This article has a correction

Please see: Ann Rheum Dis 2015;74:476

Ann Rheum Dis 67:536-541 doi:10.1136/ard.2007.071522
  • Extended report

EULAR evidence-based recommendations for the management of fibromyalgia syndrome

  1. S F Carville1,
  2. S Arendt-Nielsen2,
  3. H Bliddal3,
  4. F Blotman4,
  5. J C Branco5,
  6. D Buskila6,
  7. J A P Da Silva7,
  8. B Danneskiold-Samsøe3,
  9. F Dincer8,
  10. C Henriksson9,
  11. K G Henriksson10,
  12. E Kosek11,
  13. K Longley12,
  14. G M McCarthy13,
  15. S Perrot14,
  16. M Puszczewicz15,
  17. P Sarzi-Puttini16,
  18. A Silman17,
  19. M Späth18,
  20. E H Choy1
  1. 1
    Academic Rheumatology Unit, King’s College London, Weston Education Centre, Cutcombe Road, London SE5 9RJ, UK
  2. 2
    Aalborg University, Center for Sensory-Motor Interaction, Department of Health Sciences and Technology, Denmark
  3. 3
    The Parker Institute, Frederiksberg Hospital, Copenhagen, Denmark
  4. 4
    Rheumatology Department, Hospital Lapyeronie, Montpellier, France
  5. 5
    Serviço de Reumatologia, Hospital Egas Moniz, Centro Hospitalar Lisboa Ocidental, Lisbon, Portugal
  6. 6
    Department of Medicine, Soroka Medical Center, Beer Sheva, Israel
  7. 7
    Reumatologia, Hopitais Da Universidade, Coimbra, Portugal
  8. 8
    Department of Physical and Rehabilitation Medicine, Hacettepe Medical School, Ankara, Turkey
  9. 9
    INR, Section of Occupational Therapy, Faculty of Health Sciences, Linköping University, S-581 85 Linköping, Sweden
  10. 10
    Neuromuscular Unit, University Hospital, Linkoping
  11. 11
    Department of Clinical Neuroscience, Karolinska Institute, Stockholm, and Stockholm Spine Center, Löwenströmska Hospital, Upplans Väsby, Sweden
  12. 12
    FMA, Bath, UK
  13. 13
    Rheumatology, Mater Misericordiae University Hospital, Dublin, Ireland
  14. 14
    Pain Clinic, Hospital Cochin, Paris, France
  15. 15
    Department of Rheumatology, Rehabilitation and Internal Medicine University of Medical Sciences, Poznan, Poland
  16. 16
    Rheumatology Unit, L. Sacco University Hospital, Milan, Italy
  17. 17
    Faculty of Medical and Human Sciences ARC Epidemiology Unit, The University of Manchester, Manchester, UK
  18. 18
    Bahnhofstr. 95 82166 Graefelfing, Germany
  1. Dr Serena Carville, Sir Alfred Baring Garrod Clinical Trials Unit, Academic Rheumatology Unit, King’s College London, Weston Education Centre, Cutcombe Road, London SE5 9RJ, UK; serena.carville{at}
  • Accepted 5 July 2007
  • Published Online First 20 July 2007


Objective: To develop evidence-based recommendations for the management of fibromyalgia syndrome.

Methods: A multidisciplinary task force was formed representing 11 European countries. The design of the study, including search strategy, participants, interventions, outcome measures, data collection and analytical method, was defined at the outset. A systematic review was undertaken with the keywords “fibromyalgia”, “treatment or management” and “trial”. Studies were excluded if they did not utilise the American College of Rheumatology classification criteria, were not clinical trials, or included patients with chronic fatigue syndrome or myalgic encephalomyelitis. Primary outcome measures were change in pain assessed by visual analogue scale and fibromyalgia impact questionnaire. The quality of the studies was categorised based on randomisation, blinding and allocation concealment. Only the highest quality studies were used to base recommendations on. When there was insufficient evidence from the literature, a Delphi process was used to provide basis for recommendation.

