systematic review included trials with mindfulness‐based stress reduction (MBSR) and excluded trials with self management approaches (Bernardy 2010).

Evidence‐based guidelines on the management of FM have given different grades of recommendation for CBTs. The American Pain Society (Burckhardt 2005a) gave the highest grade of recommendation to CBTs based on a qualitative systematic review. The European League Against Rheumatism only gave a weak (expert opinion) recommendation for CBTs based on a quantitative systematic review (Carville 2008). The Association of the Scientific Medical Societies in Germany gave an open recommendation based on a quantitative systematic review (Köllner 2012). The Canadian Pain Society gave a strong recommendation of CBTs based on a quantitative systematic review (Fitzcharles 2012).

Objectives

To assess the short and long‐term benefits and harms of CBTs compared to control groups in the treatment of FM patients of any age.

Methods

Criteria for considering studies for this review

Types of studies

We selected randomized controlled trials (RCTs) of CBTs of any duration of treatment in FM. According to the eligibility criteria of The Cochrane Collaboration (Higgins 2011), a trial was eligible if, on the basis of the best available information, it was judged that the individuals were definitely or possibly assigned prospectively to one of two (or more) alternative forms of health care using random allocation or some quasi‐random method of allocation (such as alternation, date of birth, or case record number). We also considered cluster‐randomized trials to be eligible. We accepted an attention control, waiting list control, treatment as usual, no therapy and any other active therapy as controls. We included RCTs if they:

  • were available as a full publication or a report of the RCT in a peer‐reviewed journal or in a database (detailed below);

  • had a design that placed a CBT as an active treatment of primary interest;

  • had a credible CBT content (see Types of interventions);

  • had 10 or more participants in each treatment arm at the end of the assessment (Eccleston 2009).

Types of participants

We included patients of any age with a clinical diagnosis of FM by any published, recognised and standardised criteria (Häuser 2010; Smythe 1981; Wolfe 1990; Wolfe 2010; Yunus 1981). We included studies in which FM patients were mixed with patients having other chronic pain syndromes if the outcomes of FM patients were reported separately or could be provided on request.

Types of interventions

We included RCTs comparing a credible CBT treatment with controls. We judged a psychological treatment to be credible if it was based on an extant CBT model or framework (see Background) and its delivery was from, or supervised by, a healthcare professional trained in an extant CBT model or framework. In addition, the delivery by a lay leader was accepted in the case of self management education programs. The continuation of previous therapies as usual care was allowed.

We included trials comparing face‐to‐face, telephone‐based or internet‐based CBTs as an active treatment of primary interest with controls.

We excluded the following types of psychological therapies from this review.

  • Biofeedback, hypnosis, mindfulness‐based stress reduction and relaxation training as single therapies. These psychological therapies are also attributed to complementary and alternative medicine (CAM) (National Institutes of Health 2011) and are included in another Cochrane review in preparation on mind‐body therapies in FMS (Theadom 2009). Furthermore we excluded studies that included hypnosis and mindfulness‐based stress reduction as part of a complex CBTs intervention because it would not be possible to separate the effects of CBTs from these therapies.

  • Studies with education only: any combination of information on the symptoms and management of FM, discussion or emotional support without skills mastery and modelling as supplied by the facilitators.

  • Studies in which CBTs were combined with any other defined active therapy (physical exercise, physical therapy or drug therapy with defined extent and intensity (so‐called multicomponent therapy), because it is not possible to separate the effects of CBTs from these other active therapies.

Types of outcome measures

We based the selection of outcome measures on the key domains of FM developed through consensus among experts and FM patients (Mease 2009), the goals of CBTs (Bennett 2006; Eccleston 2009), the suggestions of the Initiative of Methods, Measurement and Pain Assessment in Clinical trials (IMMPACT) (Dworkin 2008; Dworkin 2009) and best practice in the reporting of systematic reviews in chronic pain (Moore 2010a). We selected outcome measures for short‐term (at final treatment) and long‐term (follow‐up of at least six months) efficacy.

The primary data type was measurement using continuous scales (Bernardy 2010; Eccleston 2009). We did not meta‐analyse dichotomous outcome data based on clinical improvement (responder analysis). These data have been rarely reported in psychological trials of FMS (Bernardy 2010; Eccleston 2009). We present in Characteristics of included studies which studies reported responder analysis for pain and disability and, if reported, reasons for dropping out. Although standard trial reporting guidance promotes the definition of major outcomes (Dworkin 2008) most psychological trials in chronic pain do not define an a priori major outcome. From each trial we selected the measure considered most appropriate for each of the outcomes. When there was more than one measure for an outcome we gave preference to the measure that was most frequently used (Eccleston 2009). Also, when there was a choice between single‐item and multi‐item self report tools, we chose multi‐item tools on the basis of inferred increased reliability (Eccleston 2009).

We analysed outcome measures at final treatment (end of therapy) and at long‐term (at least six months) follow‐up. Follow‐ups < six months were not considered for the analysis of long‐term follow‐up. If more than one follow‐up after six months had been conducted, the results of the final follow‐up visit were extracted for analysis.

Major outcomes

  1. Self reported pain at end of treatment and at long‐term (at least six months) follow‐up

  2. Self reported negative mood at end of treatment and at long‐term (at least six months) follow‐up

  3. Self reported disability at end of treatment and at long‐term (at least six months) follow‐up

  4. Acceptability: total dropout rate (patients who terminated the trial early for any reason during the treatment period (Cipriani 2009)). We analysed reasons for dropout if reported.

Minor outcomes

  1. Self reported pain self efficacy (beliefs in one's capabilities to manage one's own pain) at end of treatment and at long‐term (at least six months) follow‐up

  2. Self reported fatigue at end of treatment and at long‐term (at least six months) follow‐up

  3. Self reported sleep problems at end of treatment and at long‐term (at least six months) follow‐up

  4. Self reported disease‐specific health‐related quality of life (HRQOL) measured by the Fibromyalgia Impact Questionnaire (FIQ) at end of treatment and at long‐term (at least six months) follow‐up

Search methods for identification of studies

Electronic searches

We ran an electronic search in the Cochrane Central Register of Controlled Trials (CENTRAL (The Cochrane Library 2013, Issue 1), MEDLINE accessed through PubMed (1966 to 15 February 2013), PsycINFO (1966 to 15 February 2013) and SCOPUS (1980 to 15 February 2013). We searched http://www.clinicaltrials.gov (website of the US National Institutes of Health) and the World Health Organization Clinical Trials Registry Platform (ICTRP) (http://www.who.int/ictrp/en/) for ongoing trials.

We used the search terms fibromyalgia, CBTs and their variations (see Appendix 1).

Searching other resources

We searched bibliographies from retrieved relevant articles. We contacted content experts for further possible studies. Our search included all languages.

Data collection and analysis

Selection of studies

Two review authors (KB, PK) independently scrutinised all the titles and abstracts and selected studies based on inclusion and exclusion criteria. A third review author verified the result (WH).

Data extraction and management

Two authors extracted data on the studies (including the methods, participants, interventions, outcomes, and results) independently using a specially designed data extraction form (KB, WH). The types of treatment and reported treatment quality were rated independently by two authors (KB, WH). We resolved disagreements by discussion, if necessary a third review author (AB) was consulted. One author (WH) entered data into Review Manager (RevMan) 5 (RevMan 2011) and a second author (KB) validated the entries.

Assessment of risk of bias in included studies

Two review authors (KB, WH) independently assessed the risk of bias of each included trial. Disagreements were resolved by discussion and consensus, otherwise a third review author (AB) acted as arbiter.

For each included study, we assessed risk of bias against key criteria: random sequence generation; allocation concealment; blinding of outcome assessment; incomplete outcome data; and selective outcome reporting, in accordance with methods recommended by The Cochrane Collaboration (Higgins 2011). We excluded the option of 'blinding participants and personnel' because neither therapists nor patients can be blinded to whether they deliver or receive treatment (Williams 2012).

We explicitly judged each of these criteria as: low risk of bias, high risk of bias, or unclear (either a lack of information or uncertainty over the potential for bias) risk of bias. We present the 'Risk of bias' assessment results in the 'Risk of bias' graph and 'risk of bias' summary figures.

Assessment of quality of the treatment

We assessed the quality of the treatment using five criteria (treatment content and setting, treatment duration, manualisation of the treatment, adherence of the therapist to the manual, therapist training and client engagement) on a quality rating scale designed specifically for application to psychological treatment studies in pain. The total score ranges from 0 to 9 (Yates 2005). We considered scores 0 to 2 to indicate poor quality, scores 3 to 5 average, and scores 6 to 9 excellent treatment quality (Bernardy 2010).

Assessment of study samples

We extracted demographic data (percentage of women, mean age) and history of disease (mean durations of chronic widespread pain or FMS symptoms) from study samples. We evaluated external validity of the study samples by checking if patients with depressive or anxiety disorders were included to assess the representativeness of study samples for FMS patients in clinical practice (Bernardy 2010). We explicitly judged this criterion using the terms high risk of limited external validity (patients with depressive or anxiety disorders were excluded), no risk of limited external validity (patients with depressive or anxiety disorders were included) and unclear risk of limited external validity (insufficient details were given).

