Skip navigation

Cognitive behaviour therapy for chronic fatigue

Clinical bottom line:

There is insufficient evidence of benefit with cognitive behavioural therapy when compared with routine medical care or relaxation. Most benefit appears to be in the reduction of fatigue, though there was no significant benefit of cognitive behavioural therapy over control for other outcomes. The results were based on limited patient information from small trials.

Chronic fatigue syndrome is characterised by persistent fatigue for which there is no evidence of psychological or physical disorders being involved. It can cause considerable disability and distress. It is thought to affect about 1% of the general population. Cognitive behaviour therapy may help through its use of rehabilitative and psychological approaches.

Systematic review

Price JR, Couper J. Cognitive behaviour therapy for adults with chronic fatigue syndrome (Cochrane review). In: The Cochrane Library, Issue 1, 2001. Oxford: Update Software.

Date review completed : Last search completed June 1998

Number of trials included : Three

Number of patients : 164 in total

Control group : orthodox medical management or another intervention.

Main outcomes : symptom score (patient or clinician rated), quality of life, use of health service resources.

Inclusion criteria were randomised, controlled trials which assessed cognitive behaviour therapy in adults with chronic fatigue syndrome (CFS) defined as medically unexplained fatigue of more than six months duration, and of sufficient severity to significantly distress or disable the patient. Cognitive behaviour therapy included psychological treatment to modify thoughts and beliefs about symptoms and illness, and attempted to modify the behavioural responses to symptoms and illness (e.g. sleep, activity). Databases searched included MEDLINE, EMBASE, SIGLE and PsychLit. Published and unpublished trials were sought and no language restrictions were made. The methodological quality of trials was assessed. Numbers-needed-to-treat (NNT) and odds ratios (OR) were calculated, with 95% confidence intervals, when possible. Effect size, with confidence intervals, was calculated for other parameters.


Three trials were included; all used Type A cognitive behaviour therapy which attempted to normalise rest time, symptoms and activity levels. Two studies used the stricter Oxford criteria for diagnosis of CFS and one used the Australian criteria. In two trials patients received about 15 hours worth of treatment, compared with six in the other. Patients were assessed over 4-6 months and length of follow-up varied between 3-7 months. Cognitive behaviour therapy included orthodox medical care (reassurance, advice to increase activity, primary care follow-up), primary care follow-up, and eight weekly placebo injections which may have included dialyzable leucocyte extract (a supposedly immunological treatment for CFS). Control interventions were relaxation, routine medical care, or placebo injections with routine medical care. Data from the different studies were not pooled.

The NNT to prevent one additional unsatisfactory outcome for physical functioning was 2.1 (95% confidence interval 1.5 to 4.0) for CBT compared with relaxation over 6-7 months in both trials which assessed this outcome. Fewer patients randomised to CBT were dissatisfied with their treatment or rated treatment as unsatisfactory than with relaxation.

Fatigue was assessed in all three trials, but using different descriptors. In one trial, CBT significantly reduced fatigue compared with relaxation; NNT 2.1 (1.5 to 4.0) for the assessor rating. Patient problem rating and patient questionnaire gave similar results. Another reported reduced fatigue with CBT compared with routine care. Information from the third trial was uninterpretable.

There was no significant difference in mood or anxiety with CBT compared with relaxation, or in anxiety compared with routine care. Significant improvement in quality of life (Work and Social Adjustment Scale) was shown with CBT compared with relaxation, SMD -1.06 (95% C.I. -1.6 to -0.5) in one trial.

In one trial, individuals randomised to CBT were significantly more satisfied with their treatment than those randomised to relaxation, NNT 2.7 (1.7 to 7.5), or routine medical care, NNT 2.7 (1.7 to 7.4).

Compliance was good, with 0-10% of patients withdrawing from the studies with either active or control interventions. Use of health service resources was similar between groups.

Adverse effects

Not mentioned.


There appears to be some benefit of using CBT rather than routine medical care or relaxation for some outcomes, but not others. Patients rated CBT as an acceptable treatment and few patients dropped out of the studies. The problem is that these studies were small, the largest had 30 patients per group, and bias could easily have occurred. The outcomes assessed differed between studies, so data could not be pooled and all results expressed were based on the results of these single small trials.