Results: 146 studies were eligible for the review. 39 pharmacological intervention studies and 59 non-pharmacological were included in the final recommendation summary tables once those of a lower quality or with insufficient data were separated. The categories of treatment identified were antidepressants, analgesics, and “other pharmacological” and exercise, cognitive behavioural therapy, education, dietary interventions and “other non-pharmacological”. In many studies sample size was small and the quality of the study was insufficient for strong recommendations to be made.

Conclusions: Nine recommendations for the management of fibromyalgia syndrome were developed using a systematic review and expert consensus.

Fibromyalgia syndrome (FMS) is a common rheumatological condition characterised by chronic widespread pain and reduced pain threshold, with hyperalgesia and allodynia. Associated features include fatigue, depression, anxiety, sleep disturbance, headache, migraine, variable bowel habits, diffuse abdominal pain and urinary frequency.1 2 Although the precise pathogenesis remains unknown, peripheral and central hyperexcitability at spinal or brainstem level,35 altered pain perception6 and somatisation7 8 have been hypothesised and demonstrated in some patients.

The American College of Rheumatology (ACR) classification criteria for FMS9 are the most commonly used in clinical and therapeutic research. The healthcare utilisation by patients with FMS is high averaging over $2000 per patient per year,10 but it has been shown that positive diagnosis and management can reduce healthcare utilisation.11

Although effective treatments are available1214 no guidelines exist for the management of FMS. The objectives were to ascertain the strength of the research evidence on the effectiveness of treatment of FMS and develop recommendations for its management based on the best available evidence and expert opinion to inform healthcare professionals.



A multidisciplinary taskforce was formed consisting of 19 experts in FMS representing 11 European countries.

Search strategy

A systematic search of Medline, PubMed, EmBASE, PsycINFO, CINAHL, Web of Sciences, Science Citation Indices, Cochrane Central Register of Controlled Trials and Cochrane Database of Systematic Reviews using the keywords: “fibromyalgia”, “treatment or management” and “trial” for all publications till the end of December 2005 was carried out. A manual search of the bibliographies of trials was undertaken to verify that all published trials were identified.

Inclusion criteria

Included studies had to be clinical trials using the ACR 1990 classification criteria for FMS9 to select patients. Studies, including patients with chronic fatigue syndrome or myalgic encephalomyelitis, were excluded unless they were divided into separate comparator groups for analysis.

Assessment of literature

A “checklist” method15 was used to assess quality of each study. Data were tabulated using a customised data extraction form. This included number of patients in each arm, randomisation and blinding status. Previous reviews have identified two main outcome measures: pain assessed by the visual analogue scale (VAS) and function assessed by the fibromyalgia impact questionnaire (FIQ).16 17 The main measure of effect was the between-group difference calculated from the mean change between the pre- and post-treatment values in these outcome measures. Where possible, effect size for the “best” treatments in each category was calculated (averaged if there was more than one trial). Rosnow and Rosenthal’s modified version of the Cohen’s d calculation was used.18 The thresholds used for interpretation were: values >0.2 = small, >0.5 = medium and >0.8 = large. The number needed to harm (NNH) was also calculated if possible, using withdrawal due to adverse event as the event. Additional information included: recruitment population; duration of disease, treatment and assessment; number of tender points; and myalgic score. Other outcome measures considered were also tabulated. If required data were recorded, but not presented, or not presented in a suitable format, the author was contacted wherever possible. If data were only provided in graphical format, this was extracted where possible. Data extraction was verified by a second committee member to ensure accuracy. Any discrepancies were re-evaluated.

Categorising evidence

Owing to the large variability in outcome measures and assessments data could not be pooled to perform a formal meta-analysis; therefore studies were classified according to their randomisation and blinding level. The highest quality study (randomised controlled trial) for each treatment class was used as a basis for the recommendations. The ranking was graded as (with 1 being highest):

1. Randomised controlled double-blind trials

2. Randomised, blinded crossover trials

3. Randomised single blind trials

4. Randomised open trials/non-randomised single blind

5. Non-randomised open trials.

Evidence for each recommendation was categorised according to study design and strength of each recommendation was classified according to the criteria previously published.19

The recommendations were discussed at a final committee meeting and via email for a consensus to be reached. Delphi exercise was used to base recommendations on when limited evidence was found by systematic review. Agreement on the included recommendations was unanimous.