Measures of treatment effect

The effect measures of choice were standardised mean difference (SMD) (when different scales are used to measure outcomes) for continuous data and risk ratio (RR) for dichotomous data of CBTs groups and control groups at end of treatment and at final follow‐up. We used a random‐effects model. We expressed precision with 95% confidence intervals (CIs). We used Cohen’s categories to evaluate the magnitude of the effect size, calculated by SMD, with Hedges' g > 0.2 to 0.5 = small effect size, g > 0.5 to 0.8 = medium effect size, and g > 0.8 = large effect size (Cohen 1988). We labelled g < 0.2 to be a 'not substantial' effect size. We converted SMD to relative and absolute change by multiplying by the baseline standard deviation from the control group of a 'representative trial, and relative per cent change by dividing the absolute change by the baseline mean of the control group from the same representative trial for some results in the summary of findings table (Bliddal 2009).

Unit of analysis issues

In the case of unit of analysis issues we followed the suggestions in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). In the case of cross‐over design we used the methods of analysis for cross‐over trials: we analysed paired data if available or provided by request. If no paired data were available we used first‐period data. In the case of repeated observations on participants we selected the longest follow‐up from each study. In the case of different types of CBTs we analysed the different types of CBTs separately. If different types of CBTs were compared with only one control group we adjusted the number of patients in the control group according to the number of patients in the CBT arms. In the case of different types of control groups we used the following preference for comparison with CBTs: attention control, waiting list control, treatment as usual, no therapy, and any other active therapy. We did not combine different types of control groups.

Dealing with missing data

Where means or standard deviations (SDs) were missing, we attempted to obtain these data through contacting the trial authors. Where SDs were not available from trial authors, we calculated them from t values, confidence intervals or standard errors, where reported in articles (Higgins 2011). If these data were not available, we substituted the missing SD by a validated imputation method (Furukawa 2006).

Assessment of heterogeneity

We extracted demographic (average age, percentages of women) and clinical characteristics of the patients (duration of FMS symptoms) as well as study characteristics (country and setting of study, type and duration of CBTs) as potential sources of clinical heterogeneity. We used the I2 statistic to describe the percentage variability of effect estimates that is due to heterogeneity. We combined results in a meta‐analysis using a random‐effects model. I2 values above 50% indicate high heterogeneity, between 25% and 50% moderate heterogeneity, and below 25% low heterogeneity.

Assessment of reporting biases

We avoided language publication bias by including studies irrespective of the language of publication.

We addressed publication bias by visual inspection of funnel plots and tests for funnel plot asymmetry (Begg 1994; Egger 1997) when there were at least 10 studies included in the meta‐analysis (Higgins 2011).

We addressed outcome reporting bias by checking if the means and SDs of all primary and secondary outcomes, as outlined in the methods section of the published studies, had been reported or had been provided on request.

Data synthesis

We examined the combined results using a random‐effects model (inverse variance method) because this model is more conservative than the fixed‐effect model and incorporates both within‐study and between‐study variance and because we expected clinical and statistical heterogeneity. We used the GRADE approach to grade the quality of evidence (Brozek 2009; Higgins 2011). We used the software GradePro (Guyatt 2006). We presented a 'Summary of findings' table with the major outcomes (pain, disability, mood, and acceptability).

The numbers needed to treat for an additional outcome of benefit (NNTB) for continuous variables (pain, fatigue, sleep problems, negative mood, disability, disease‐specific quality of life, self reported pain self efficacy) were calculated using the Wells calculator software available at the Cochrane Musculoskeletal Group editorial office, which estimates the proportion of patients who will benefit from treatment from SMDs. The estimation of responders is nearly independent from the minimally important difference (MID) (Norman 2001). We used a minimal clinically important difference of 15% for the calculation of NNTB from SMDs for all continuous outcomes.

Subgroup analysis and investigation of heterogeneity

We analysed the effects of all types of CBTs pooled together compared to all types of control groups pooled together for major and minor outcomes. We performed subgroup analyses of the efficacy of the different types of CBTs (operant (behavioural) therapies, traditional cognitive behavioural therapies, self management approaches, acceptance‐based cognitive behaviour therapies); face‐to‐face versus other (telephone, internet‐based) CBTs; CBTs in adults (persons ≥18 years) versus children and adolescents (persons < 18 years); and different types of control groups (attention controls, active controls and other types of controls [treatment as usual, waiting list control]) compared to all types of CBTs pooled together. At least two studies should be available for subgroup analysis. We tested for subgroup differences using the test of interaction (Altman 2003).

We performed a subgroup analysis of ultra‐short (< 5 sessions), short‐term (5 to 25 sessions) and long‐term CBTs studies (> 25 sessions) with the primary outcome measures of pain, negative mood and disability to test for dosage effects. We performed a subgroup analysis of studies with low, moderate and high reported treatment quality to test for effects of treatment quality on outcomes. We expected better treatment outcomes in studies with high treatment quality compared to studies with low treatment quality.

Sensitivity analysis

We performed sensitivity analysis of all types of CBTs pooled together compared to all types of control groups pooled together for the three primary outcomes of pain, mood and disability: a) based on the need to substitute means, SDs, or both, by excluding studies with substituted values or values visually extracted from figures; b) based on the risk of bias by excluding studies with high and unclear selection bias, studies with high and unclear attrition bias, studies with high and unclear reporting bias, and studies with high and unclear risk of limited external validity; c) (post hoc decision) excluding studies with < 20 participants per treatment arm in accordance with the Cochrane reviews on psychological therapies for the management of chronic pain in children and adolescents (Eccleston 2012) and in adults (Williams 2012).

Results

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification

Results of the search

We identified 1519 studies. We excluded 1477 references as they did not fulfil inclusion criteria related to the interventions evaluated in this review. We identified 42 studies potentially related to CBTs, and the full text was obtained for each of them. Of these 42 studies, 15 did not meet the inclusion criteria and were excluded. Four studies which were identified by a second and third search were included in Characteristics of studies awaiting classification for a total of 23 studies with 24 pairs of study arms to be included in the analysis (Figure 1).

Figure 1
Study flow diagram.

Study flow diagram.

The following studies are awaiting classification and will be included in the update of this review: Three studies with cognitive behavioural therapy (Jensen 2012; Sanchez 2012; Martinez 2013) and one study with acceptance and commitment therapy (Wicksell 2012) which were found in the second search of 28 February 2012 and in the third search of 28 August 2013.

Included studies

The main characteristics of the studies are summarized in Table 1.

Open in table viewer
Table 1. Main characteristics of included studies

Author

Country

Type of CBT

Type of control group

Duration CBT (weeks)

Number of CBT sessions

Total treatment time CBT (hours)

Number of patients in CBT group

% women

Number of patients in control group

% women

Alda 2011 *

Spain

CBT

TAU

12

6

15

57

95

56

96

Ang 2010 *

USA

CBT

TAU

12

6

3

17

100

15

100

Burckhardt 1994 *

Sweden

CBT

Delayed treatment

6

6

9

28

100

30

100

Castel 2009 *

Spain

CBT

TAU

11

12

18

18

94

12

86

Castel 2012 *

Spain

CBT

TAU

14

14

28

34

94

30

100

Edinger 2005 *

USA

CBT

TAU

6

6

6

16

94

12

100

Falcao 2008 *

Brazil

CBT

TAU

10

20

30

30

100

30

100

Kashikar‐Zuck 2005 **

USA

CBT

Active control

8

8

12

14

100

14

100

Kashikar‐Zuck 2012 **

USA

CBT

Active control

8

8

6

57

95

57

90

King 2002 *

USA

CBT

Delayed treatment

12

12

18

48

100

39

100

Luciano 2011 *

Spain

CBT

TAU

8

8

16

108

95

108

98

Miro 2011 *

Spain

CBT

Active control

6

6

9

20

100

20

100

Nicassio 1997 *

USA

CBT

Active control

10

10

15

36

89

35

89

Oliver 2002 *

USA

Self‐management

TAU

52

10

20

207

96

193

94

Redondo 2004 *

Spain

CBT

Active control

8

8

20

21

100

19

100

Rooks 2007 *

USA

Self‐management

Active control

16

8

16

51

100

50

100

Soares 2002 *

Sweden

CBT

Attention control

10

10

120

18

100

18

100

Thieme 2003 *

Germany

Operant therapy

Active control

5

25

75

42

100

21

100

Thieme 2006a *

Germany

Operant therapy

Attention control

15

15

30

42

100

20

100

Thieme 2006b *

Germany

CBT

Attention control

15

15

30

43

100

20

100

Vlayen 1996 *

Netherlands

CBT

Active control

6

12

18

49

93

43

82

Wigers 1996 *

Norway

CBT

TAU

14

15

30

20

90

20

95

Williams 2010 *

USA

Self‐management

TAU

26

NR

59

95

59

95

Woolfolk 2012 *

USA

CBT

TAU

NR

NR

38

89

38

87

* Studies included only adults  

** Studies included only children and adolescents

NR = Not reported and not provided on request

TAU = Treatment as usual

Settings

Twenty‐three studies with 24 pairs of study arms were analysed. Twelve studies were conducted in Europe (Alda 2011; Burckhardt 1994; Castel 2009; Castel 2012; Luciano 2011; Miro 2011; Redondo 2004; Soares 2002; Thieme 2003; Thieme 2006; Vlayen 1996; Wigers 1996), 10 in North America (Ang 2010; Edinger 2005; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; King 2002; Nicassio 1997; Oliver 2001; Rooks 2007; Williams 2010; Woolfolk 2012) and one in South America (Falcao 2008). Four studies had been conducted before 2000 (Burckhardt 1994; Nicassio 1997; Vlayen 1996; Wigers 1996), the remaining studies were published after 2000. All studies but one (Thieme 2003) were outpatient‐based. Three studies were conducted in primary care (Alda 2011; Luciano 2011; Rooks 2007), two studies in secondary care (Kashikar‐Zuck 2005; Thieme 2003) and the remaining studies in tertiary care (university centres). All studies except five (Alda 2011; Kashikar‐Zuck 2012; Luciano 2011; Oliver 2001; Rooks 2007) were single‐centre studies.

Types of therapies

Nineteen studies provided traditional CBT (Alda 2011; Ang 2010; Burckhardt 1994; Castel 2009; Castel 2012; Edinger 2005; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; King 2002; Luciano 2011; Miro 2011; Nicassio 1997; Redondo 2004; Soares 2002; Thieme 2006; Vlayen 1996; Wigers 1996; Woolfolk 2012), three studies provided self management education (Oliver 2001; Rooks 2007; Williams 2010) and two studies provided operant therapy (Thieme 2003; Thieme 2006). All studies were conducted by live face‐to‐face contact except one study which was provided by the internet (Williams 2010) and one which was delivered by telephone (Ang 2010). The median duration of all CBTs was 10 (5 to 54) weeks. The median number of sessions was 10 (6 to 60) and the median total hours was 18 (3 to 102) hours.The median of follow‐ups which were performed by 17 of 23 studies was 6 (3 to 48) months. Fourteen studies performed follow‐ups at equal to or greater than six months (Alda 2011; Burckhardt 1994; Castel 2012; Edinger 2005; Kashikar‐Zuck 2012; Nicassio 1997; Redondo 2004; Rooks 2007; Soares 2002; Thieme 2003; Thieme 2006; Vlayen 1996; Wigers 1996; Woolfolk 2012).

Controls

Two studies used waiting list controls (Burckhardt 1994; King 2002), two studies used attention controls (Soares 2002; Thieme 2006), eight studies used active controls (Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; Miro 2011; Nicassio 1997; Redondo 2004; Rooks 2007; Thieme 2003; Vlayen 1996), the remaining studies compared CBTs to treatment as usual.

Patients

A total of 1073 patients in treatment groups and 958 patients in control groups were included in the analysis. The median number of patients in CBTs groups was 36 (14 to 207), in controls 30 (11 to 193). Participants were referred and recruited from a wide range of healthcare settings and other sources (self‐help groups, media advertisements). The median percentage of women in CBTs groups was 96% (89% to 100%). Ten studies included only women (Ang 2010; Burckhardt 1994; Falcao 2008; Kashikar‐Zuck 2005; King 2002; Miro 2011; Redondo 2004; Soares 2002; Thieme 2003; Thieme 2006). The median of the mean age was 47.5 (15.2 to 55.4) years. Two studies included only children and adolescents (Kashikar‐Zuck 2005; Kashikar‐Zuck 2012); the other studies included only adults. The median percentage of Caucasians was 93% (79% to 100%). The percentage of Caucasians in the whole sample was probably high because most European studies did not report ethnicity of the patients included. The studies used different criteria of disease duration. Therefore, we did not calculate median values. Overall, the studies included patients with a long disease duration (more than five years) except in three studies (Falcao 2008; Miro 2011; Soares 2002) which reported a disease duration of less than five years. Disease duration in children and adolescents was reported to be three years in one study (Kashikar‐Zuck 2012).

Diagnosis of FM

FM was diagnosed in the two studies with children and adolescents (Kashikar‐Zuck 2005; Kashikar‐Zuck 2012) according to the Yunus criteria (Yunus 1981). Of the studies with adults, one study (Wigers 1996) used the Smythe criteria (Smythe 1981). The remaining studies used the American College of Rheumatology (ACR) 1990 classification criteria (Wolfe 1990) for diagnosis.

Exclusion of anxiety or depressive disorder

Twelve studies included patients with depressive or anxiety disorders, or both (Alda 2011; Ang 2010; Burckhardt 1994; Castel 2009; Castel 2012; King 2002; Nicassio 1997; Luciano 2011; Thieme 2006; Wigers 1996).

Reported treatment quality

Three studies had a low (Castel 2009; Nicassio 1997; Vlayen 1996), 12 studies had a moderate (Burckhardt 1994; Edinger 2005; Falcao 2008; King 2002; Luciano 2011; Oliver 2001; Redondo 2004; Rooks 2007; Soares 2002; Thieme 2003; Wigers 1996; Woolfolk 2012) and nine studies had a high treatment quality (Alda 2011; Ang 2010; Castel 2012; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; Miro 2011; Thieme 2006; Williams 2010) (see Table 2).

Open in table viewer
Table 2. Reported treatment quality

 

Treatment content and setting

Treatment duration

Manualisation

Adherence to manual

Therapist training

Client engagement

Sum

Alda 2011

2

1

2

1

1

0

7

Ang 2010

2

1

2

1

2

0

8

Burckhardt 1994

1

1

0

0

1

0

3

Castel 2009

1

1

0

0

0

0

2

Castel 2012

1

1

2

1

0

1

6

Edinger 2005

1

1

2

0

1

0

5

Falcao 2008

1

1

0

0

1

 

1

4

Kashikar‐Zuck 2005

2

1

2

1

2

1

9

Kashikar‐Zuck 2012

2

1

2

1

2

1

9

King 2002

1

1

0

0

0

1

3

Luciano 2011

2

1

1

0

1

0

5

Miro 2011

2

1

2

0

1

0

6

Nicassio 1997

1

1

0

0

0

0

2

Oliver 2002

2

1

0

0

0

0

3

Redondo 2004

2

1

0

0

0

0

3

Rooks 2007

2

1

0

0

0

0

3

Soares 2002

1

1

0

0

1

0

3

Thieme 2003

2

1

2

0

0

0

5

Thieme 2006

2

1

1

0

2

0

6

Vlayen 1996

1

1

0

0

0

0

2

Wigers 1996

1

1

0

0

1

1

4

Williams 2010

2

1

1

0

1

1

6

Woolfolk 2012

1

1

2

0

0

0

4

Items and scores of treatment quality scale (Yates 2005)

1. Treatment content and setting: 2 ‐ Adequate: a clear rationale for the treatment has been reported along with an adequate description of its content; 1 ‐ Partial: either a clear rationale or a description of the content of the treatment is reported; 0 ‐ Inadequate:neither the rationale for treatment or the treatment content are adequately reported.

2. Treatment duration: 1 – Reported; 0 ‐ Unknown.

3. Manualistion of treatment: 2 ‐ Adequate: there is reference to use of a manual that describes the active components of the treatment of study. If more than one treatment arm, manuals were used for all the appropriate treatments; 1 ‐ Partial:in trials with more than one treatment arm, the use of a manual is described but not for all the treatments that would be expected to be manualised; 0 ‐ Inadequate: no evidence that a manual has been used, but reference is made to various principles.

4. Adherence to the manual: 1 ‐ Adequate: there is evidence that the investigators have checked adherence to the manual during the period of study via direct observations, tape recording or supervisory processes that explicitly state adherence to the manual; 0 ‐ Inadequate: no evidence of adherence checks reported.

5. Therapist training: 2 ‐ Adequate: there is documentation of explicit training for the treatment of the trial; 1 ‐ Partial: the general level of therapist training is reported and is adequate (professionally qualified) but there is no mention of explicit training for the trial; 0 ‐ Inadequate: there is no convincing evidence that the therapists have an adequate level of training (e.g. graduate level) or explicit training for the trial.

6. Client Engagement: 1 ‐ Adequate: documented that evidence of engagement was sought e.g. checks on homework were made, skills practice in sessions; 0 – inadequate: no evidence that checks were made on level of engagement.

Major outcomes

Pain was assessed by a visual analogue scale (VAS) in 12 studies (Alda 2011; Ang 2010; Burckhardt 1994; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; Luciano 2011; Miro 2011; Redondo 2004; Rooks 2007; Wigers 1996; Woolfolk 2012), by a numeric rating scale in five studies (Castel 2009;Castel 2012; Thieme 2003; Thieme 2006; Williams 2010), by the McGill Pain Questionnaire in three studies (Edinger 2005; Soares 2002; Vlayen 1996) and by a scale with a composed score by one study (Nicassio 1997).

Negative mood was assessed by the Beck Depression Inventory in five studies (Falcao 2008; Redondo 2004; Rooks 2007; Vlayen 1996; Woolfolk 2012), by a single item visual analogue scale (VAS scale) in four studies (Burckhardt 1994; Castel 2009; Luciano 2011; Wigers 1996), by the Center for Epidemiological Studies Depression Scale in three studies (Nicassio 1997; Oliver 2001; Williams 2010), by the Hospital Anxiety and Depression Scale in two studies (Castel 2012; Miro 2011), by the Children Depression Inventory in two studies (Kashikar‐Zuck 2005; Kashikar‐Zuck 2012), and in one study each by the Patient Health Questionnaire 8 (Ang 2010), the Profile of Mood States (Edinger 2005), the depression subscale of the Multidimensional Pain Inventory (MPI) (Thieme 2003) and the Hamilton Rating Scale Depression (Alda 2011).

Disability was assessed by a single item VAS scale in seven studies (Alda 2011; Ang 2010; Burckhardt 1994; Castel 2009; Castel 2012; Falcao 2008; Luciano 2011), by the Short Form Health Survey (SF)‐36 subscale physical functioning in three studies (Rooks 2007; Williams 2010; Woolfolk 2012), in two studies each by the MPI disability subscale (Thieme 2003; Thieme 2006) and by the Functional Disability Index (Kashikar‐Zuck 2005; Kashikar‐Zuck 2012), and in one study by the quality of well‐being index (Nicassio 1997).

Dropout rates suitable for analysis with reasons for discontinuation were reported by 17 studies (Alda 2011; Ang 2010; Burckhardt 1994; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; Luciano 2011; Miro 2011; Oliver 2001; Redondo 2004; Rooks 2007; Soares 2002; Thieme 2003; Thieme 2006; Vlayen 1996; Wigers 1996; Williams 2010).