Publication bias analysis

Abstracts published between 2002 and 2005 inclusive in Annals of the Rheumatic Diseases, Pain, Arthritis & Rheumatism and Journal of Musculoskeletal Pain were reviewed to guard against non-inclusion of any negative studies that had not been fully published. If available, data were extracted. Any contradictory data would be included when forming the recommendations.

Future research plan

The committee proposed that these recommendations should be reviewed and updated in 4 years time, to see if (a) quality of trials and reporting in FMS had improved, and (b) if there was new evidence to suggest recommendation of new treatments, or to alter the recommendations of treatments already included.


Research evidence identified

In the preliminary search, 508 studies were identified. Tables 1 and 2 demonstrate how these were short-listed.

Table 1 Study breakdown from initial literature search
Table 2 Breakdown of the short-listed studies to base recommendations on, and those eliminated from further analysis

Sensitivity analysis

Effect size and NNH for the interventions recommended were calculated where possible (table 3).

Table 3 Effect size calculated using modified Cohen’s d method for recommended treatments where data available

EULAR recommendations

From tables 1–3 the following recommendations were made (table 4).

Table 4 EULAR recommendations for the management of fibromyalgia

Assessment of recommendations

There was no weighting in terms of order of the recommendations. √ denotes recommendation derived from expert opinion.

√ Full understanding of fibromyalgia requires comprehensive assessment of pain, function and psychosocial context. Fibromyalgia should be recognised as a complex and heterogeneous condition where there is abnormal pain processing and other secondary features.

This is based on expert opinion. It is important to recognise that FMS is a heterogeneous condition comprising a range of symptoms and features, effective management must take all of these factors into account. The nociceptive system also has connections with the stress regulating, immune and the sleep system in the limbic brain. It is these links that probably lead to the “syndrome” incorporating numerous symptoms and features.

√ Optimal treatment requires a multidisciplinary approach with a combination of non-pharmacological and pharmacological treatment modalities tailored according to pain intensity, function, associated features, such as depression, fatigue and sleep disturbance in discussion with the patient.

This is a logical progression from the first recommendation. It represents general practice, but is based solely on expert opinion. As FMS is polysymptomatic, lacking one treatment that acts on all symptoms, a multidisciplinary approach tailored to the needs of the individual is often required. This may need to include self-management via patient education.4749 Only two multidisciplinary trials were short-listed in the summary tables for further analysis.50 51 Other reviews have supported the use of multidisciplinary treatment,47 48 but highlighted the lack of high-quality trials in this area.48 52

Heated pool treatment with or without exercise is effective in fibromyalgia

Heated pool treatment or balneotherapy was reported to be effective in improving pain and function. Three of five trials included exercise in the intervention34 35 38 (two positive for function and two for pain). Of those without exercise, two were positive for pain and function.34 36 In the third trial only the heated pool treatment group improved in pain, but no comparison was made with the control. Function was not assessed.37 Drop-out for adverse events was very low. Sample sizes ranged from medium to large. Three of the studies restricted the use of medications (not stated in the remaining two). The fairly high quality of this small number of studies with positive results has led to this recommendation and there is agreement with previous reviews.47 48

√ Individually tailored exercise programmes, including aerobic exercise and strength training can be beneficial to some patients with fibromyalgia.

This is based largely on expert opinion with a combination of some experimental evidence and previous reports.