Minor outcomes

Self reported pain self efficacy was assessed by the Self Efficacy Pain Scale in four studies (Burckhardt 1994; King 2002; Oliver 2001; Rooks 2007), in two studies by the Chronic Pain Self Efficacy Scale (Redondo 2004; Woolfolk 2012), and in one study each by the Coping Strategies Questionnaire (Vlayen 1996), the Arthritis Self Efficacy Scale (Soares 2002), the Pain Castastrophizing Scale (Alda 2011), the Pain Coping Questionnaire (Kashikar‐Zuck 2005), the Pain Management Inventory (Nicassio 1997), the MPI Pain Coping Scale (Thieme 2003) and the Pain‐related Self‐statements Scale (Thieme 2006).

Fatigue was assessed by a single item VAS scale in 10 studies (Alda 2011; Ang 2010; Burckhardt 1994; Castel 2009; Castel 2012; Falcao 2008; Luciano 2011; Redondo 2004; Rooks 2007; Wigers 1996) and in one study by the Multidimensional Fatigue Inventory (Williams 2010).

Sleep problems were assessed by single item VAS scale in three studies (Kashikar‐Zuck 2012; Redondo 2004; Wigers 1996) and in one study each by the Karolinska Sleep Questionnaire (Soares 2002), the Insomnia Symptom Questionnaire (Edinger 2005), the Pittsburg Sleep Quality Index (Miro 2011) and the SF‐36 Sleep Scale (Williams 2010).

Disease‐specific health‐related quality of life was assessed by the FIbromyalgia Impact Questionnaire in 12 studies (Alda 2011; Burckhardt 1994; Castel 2009; Castel 2012; Falcao 2008; Luciano 2011; Miro 2011; Oliver 2001; Redondo 2004; Rooks 2007; Soares 2002; Thieme 2006).

Excluded studies

Fourteen studies were excluded: four studies because they were not randomised studies (Anderson 2007; De Voogd 2003; Goldenberg 1994; Lommel 2011), two studies because the predefined criteria of CBTs were not met (Carleton 2011; Goeppinger 2009), two studies because no separate data from FM patients were reported and were not provided on request (Haugli 2008; Solomon 2002), two studies because of < 10 patients per treatment arm (Garcia 2006; Martinez‐Valero 2008), one study because of the combination of CBT with psychodynamic therapy (Langford 2010), and one study each because the authors did not report the continuous outcomes assessed and did not provide these outcomes on request (Williams 2002), because no details of FM diagnosis were reported (Stuifbergen 2010) and because FM diagnosis was not established according to the inclusion criteria of the study (Lorig 2008).

Risk of bias in included studies

Risk of bias could not be properly assessed in some studies due to poor method reporting. Most of the studies reviewed were published prior to the standardisation of RCT reporting, established by the CONSORT statement (Schulz 2010). In general, the risk of bias of included studies was high for incomplete outcome data and selective reporting (Figure 2, Figure 3 for risk of bias summary and graph). Only two studies were without any risk of bias (Alda 2011; Williams 2010). Detailed information regarding risk of bias assessments of every study are given in the Characteristics of included studies table. Moreover, individual treatment groups were relatively small in size, which makes them susceptible to random chance and small sample bias.

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

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Reported random sequence generation was adequate in 13 studies (Alda 2011; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; King 2002; Luciano 2011; Miro 2011; Nicassio 1997; Redondo 2004; Rooks 2007; Wigers 1996; Williams 2010; Woolfolk 2012). Reported allocation concealment was adequate in seven studies (Alda 2011; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; Miro 2011; Rooks 2007; Williams 2010).

Blinding

Ten studies reported blinding of the outcome assessor (Alda 2011; Castel 2012; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; King 2002; Luciano 2011; Miro 2011; Williams 2010; Woolfolk 2012).

Incomplete outcome data

Only 11 studies performed an intention‐to‐treat (ITT) analysis (Alda 2011; Castel 2012; Kashikar‐Zuck 2012; King 2002; Luciano 2011; Redondo 2004; Soares 2002; Thieme 2006; Wigers 1996; Williams 2010; Woolfolk 2012).

Selective reporting

Visual inspection of funnel plots was not indicative of publication bias. In the Egger’s test the intercept of the effect size on pain at end of treatment was ‐0.96 (95% CI ‐3.49 to 1.57) with t = 0.79 (two‐tailed P = 0.46). In the Begg’s test Kendall’s tau without continuity correction was ‐0.08 and z = 0.56 (two‐tailed P = 0.43). Both tests were not indicative of publication bias.

Fourteen studies reported all study outcomes or provided these data on request (Alda 2011; Castel 2009; Castel 2012; Falcao 2008; Kashikar‐Zuck 2005; Kashikar‐Zuck 2012; Luciano 2011; Nicassio 1997; Redondo 2004; Rooks 2007; Thieme 2003; Thieme 2006; Wigers 1996; Williams 2010). Nine authors (Ang 2010; Burckhardt 1994; Edinger 2005; King 2002; Miro 2011; Oliver 2001; Soares 2002; Vlayen 1996; Woolfolk 2012) did not provide missing outcomes on request.

Effects of interventions

See: Summary of findings for the main comparison ; Summary of findings 2 Cognitive behavioural therapies versus controls for fibromyalgia

Of the 15 analyses which compared CBTs with controls, 14 had high heterogeneity and one had moderate heterogeneity. In the subgroup analyses, the heterogeneity of the effect size of operant therapy on pain and disability at end of treatment was > 90%, which is indicative of highly inconsistent findings of the two operant therapy studies (Thieme 2003; Thieme 2006). Subgroup analyses revealed that the high heterogeneity was due to studies with active controls and that heterogeneity in studies with other types of controls was low to moderate (Appendix 2).

Some missing SDs of one study (Burckhardt 1994) had to be calculated by a validated imputation method. Means from one study (Woolfolk 2012) had to be extracted from figures.

CBTs versus controls at end of treatment

Major outcomes

Twenty‐one studies with 1453 participants were entered into an analysis of the effects of CBTs on pain. CBTs had a small effect size on pain of ‐0.29 (95% CI ‐0.47 to ‐0.11) (Figure 4). Nineteen studies with 1649 participants were entered into an analysis of the effects of CBTs on negative mood. CBTs had a small effect size on mood of ‐0.33 (95% CI ‐0.49 to ‐0.17) (Figure 5). Sixteen studies of 1234 participants were entered into an analysis of the effects of CBTs on disability. CBTs had a small effect size on disability of ‐0.30 (95% CI ‐0.51 to ‐0.08) (Figure 6). Twenty‐one studies with 1914 participants were entered into an analysis of the acceptability of CBTs. The risk ratio of dropping out for any reason did not differ between CBTs (15.4%) and controls (14.5%) (RR 0.94; 95% CI 0.65 to 1.35) (see summary of findings Table for the main comparison).

Figure 4
Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.1 Pain.

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.1 Pain.

Figure 5
Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.2 Negative mood.

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.2 Negative mood.

Figure 6
Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.3 Disability.

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.3 Disability.

Minor outcomes

Eleven studies with 1047 participants were entered into an analysis of the effects of CBTs on pain self efficacy. CBTs had a small effect size on pain self efficacy of ‐0.49 (95% CI ‐0.80 to ‐0.17). Eleven studies with 910 participants were entered into a study of the effects of CBTs on fatigue. CBTs had a small effect size on fatigue of ‐0.25 (95% CI ‐0.49 to ‐0.02). Eight studies of 422 participants were entered into a study of the effects of CBTs on sleep problems. The overall effect on CBTs on sleep problems was ‐0.40 (95% CI ‐ 0.85 to 0.05), which was not statistically significant (P = 0.06). Thirteen studies of 1238 participants were entered into a study of the effects of CBTs on disease‐specific quality of life. CBTs had a small effect size on disease‐specific quality of life of ‐0.23 (95% CI ‐0.38 to ‐0.08) (see summary of findings Table 2).

CBTs versus controls at long‐term follow‐up

Major outcomes

Fifteen studies with 893 participants were entered into an analysis of the effects of CBTs on pain. CBTs had a small effect size on pain of ‐0.40 (95% CI ‐0.62 to ‐0.17). Twelve studies with 792 participants were entered into an analysis of the effects of CBTs on negative mood. CBTs had a small effect size on negative mood of ‐0.43 (95% CI ‐0.75 to ‐0.11). Ten studies of 735 participants were entered into an analysis of the effects of CBTs on disability. CBTs had a moderate effect size on disability of ‐0.52 (95% CI ‐0.86 to ‐0.18) (see summary of findings Table for the main comparison).

Minor outcomes

Nine studies of 617 participants were entered into an analysis of the effects of CBTs on pain self efficacy. CBTs had a moderate effect size on pain self efficacy of ‐0.75 (95% CI ‐1.27 to ‐0.24). Six studies of 429 participants were entered into a study of the effects of CBTs on fatigue. CBTs had a small effect size on fatigue of ‐0.46 (95% CI ‐0.77 to ‐0.15). Six studies of 425 participants were entered into a study of the effects of CBTs on sleep problems. The overall effect of CBTs on sleep problems of ‐0.64 (95% CI ‐1.31 to 0.03) was not statistically significant. Six studies of 425 participants were entered into a study of the effects of CBTs on disease‐specific quality of life. The overall effect of CBTs on disease‐specific quality of life of ‐0.19 (95% CI ‐0.58 to 0.21) was not statistically significant (see summary of findings Table 2).

Subgroup analyses

Different types of CBTs

a. End of treatment

There were no statistically significant subgroup differences for the outcomes pain (Chi2 = 0.69, P = 0.71) (Analysis 1.1), negative mood (Chi2 = 5.17, P = 0.08) (Analysis 1.2), disability (Chi2 = 1.45, P = 0.48) (Analysis 1.3) and pain self efficacy (Chi2 = 2.11, P = 0.35) (Analysis 1.4). The effect sizes of operant therapy and self management education on these outcomes were not statistically significant. The effect size of operant therapy on pain was ‐0.67 (95% CI ‐2.56 to 1.23; P = 0.49) and of self management education was 0.02 (95% CI ‐0.81 to 0.84; P = 0.97). Operant therapy had a small effect size on disability of ‐0.30 (95% CI ‐0.44 to ‐0.17; P < 0.0001) (Analysis 1.3). The effect size of operant therapy on negative mood was ‐0.90 (95% CI ‐2.12 to 0.42; P = 0.18) and of self management education was ‐0.10 (95% CI ‐0.27 to 0.07; P = 0.26) (Analysis 1.2). The effect size of operant therapy on disability was ‐0.88 (95% CI ‐3.10 to 1.33); P = 0.43) and of self management education was 0.07 (95% CI ‐0.60 to 0.74; P = 0.06) (Analysis 1.3). The effect size of operant therapy on pain self efficacy was ‐1.18 (95% CI ‐3.01 to 0.64; P = 0.20) and of self management education was ‐0.18 (95% CI ‐0.39 to 0.04; P = 0.11) (Analysis 1.4).

b. Long‐term follow‐up

There was no statistically significant group difference between traditional CBT and operant therapy on pain (Chi2 = 3.46, P = 0.06) (Analysis 2.1) but there was on negative mood (Chi2 = 6.9, P = 0.0009) (Analysis 2.2), disability (Chi2 = 12.34, P = 0.0004) (Analysis 1.3), pain self efficacy (Chi2 = 3.71, P = 0.05) (Analysis 1.4), fatigue (Chi2 = 3.89, P = 0.05) (Analysis 2.5) and sleep problems (Chi2 = 7.03, P = 0.008) (Analysis 2.6) favouring operant therapy; and in HRQOL (Chi2 = 3.92, P = 0.05) Analysis 2.7) favouring traditional CBT. Of note, the effect sizes of operant therapy on pain (‐1.27; 95% CI ‐2.30 to ‐0.24; P = 0.02), negative mood (‐1.28; 95% CI ‐1.97 to ‐0.49; P = 0.003), disability (‐1.68; 95% CI ‐2.40 to ‐0.96; P < 0.0001) and pain self efficacy (‐1.02; 95% CI ‐1.59 to ‐0.46; P = 0.02) were large and statistically significant. The effect size of traditional CBT on pain of ‐0.28 (95% CI ‐0.43 to ‐0.14; P = 0.001) was small and on negative mood of (‐0.28; 95% CI ‐0.58 to 0.02; P = 0.07) was not statistically significant; on disability (‐0.32; 95% CI ‐0.55 to ‐0.09; P = 0.007) the effect size was small and statistically significant. No study with self management education performed long‐term follow‐up.

Types of controls

The effect sizes of CBTs on pain, negative mood and disability at end of treatment and at long‐term follow‐up compared to active controls were not statistically significant. The effect sizes of CBTs on pain, negative mood and disability at end of treatment compared to attention controls were only statistically significant for negative mood, but not for pain and disability at end of therapy. The effects sizes of CBTs on pain, disability and negative mood compared to treatment as usual and waiting list control at end of treatment were small and statistically significant. The effect sizes of CBTs on pain and on disability at long‐term follow‐up compared treatment as usual or waiting list control were small and statistically significant, the one on negative mood was not statistically significant (Appendix 2).

Type of delivery of treatment

The effect sizes of internet or telephone therapy CBTs on pain, negative mood and disability at end of treatment were not statistically significant. The effect sizes of face‐to‐face CBTs on pain, negative mood and disability at end of treatment were small and statistically significant (Appendix 3).

Age of study participants

The effect size of traditional CBT on pain at end of treatment in children and adolescents was small and statistically significant. The effect sizes on negative mood and disability at end of treatment were not statistically significant. The effect sizes of traditional CBT on pain, negative mood and disability in adults at end of treatment were small and statistically significant (Appendix 4).

Treatment duration

Two studies which did not report treatment duration (Williams 2010; Woolfolk 2012) were excluded from this analysis. One study (Ang 2010) with < five hours was labelled 'ultra‐short term' CBTs. This study had no statistically significant effects on pain and disability at end of treatment. Studies with > 25 hours (Castel 2012; Falcao 2008; Thieme 2006; Wigers 1996) and > 50 hours (Soares 2002; Thieme 2003) were included in a group labelled 'long‐term' CBTs. At end of treatment, the effect sizes of long‐term CBTs on pain and disability were not statistically significant, the effect size on negative mood was moderate and statistically significant. The remaining studies with a study intensity of 5 to 25 hours were included in a group labelled 'short‐term' CBTs. At end of treatment, the effect sizes of short‐term CBTs on negative mood and disability were small and statistically significant, the effect size on pain was not statistically significant (Appendix 5).

Reported treatment quality

The effect sizes of CBTs with low reported treatment quality on pain, negative mood and disability at end of treatment were not statistically significant. The effect sizes of CBTs with moderate reported treatment quality on pain and negative mood at end of treatment were small and statistically significant, the effect size on disability was not statistically significant. The effect sizes of CBTs with high reported treatment quality on pain and negative mood at end of treatment were small and statistically significant, the effect size on disability was not statistically significant (Appendix 6).

Sensitivity analyses

Removing studies with data substituted or extracted from figures, with selection bias, with attrition bias, with reporting bias, without ITT, and with exclusion of patients with depressive or anxiety disorders did not change the magnitude and significance of the effect sizes of CBTs on pain, negative mood and disability at end of treatment; except that the effect size of CBTs on disability at end of treatment was not statistically significant after excluding studies with data extracted from figures or substituted values. Moreover, the effect size of CBTs on pain and disability was not statistically significant in studies after excluding studies which did not include patients with depressive or anxiety disorders, or both. Removing studies with < 20 participants per treatment arm did not change the magnitude and significance of the effect sizes of CBTs on pain, negative mood and disability at end of treatment (Appendix 7).

Discussion

Summary of main results

Low quality evidence from 20 trials (1382 patients with FM) indicates that CBTs provided a small decrease in pain at the end of about 12 weeks treatment, and from 14 trials (822 patients with FM) that CBTs provided a small decrease in pain at the end of about six months follow‐up. Low quality evidence from 18 trials (1578 patients with FM) indicates that CBTs provided a small decrease in negative mood at the end of about 12 weeks treatment, and from 11 trials (721 patients with FM) that CBTs provided a small decrease of negative mood at the end of about six months follow‐up. Low quality evidence from 15 trials (1163 patients with FM) indicates that CBTs provided a small decrease in disability at the end of about 12 weeks treatment, and from nine trials (664 patients with FM) that CBTs provided a moderate decrease in disability at the end of about six months follow‐up (summary of findings Table for the main comparison). Subgroup analyses demonstrated that positive effects of CBTs were only detectable for CBT at end of treatment and at long‐term follow‐up and for operant therapy at follow‐up, but not for operant therapy at end of treatment and not for self management education programs at end of treatment. Positive effects were only verifiable for face‐to‐face CBTs but not for internet‐based and telephone‐based CBTs at end of treatment. Positive effects were only traceable in the comparison of CBTs with treatment as usual and waiting lists controls but not with other active treatments (for example aerobic exercise) or with attention control (except negative mood) at end of treatment. Studies which included patients with anxiety and depression disorders exerted only a reduction of negative mood but not of pain and disability at end of treatment.

Overall completeness and applicability of evidence

The tests conducted were not indicative of a publication bias. Nevertheless, we cannot rule out the possibility that negative study results with CBTs had not been published or had been missed by our search strategy.

The applicability (external validity) of evidence is strong for the following reasons. 1. The studies were not only performed in university centres but also in primary and secondary care. 2. Patients with anxiety or depressive disorder, or both, which are frequently associated with FM were included in some studies. 3. Subgroup analysis demonstrated the efficacy of CBTs in children and adolescents and in adults. However, the majority of the patients were middle‐aged Caucasian women, making it difficult to apply the results to the total FM population, especially to male and non‐Caucasian patients. No study performed a subgroup analysis for male and non‐Caucasian patients. In summary, the external validity of CBTs studies in FM is much stronger than in the phase III clinical trials for drugs approved for FM treatment by the Food and Drug Administration (duloxetine, milnacipran and pregabalin). These studies were conducted only in research centres and excluded patients with anxiety and depressive disorder, except the duloxetine studies (Sommer 2012b).

Quality of the evidence

The quality of evidence of this review was based on the data presented in peer reviewed journals and some additional details which were provided on request by the study authors. The overall quality of evidence for the primary outcomes in the majority of studies was low (summary of findings Table for the main comparison) due to studies without or with unclear randomisation, allocation concealment, ITT analysis and selective reporting. However, sensitivity analyses demonstrated that the study results were robust against these risks of bias.

Potential biases in the review process

We cannot be certain that other studies that have not been published (with positive or negative results) were not identified.

We might have underestimated the methodological and treatment quality of some studies which might not have reported some details required for the treatment quality score used. We relied on the reported data for quality assessment and did not ask authors for further details because we did not want to introduce a 'response' bias. We have experienced in previous reviews on psychological therapies that it was impossible to get any details, on request, of studies conducted before 2005 (Bernardy 2010; Bernardy 2011). Even some authors of studies conducted after 2010 did not respond to our requests for outcomes not reported for this review.