For aerobic exercise the majority of trials were open (seven of 11). The best quality were a randomised, assessor blind 12-week study by Richards and Scott, with large sample size,53 and a smaller randomised single blind study by Valim et al.42 Valim et al reported an improvement in VAS pain and FIQ compared with control. Richards and Scott did not report significant between-group improvements in either of our chosen outcome measures although the FIQ score did improve more in the treatment group, and significant between-group improvements were seen at 12 months follow-up. All three strength training studies were randomised but only one single blind. This had no significant between-group differences in pain or function, although both improved in the exercise group only.44

In general the quality of studies among exercise trials was considerably variable. Blinding and/or control was frequently inadequate. Those that did show some differences in favour of exercise used usual activity and care for their controls40 41 (with the exception of Valim et al who had a stretching control group42). The majority of exercise studies asked for participants not to change their medication intake while on the trial.9

Although evidence in the literature was poor, the committee felt that given the safety and benefit of exercise to general health exercise should be included as a recommendation. The poor quality of the trials and our predetermined outcome measures were likely precluding positive outcomes from being shown. In previous reviews, exercise has been recommended12 16 17 47 48 with aerobic exercise gaining the most support. It is likely that different forms of exercise would suit different subgroups of patients, hence these programmes should be tailored to the individual.

√ Cognitive behavioural therapy may be of benefit to some patients with fibromyalgia.

This is based on expert opinion. The only two studies identified for our review with pure cognitive behavioural therapy (CBT) were of poor quality; neither had a control group, both allowed patients to remain on their usual medication and only one used either of our predetermined outcome measures.

This is another area in which the poor quality of trials has masked what experts believe to be a realistic reflection of possible benefits. While previous review work has also been hampered by the inadequacy of research in this field, strong evidence has been reported for CBT with positive results for pain and function.47

√ Other therapies such as relaxation, rehabilitation, physiotherapy and psychological support may be used depending on the needs of the individual patient.

This is based on expert opinion and some experimental evidence. Two studies of moderate quality were identified for physiotherapy. An open study54 for connective tissue massage, which had larger subject numbers (25 control and 23 treated) and lasted 10 weeks, reported improvement in both pain and function compared with control. Other relaxation and rehabilitation techniques are recommended due to expert opinion. Clinical trial evidence is lacking in these areas, although reviews report some benefits.47

Tramadol is recommended for the management of pain in fibromyalgia

Simple analgesics such as paracetamol and other weak opioids can also be considered in the treatment of fibromyalgia. Corticosteroids and strong opioids are not recommended.

Regarding tramadol, two randomised controlled trials were identified as eligible for the review.30 31 One was a high-quality study of large sample size and 13 weeks duration.31 The second was preceded by an open label study and only included responders.30 Bennett et al reported positive effects for pain and function, and Russell et al reported improved pain levels but no change in function. There was no difference between placebo and treated group for adverse event withdrawals (high but non-serious). Bennett et al restricted concomitant medications, but Russell et al disallowed sedative hypnotics only. Tramadol should be used with some caution due to the possibility of typical opiate withdrawal symptoms with discontinuation and the risk of abuse and dependence.55

The recommendation for simple analgesics and other weak opioids is based mainly on expert opinion due to insufficient data.56

The negative recommendation for use of strong opioids and corticosteroids is based on expert opinion. These medications have significant long-term side-effects and no clinical trials were identified in FMS. Previous reviews support our recommendation.47 57

Antidepressants: amitriptyline, fluoxetine, duloxetine, milnacipran, moclobemide and pirlindole, reduce pain and often improve function therefore they should be considered for the treatment of fibromyalgia

Four of five trials of amitriptyline that assessed VAS pain had positive outcomes. Only two used the FIQ, one positive. However, it is important to note, that as highlighted in previous reviews,14 the only trial that lasted longer than 12 weeks did not show a significant improvement in pain compared with control.58 Two trials assessing fluoxetine reported positive outcomes for both pain and function.22 28 These trials were of moderate to high quality, reasonable samples sizes and 6 and 12 weeks duration. Duloxetine improved function in two trials and pain in one.25 27 The milnacipran trial reported an improvement in pain.26 These were all large, high-quality trials of 12 weeks duration. Moclobemid and pirlindole were assessed in one trial each, both of high quality and with improvements in pain.21 23 FIQ was not assessed in either trial. For all the trials withdrawals due to adverse events were generally low and non-serious.