Efficacy outcomes were analysed by some studies using last observation carried forward to impute for missing data. This procedure may lead to an overestimation of efficacy. However, the recommended baseline observation carried forward method (Moore 2010b) has been used by some other studies.

We had to calculate missing values by established imputation methods and to extract data from figures, for one study each. Excluding these studies from analyses did not change the significance and magnitude of the effect sizes.

The results of subgroup analyses are not robust due to the small number of studies in the operant and self‐management group and in the internet/telephone‐based groups.

Agreements and disagreements with other studies or reviews

The results of this review were more favourable for CBTs than previous reviews on CBTs in FM which searched the literature to June 2009 (Bernardy 2010) and December 2010 (Köllner 2012). By including new studies with large sample sizes and positive results into this review, positive effects of CBTs on pain, fatigue and disability could be demonstrated that were not detectable in our previous reviews (Bernardy 2010; Köllner 2012). Positive effects of CBTs on depression and self efficacy pain in FM patients were demonstrated in the previous reviews (Bernardy 2010; Köllner 2012)..

A Cochrane review on psychological therapies in chronic pain syndromes, except headache, in adults searched the literature until September 2011 (Williams 2012). The authors found that CBTs had small positive effects on disability, but not on pain or mood, when compared with active controls. CBTs had small to moderate effects on pain, disability and mood at end of treatment when compared with treatment as usual or waiting list. All except a small effect on mood disappeared at follow‐up. Subgroup analyses stratified according to the type of pain syndrome were not conducted. Our results are mainly in line with this review (Williams 2012): in FM, CBTs were only superior with small effect sizes in reducing pain, negative mood and disability at the end of treatment when compared to treatment as usual and waiting list but not when compared to attention controls (except for negative mood) and to other active therapies such as exercise.

A Cochrane review on psychological therapies in chronic pain syndromes in children and adolescents searched the literature until August 2008 (Eccleston 2012). The authors included one study in FM (Kashikar‐Zuck 2005). They found a significant and large effect of CBTs compared to controls in non‐headache treatments on pain at end of treatment and at long‐term follow‐up (Eccleston 2012). The effects of the two studies of CBTs in FM children and adolescents compared to controls were small and statistically significant on pain at end of treatment but were not statistically significant on negative mood and disability.

Of note, the effects of operant therapy compared to active controls in FM were not statistically significant at end of treatment, but they were large and statistically significant at long‐term follow‐up in this review. This time‐dependent efficacy of operant therapy in FM has not been sufficiently outlined by previous reviews on CBTs in FM (Bernardy 2010)).

Study flow diagram.
Figures and Tables -
Figure 1

Study flow diagram.

Navigate to figure in ReviewOpen in new tab

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Figures and Tables -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Navigate to figure in ReviewOpen in new tab

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Figures and Tables -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.1 Pain.
Figures and Tables -
Figure 4

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.1 Pain.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.2 Negative mood.
Figures and Tables -
Figure 5

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.2 Negative mood.

Navigate to figure in ReviewOpen in new tab

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.3 Disability.
Figures and Tables -
Figure 6

Forest plot of comparison: 1 Cognitive behavioural therapies versus controls at end of treatment, outcome: 1.3 Disability.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 1 Pain.
Figures and Tables -
Analysis 1.1

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 1 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 2 Negative mood.
Figures and Tables -
Analysis 1.2

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 2 Negative mood.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 3 Disability.
Figures and Tables -
Analysis 1.3

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 3 Disability.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 4 Self‐efficacy pain.
Figures and Tables -
Analysis 1.4

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 4 Self‐efficacy pain.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 5 Acceptability.
Figures and Tables -
Analysis 1.5

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 5 Acceptability.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 6 Fatigue.
Figures and Tables -
Analysis 1.6

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 6 Fatigue.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 7 Sleep problems.
Figures and Tables -
Analysis 1.7

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 7 Sleep problems.

Navigate to figure in ReviewOpen in new tab

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 8 Health‐related quality of life.
Figures and Tables -
Analysis 1.8

Comparison 1 Cognitive behavioural therapies versus controls at end of treatment, Outcome 8 Health‐related quality of life.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 1 Pain.
Figures and Tables -
Analysis 2.1

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 1 Pain.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 2 Negative mood.
Figures and Tables -
Analysis 2.2

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 2 Negative mood.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 3 Disability.
Figures and Tables -
Analysis 2.3

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 3 Disability.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 4 Self‐efficacy pain.
Figures and Tables -
Analysis 2.4

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 4 Self‐efficacy pain.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 5 Fatigue.
Figures and Tables -
Analysis 2.5

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 5 Fatigue.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 6 Sleep problems.
Figures and Tables -
Analysis 2.6

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 6 Sleep problems.

Navigate to figure in ReviewOpen in new tab

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 7 Health‐related quality of life.
Figures and Tables -
Analysis 2.7

Comparison 2 Cognitive behavioural therapies versus controls at long‐term follow‐up, Outcome 7 Health‐related quality of life.

Navigate to figure in ReviewOpen in new tab

Cognitive behavioural therapies compared to controls for fibromyalgia

Patient or population: Patients with fibromyalgia

Settings: In‐ and outpatient

Intervention: Cognitive behavioural therapies

Comparison: Controls (attention control, treatment as usual, waiting list, other active therapy)

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

controls

Cognitive behavioural therapies

Pain

end of treatment

(0‐10 scale)

Higher scores indicate higher pain levels

Mean pain baseline

7.37 (SD 2.10) 3

The mean pain in the intervention groups was 0.29 standard deviations lower (0.49 to 0.11 lower)

1382
(20)

⊕⊕⊝⊝
low1,2

SMD ‐0.29 (95% CI ‐0.47 to ‐0.11)

8.5% (95% CI 3.1% to 14.0%) relative improvement

6.3 % (95% 2.3% to 10.3%) CI) fewer points on the scale (absolute change)

NNTB 7 (95% CI 5 to19)

Pain
follow‐up median 6 months

(0‐10 scale)

Higher scores indicate higher pain levels

Mean pain baseline

64.72 (SD 10.44) 4

The mean pain in the intervention groups was 0.40 standard deviations lower (0.64 to 0.16 lower)

822
(14)

⊕⊕⊝⊝
low1,2

SMD ‐0.40 (95% CI ‐0.62 to ‐0.17)

6.4% (95% CI 2.7% to 9.9%) relative improvement

4.2% (95% CI 1.8% to 6.5%) fewer points on the scale (absolute change)

NNTB 10 (95% CI 6 to 24)

Negative mood

end of treatment

(0‐10 scale)

Higher scores indicate higher negative mood levels

Mean depression

baseline 6.82 (SD 3.11) 5

The mean negative mood in the intervention groups was 0.33 standard deviations lower (0.49 to 0.17 lower)

1578
(18)

⊕⊕⊝⊝
low1,2

SMD ‐0.33 (95% CI ‐0.49 to ‐0.17)

15.0% (95% CI 7.7% to 22.3%) relative improvement

10.2% (95% CI 5.2% to 15.2%) fewer points on the scale (absolute change)

NNTB 6 (95% CI 4 to12)

Negative mood
follow‐up median 6 months

(0‐50 scale)

Higher scores indicate higher negative mood levels

Mean depression

baseline 14.94 (SD 3.11) 6

The mean negative mood in the intervention groups was 0.43 standard deviations lower (0.75 to 0.11 lower)

721
(11)

⊕⊕⊝⊝
low1,2

SMD ‐0.43 (95% CI ‐0.75 to ‐0.11)

8.9% (95% CI 2.3% to 15.8%) relative improvement

2.7% (95% CI 0.1% to 4.7%) fewer points on the scale (absolute change)

NNTB 11 (95% CI 6 to 43)

Disability

end of treatment

(0‐10 scale)

Higher scores indicate disability levels

Mean physical impairment baseline 2.80 (SD 2.40) 7

The mean disability in the intervention groups was 0.30 standard deviations lower (0.51 to 0.08 lower)

1163
(15)

⊕⊕⊝⊝
low1,2

SMD ‐0.30 (95% CI ‐0.51 to ‐0.08)

25.8 % (95% CI 6.9% to 43.7% relative improvement

7.2% (95% CI 1.9% to 12.2%) fewer points on the scale (absolute change)

NNTB 7 (95% CI 4 to 26)

Disability
follow‐up median 6 months

(0‐10 scale)

Higher scores indicate disability levels

Mean physical impairment baseline 3.24 (SD 2.26) 8

The mean disability in the intervention groups was 0.52 standard deviations lower (0.86 to 0.18 lower)

664
(9)

⊕⊕⊝⊝
low1,2

SMD ‐0.52 (95% CI ‐0.86 to ‐0.18)

36.4% (95% CI 1.3% to 60.2%) relative improvement

11.7% (95% CI 4.1% to 19.4%) fewer points on the scale (absolute change)

NNTB 4 (95% CI 3 to12)

Acceptability

end of treatment (dropouts from study due to any reasons)

136 (94 to 195) per 1000

127 (88 to 182)

RR 0.94 (0.65 to 1.35)

1914 (21)

⊕⊕⊝⊝
low1

Absolute risk difference
0% (95% CI ‐1 to 0)

Relative per cent change
6% (95% CI 15%
improvement to 35%
worsening)

Not statistically significant

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Some studies with lack of reported allocation concealment, without intention‐to‐treat analysis and with selective outcome reporting

2 High heterogeneity of treatment effect

3 Luciano 2011: N=216 patients; Pain VAS 0‐10 scale

4 Alda 2011: N=113 patients; Pain VAS 0‐100 scale

5 Luciano 2011: N=216 patients; Depression VAS 0‐10 scale

6 Alda 2011; N=113 patients; Hamilton Rating Scale for Depression (0‐50)

7 Luciano 2011: N=216 patients; Physical impairment VAS 0‐10 scale

8 Alda 2011; N=113 patients; Physical impairment VAS 0‐10 scale

Figures and Tables -
Navigate to table in Review
Summary of findings 2. Cognitive behavioural therapies versus controls for fibromyalgia

Cognitive behavioural therapies versus controls for fibromyalgia

Patient or population: Patients with fibromyalgia

Settings: In‐ and outpatients

Intervention: Cognitive behavioural therapies

Comparison: Controls (attention control, treatment as usual, waiting list, other active therapy)

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Control

Cognitive behavioural therapies versus controls final treatment

Fatigue

end of treatment

(0‐10 scale)

Higher scores indicate higher fatigue levels

Mean fatigue score 8.13 (SD 1.89) 3

The mean fatigue in the intervention groups was
0.25 standard deviations lower
(0.49 to 0.02 lower)

910
(11 studies)

⊕⊕⊝⊝
low1

SMD ‐0.25 (95% CI ‐0.49 to ‐0.02)

5.8% (95% CI 0.05% to 11.3%) relative improvement

4.7% (95% CI 0.4% to 9.3%) fewer points on the scale (absolute change)

NNTB 9 (95% CI 5 to109)

Sleep problems

end of treatment

(0‐50 scale)

Higher scores indicate more sleep problems

Mean sleep problems score 27.9 (SD 8.8) 4

The mean sleep problems in the intervention groups was
0.4 standard deviations lower
(0.85 lower to 0.05 higher)

422
(8 studies)

⊕⊕⊝⊝
low2

SMD ‐0.40 (95% CI ‐0.85 to 0.05)

0.3% (95% CI ‐0.03% to 1.7%) relative improvement

7.0% (95% CI ‐0.90% to 15.0%) fewer points on the scale (absolute change)

NNTB 5 (95% CI ‐45 to 3)

Health‐related

quality of life

end of treatment

(0‐80 scale)

Higher scores indicate lower health‐related quality of life

Mean health‐related quality of life score 55.97 (SD 15.95) 5

The mean health‐related quality of life in the intervention groups was
0.23 standard deviations lower
(0.38 to 0.08 lower)

1238
(13 studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.23 (95% CI ‐0.38 to ‐0.08)

0.08% (95% CI 0.03% to 0.13%) relative improvement

4.6% (95% CI 1.6% to 7.6%) fewer points on the scale (absolute change)

NNTB 9 (95% CI 6 to 27)

Fatigue

Follow‐up median 6 months

(0‐10 scale)

Higher scores indicate higher fatigue levels

Mean fatigue score

Mean 8.32 (SD 2.17)6

The mean fatigue in the intervention groups was
0.46 standard deviations lower
(0.77 to 0.15 lower)

429
(6 studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.46 (95% CI ‐0.77 to ‐0.15)

1.2% (95% CI 0.4% to 2.0%) relative improvement

10.0% (95% CI 3.2% to 16.7%) fewer points on the scale (absolute change)

NNTB 5 (95% CI 3 to 14)

Sleep problems
Follow‐up median 6 months

(0‐50 scale)

Higher scores indicate more sleep problems

Mean sleep problems score 27.9 (SD 8.8) 4

The mean sleep problems in the intervention groups was
0.64 standard deviations lower
(1.31 lower to 0.03 higher)

378
(7 studies)

⊕⊕⊝⊝
low1,2

SMD ‐0.64 (95% CI ‐1.31 to 0.03)

0.4% (95% CI ‐0.02% to 0.8%) relative improvement

11.2% (95% CI ‐0.53% to 23.1%) fewer points on the scale (absolute change)

NNTB 4 (95% CI ‐74 to 2)

Health‐related

quality of life

Follow‐up median 6 months

(0‐80 scale)

Higher scores indicate lower health‐related quality of life

Mean health‐related quality if life score 64.48

(SD 10.50) 7

The mean health‐related quality of life in the intervention groups was
0.19 standard deviations lower
(0.58 lower to 0.21 higher)

425
(6 studies)

⊕⊕⊝⊝
low1

SMD ‐0.19 (95% CI ‐0.58 to 0.21)

0.03% (95% CI ‐0.03% to 0.15%) relative improvement

2.0% (95% CI ‐2.2% to 6.1%) fewer points on the scale (absolute change)

NNTB 12 (95% CI ‐17 to 6)

Acceptability
Follow‐up: median 6 months

See comment

See comment

Not estimable

‐

See comment

Not assessed

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Some studies with lack of reported allocation concealment, without intention‐to‐treat analysis and with selective reporting
2 High heterogeneity of treatment effect

3 Luciano 2011: N=216 patients; VAS 0‐10 scale

4 Castel 2012; N=60 patients; NRS 0‐50 scale

5 Luciano 2011: N=216 patients; VAS 0‐80 scale

6 Alda 2010; N= 113; VAS 0‐10 scale

7 Alda 2010; N= 113; VAS 0‐100 scale

Figures and Tables -
Summary of findings 2. Cognitive behavioural therapies versus controls for fibromyalgia
Navigate to table in Review
Table 1. Main characteristics of included studies

Author

Country

Type of CBT

Type of control group

Duration CBT (weeks)

Number of CBT sessions

Total treatment time CBT (hours)

Number of patients in CBT group

% women

Number of patients in control group

% women

Alda 2011 *

Spain

CBT

TAU

12

6

15

57

95

56

96

Ang 2010 *

USA

CBT

TAU

12

6

3

17

100

15

100

Burckhardt 1994 *

Sweden

CBT

Delayed treatment

6

6

9

28

100

30

100

Castel 2009 *

Spain

CBT

TAU

11

12

18

18

94

12

86

Castel 2012 *

Spain

CBT

TAU

14

14

28

34

94

30

100

Edinger 2005 *

USA

CBT

TAU

6

6

6

16

94

12

100

Falcao 2008 *

Brazil

CBT

TAU

10

20

30

30

100

30

100

Kashikar‐Zuck 2005 **

USA

CBT

Active control

8

8

12

14

100

14

100

Kashikar‐Zuck 2012 **

USA

CBT

Active control

8

8

6

57

95

57

90

King 2002 *

USA

CBT

Delayed treatment

12

12

18

48

100

39

100

Luciano 2011 *

Spain

CBT

TAU

8

8

16

108

95

108

98

Miro 2011 *

Spain

CBT

Active control

6

6

9

20

100

20

100

Nicassio 1997 *

USA

CBT

Active control

10

10

15

36

89

35

89

Oliver 2002 *

USA

Self‐management

TAU

52

10

20

207

96

193

94

Redondo 2004 *

Spain

CBT

Active control

8

8

20

21

100

19

100

Rooks 2007 *

USA

Self‐management

Active control

16

8

16

51

100

50

100

Soares 2002 *

Sweden

CBT

Attention control

10

10

120

18

100

18

100

Thieme 2003 *

Germany

Operant therapy

Active control

5

25

75

42

100

21

100

Thieme 2006a *

Germany

Operant therapy

Attention control

15

15

30

42

100

20

100

Thieme 2006b *

Germany

CBT

Attention control

15

15

30

43

100

20

100

Vlayen 1996 *

Netherlands

CBT

Active control

6

12

18

49

93

43

82

Wigers 1996 *

Norway

CBT

TAU

14

15

30

20

90

20

95

Williams 2010 *

USA

Self‐management

TAU

26

NR

59

95

59

95

Woolfolk 2012 *

USA

CBT

TAU

NR

NR

38

89

38

87

* Studies included only adults  

** Studies included only children and adolescents

NR = Not reported and not provided on request

TAU = Treatment as usual
Figures and Tables -
Table 1. Main characteristics of included studies
Navigate to table in Review
Table 2. Reported treatment quality

 

Treatment content and setting

Treatment duration

Manualisation

Adherence to manual

Therapist training

Client engagement

Sum

Alda 2011

2

1

2

1

1

0

7

Ang 2010

2

1

2

1

2

0

8

Burckhardt 1994

1

1

0

0

1

0

3

Castel 2009

1

1

0

0

0

0

2

Castel 2012

1

1

2

1

0

1

6

Edinger 2005

1

1

2

0

1

0

5

Falcao 2008

1

1

0

0

1

 

1

4

Kashikar‐Zuck 2005

2

1

2

1

2

1

9

Kashikar‐Zuck 2012

2

1

2

1

2

1

9

King 2002

1

1

0

0

0

1

3

Luciano 2011

2

1

1

0

1

0

5

Miro 2011

2

1

2

0

1

0

6

Nicassio 1997

1

1

0

0

0

0

2

Oliver 2002

2

1

0

0

0

0

3

Redondo 2004

2

1

0

0

0

0

3

Rooks 2007

2

1

0

0

0

0

3

Soares 2002

1

1

0

0

1

0

3

Thieme 2003

2

1

2

0

0

0

5

Thieme 2006

2

1

1

0

2

0

6

Vlayen 1996

1

1

0

0

0

0

2

Wigers 1996

1

1

0

0

1

1

4

Williams 2010

2

1

1

0

1

1

6

Woolfolk 2012

1

1

2

0

0

0

4

Items and scores of treatment quality scale (Yates 2005)

1. Treatment content and setting: 2 ‐ Adequate: a clear rationale for the treatment has been reported along with an adequate description of its content; 1 ‐ Partial: either a clear rationale or a description of the content of the treatment is reported; 0 ‐ Inadequate:neither the rationale for treatment or the treatment content are adequately reported.

2. Treatment duration: 1 – Reported; 0 ‐ Unknown.

3. Manualistion of treatment: 2 ‐ Adequate: there is reference to use of a manual that describes the active components of the treatment of study. If more than one treatment arm, manuals were used for all the appropriate treatments; 1 ‐ Partial:in trials with more than one treatment arm, the use of a manual is described but not for all the treatments that would be expected to be manualised; 0 ‐ Inadequate: no evidence that a manual has been used, but reference is made to various principles.

4. Adherence to the manual: 1 ‐ Adequate: there is evidence that the investigators have checked adherence to the manual during the period of study via direct observations, tape recording or supervisory processes that explicitly state adherence to the manual; 0 ‐ Inadequate: no evidence of adherence checks reported.

5. Therapist training: 2 ‐ Adequate: there is documentation of explicit training for the treatment of the trial; 1 ‐ Partial: the general level of therapist training is reported and is adequate (professionally qualified) but there is no mention of explicit training for the trial; 0 ‐ Inadequate: there is no convincing evidence that the therapists have an adequate level of training (e.g. graduate level) or explicit training for the trial.

6. Client Engagement: 1 ‐ Adequate: documented that evidence of engagement was sought e.g. checks on homework were made, skills practice in sessions; 0 – inadequate: no evidence that checks were made on level of engagement.

Figures and Tables -
Table 2. Reported treatment quality
Navigate to table in Review
Comparison 1. Cognitive behavioural therapies versus controls at end of treatment

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Pain Show forest plot

21

1453

Std. Mean Difference (IV, Random, 95% CI)

‐0.29 [‐0.47, ‐0.11]

1.1 Traditional cognitive behavioural therapy

18

1150

Std. Mean Difference (IV, Random, 95% CI)

‐0.30 [‐0.44, ‐0.15]

1.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐0.67 [‐2.56, 1.23]

1.3 Self‐management

2

180

Std. Mean Difference (IV, Random, 95% CI)

0.02 [‐0.81, 0.84]

2 Negative mood Show forest plot

19

1649

Std. Mean Difference (IV, Random, 95% CI)

‐0.33 [‐0.49, ‐0.17]

2.1 Traditional cognitive‐behavioural therapy

15

1010

Std. Mean Difference (IV, Random, 95% CI)

‐0.34 [‐0.48, ‐0.19]

2.2 Operant therapy

2

124

Std. Mean Difference (IV, Random, 95% CI)

‐0.90 [‐2.21, 0.42]

2.3 Self‐management education

3

515

Std. Mean Difference (IV, Random, 95% CI)

‐0.10 [‐0.27, 0.07]

3 Disability Show forest plot

16

1234

Std. Mean Difference (IV, Random, 95% CI)

‐0.30 [‐0.51, ‐0.08]

3.1 Traditional cognitive‐behavioural therapy

13

931

Std. Mean Difference (IV, Random, 95% CI)

‐0.31 [‐0.45, ‐0.18]

3.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐0.88 [‐3.10, 1.33]

3.3 Self‐management education

2

180

Std. Mean Difference (IV, Random, 95% CI)

0.07 [‐0.60, 0.74]

4 Self‐efficacy pain Show forest plot

11

1047

Std. Mean Difference (IV, Random, 95% CI)

‐0.49 [‐0.80, ‐0.17]

4.1 Traditional cognitive‐behavioural therapy

9

589

Std. Mean Difference (IV, Random, 95% CI)

‐0.39 [‐0.72, ‐0.05]

4.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐1.18 [‐3.01, 0.64]

4.3 Self‐management education

1

335

Std. Mean Difference (IV, Random, 95% CI)

‐0.18 [‐0.39, 0.04]

5 Acceptability Show forest plot

21

1914

Risk Ratio (M‐H, Random, 95% CI)

0.94 [0.65, 1.35]

5.1 Traditional cognitive‐behavioural therapy

17

1169

Risk Ratio (M‐H, Random, 95% CI)

0.98 [0.65, 1.46]

5.2 Operant therapy

2

126

Risk Ratio (M‐H, Random, 95% CI)

0.43 [0.03, 7.34]

5.3 Self‐management education

3

619

Risk Ratio (M‐H, Random, 95% CI)

1.33 [0.81, 2.19]

6 Fatigue Show forest plot

11

910

Std. Mean Difference (IV, Random, 95% CI)

‐0.25 [‐0.49, ‐0.02]

6.1 Traditional cognitive‐behavioural therapy

9

667

Std. Mean Difference (IV, Random, 95% CI)

‐0.38 [‐0.65, ‐0.10]

6.2 Operant therapy

1

63

Std. Mean Difference (IV, Random, 95% CI)

0.09 [‐0.44, 0.62]

6.3 Self‐management education

2

180

Std. Mean Difference (IV, Random, 95% CI)

0.04 [‐0.33, 0.40]

7 Sleep problems Show forest plot

8

422

Std. Mean Difference (IV, Random, 95% CI)

‐0.40 [‐0.85, 0.05]

7.1 Traditional cognitive‐behavioural therapy

6

244

Std. Mean Difference (IV, Random, 95% CI)

‐0.50 [‐1.11, 0.11]

7.2 Operant therapy

1

60

Std. Mean Difference (IV, Random, 95% CI)

‐0.37 [‐0.91, 0.17]

7.3 Self‐management education

1

118

Std. Mean Difference (IV, Random, 95% CI)

0.09 [‐0.27, 0.45]

8 Health‐related quality of life Show forest plot

13

1238

Std. Mean Difference (IV, Random, 95% CI)

‐0.23 [‐0.38, ‐0.08]

8.1 Traditional cognitive‐behavioural therapy

11

778

Std. Mean Difference (IV, Random, 95% CI)

‐0.34 [‐0.49, ‐0.18]

8.2 Operant therapy

1

63

Std. Mean Difference (IV, Random, 95% CI)

0.19 [‐0.34, 0.72]

8.3 Self‐management education

2

397

Std. Mean Difference (IV, Random, 95% CI)

‐0.01 [‐0.36, 0.33]
Figures and Tables -
Comparison 1. Cognitive behavioural therapies versus controls at end of treatment
Navigate to table in Review
Comparison 2. Cognitive behavioural therapies versus controls at long‐term follow‐up

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Pain Show forest plot

14

893

Std. Mean Difference (IV, Random, 95% CI)

‐0.40 [‐0.62, ‐0.17]

1.1 Traditional cognitive‐behavioural therapy

13

770

Std. Mean Difference (IV, Random, 95% CI)

‐0.28 [‐0.43, ‐0.14]

1.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐1.27 [‐2.30, ‐0.24]

1.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

2 Negative mood Show forest plot

12

792

Std. Mean Difference (IV, Random, 95% CI)

‐0.43 [‐0.75, ‐0.11]

2.1 Traditional cognitive‐behavioural therapy

11

669

Std. Mean Difference (IV, Random, 95% CI)

‐0.28 [‐0.58, 0.02]

2.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐1.28 [‐1.97, ‐0.59]

2.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

3 Disability Show forest plot

10

735

Std. Mean Difference (IV, Random, 95% CI)

‐0.52 [‐0.86, ‐0.18]

3.1 Traditional cognitive‐behavioural therapy

9

612

Std. Mean Difference (IV, Random, 95% CI)

‐0.32 [‐0.55, ‐0.09]

3.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐1.68 [‐2.40, ‐0.96]

3.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

4 Self‐efficacy pain Show forest plot

9

617

Std. Mean Difference (IV, Random, 95% CI)

‐0.75 [‐1.27, ‐0.24]

4.1 Traditional cognitive‐behavioural therapy

8

494

Std. Mean Difference (IV, Random, 95% CI)

‐0.52 [‐1.04, ‐0.00]

4.2 Operant therapy

2

123

Std. Mean Difference (IV, Random, 95% CI)

‐1.69 [‐2.76, ‐0.62]

4.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

5 Fatigue Show forest plot

6

429

Std. Mean Difference (IV, Random, 95% CI)

‐0.46 [‐0.77, ‐0.15]

5.1 Traditional cognitive‐behavioural therapy

6

366

Std. Mean Difference (IV, Random, 95% CI)

‐0.38 [‐0.69, ‐0.07]

5.2 Operant therapy

1

63

Std. Mean Difference (IV, Random, 95% CI)

‐1.02 [‐1.59, ‐0.46]

5.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

6 Sleep problems Show forest plot

7

378

Std. Mean Difference (IV, Random, 95% CI)

‐0.64 [‐1.31, 0.03]

6.1 Traditional cognitive‐behavioural therapy

6

318

Std. Mean Difference (IV, Random, 95% CI)

‐0.46 [‐1.11, 0.19]

6.2 Operant therapy

1

60

Std. Mean Difference (IV, Random, 95% CI)

‐1.68 [‐2.30, ‐1.06]

6.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]

7 Health‐related quality of life Show forest plot

6

425

Std. Mean Difference (IV, Random, 95% CI)

‐0.19 [‐0.58, 0.21]

7.1 Traditional cognitive‐behavioural therapy

6

362

Std. Mean Difference (IV, Random, 95% CI)

‐0.28 [‐0.68, 0.11]

7.2 Operant therapy

1

63

Std. Mean Difference (IV, Random, 95% CI)

0.39 [‐0.15, 0.92]

7.3 Self‐management education

0

0

Std. Mean Difference (IV, Random, 95% CI)

0.0 [0.0, 0.0]
Figures and Tables -
Comparison 2. Cognitive behavioural therapies versus controls at long‐term follow‐up
Navigate to table in Review