In general these trials excluded other medications prescribed for FMS, with the exception of paracetamol. The only exception was the Arnold et al trial that also allowed NSAIDs.28 Previous reviews have agreed with the recommendation of antidepressants with the strongest evidence for amitriptyline (or tricyclic antidepressants).12 14 47 57

Tropisetron, pramipexole and pregabalin reduce pain and should be considered for the treatment of fibromyalgia

Two tropisetron clinical trials were eligible. One had positive results for pain at a dose of 5 mg.59 Späth et al did not report significantly positive results, but sample size was small and there was a positive trend in the treated group.32 FIQ was only assessed in the trial by Späth et al with negative results; therefore, no firm comment can be made on this outcome measure. Fäber et al made no comment on whether concomitant medications had been controlled, but Späth et al disallowed antidepressants, tranquillisers and sedatives. This treatment appears well tolerated. These were short-term studies, so further research into longer-term effects is required.

One trial for pramipexole was positive for both pain and function.33 Frequency of mild/moderate adverse events was high and this trial did not restrict concomitant medications, although dosages were kept stable. A monotherapy trial is required for more conclusive assessment of effect.

One trial reported pregabalin 450 mg reduced pain, but FIQ was not assessed.60 Drop-outs due to adverse events were largely classed mild to moderate in severity. All medications for pain and sleep disorders were restricted, with the exception of paracetamol.

These are recent studies and suggest further research into the use of these promising medications for FMS. Previous reviews have also mentioned their potential benefit47 57 (neither include the pramipexole study as this was not published).


These EULAR recommendations are based on expert opinion and changes in pain assessed by VAS and function assessed by the FIQ in clinical trials. Positive effects in other outcome measures were not considered, neither were pain or function if assessed by different instruments. Consequently some studies were excluded from our review due to not using these outcome measures, or not presenting the data. Although other instruments might be more sensitive in FMS it was decided that setting a standard for outcome measures was vital so that comparisons could be made fairly between trials and therefore using those most frequently reported allowed better analysis.47 61 Previous reviews have used different inclusion/exclusion criteria and/or assessed more or different outcome measures producing different evidence.16 47 48

The high variability in outcome measures used, reporting of results, as well as the inadequacy of methodological quality were barriers to conducting meta-analysis.12 14 16 17 57 62 This led to difficulties in producing strict evidence-based recommendations. In some areas evidence is lacking due to the poor quality of the studies, where expert opinion suggests otherwise, eg, exercise.

Outcome measures may be decided according to desired treatment effect. Non-pharmacological interventions have previously been suggested to have a significantly better effect on function than medications,62 reflected by its wider assessment in these studies. However, if this outcome measure is not frequently assessed in pharmacological trials, results could be biased.

Guidance on how to conduct good randomised controlled trials in FMS, including standardised outcome measures and validated, sensitive instruments is important for future research.

For the treatments that were recommended, effect sizes generally range from medium to high. Although these results give an indication of the efficacy of each treatment, they should be interpreted with some caution as they were only calculated where data were available and could be biased by factors such as whether or not the outcome measure was assessed. We have not collected any information on the cost-effectiveness of these treatments. Further analysis of disease duration and baseline values does not reveal any obvious pattern that would affect the outcomes of this review. Review of the abstracts published between 2002 and 2005 revealed no conflicting evidence to that derived from the published articles identified.

The assessment of strength of evidence tends to favour pharmacological studies as double blinding and placebo controls are impossible in many non-pharmacological studies. However, most non-pharmacological interventions are safe and have other health benefits. These important factors were taken into account in formulating these recommendations.


These recommendations are the first to be commissioned for FMS, although previous reviews have addressed the area.47 62 The standard operating procedures published by EULAR63 were followed. They will be updated every 5 years and it is hoped that good quality clinical trials in this area will add to the evidence currently available. These recommendations should assist healthcare providers, with a secondary intention to incorporate information into materials for patients.

The nine recommendations included eight management categories, three of which had strong evidence from the current literature, and three were based on expert opinion.


We acknowledge EULAR for their financial support and the ESCISIT steering committee for their endorsement.


  • Competing interests: FB has been reimbursed by Laboratoires Procter and Gamble, Sanofi-Aventis, Roche and Bristol Meyers Squibb for attending medical conferences and had honorarias for speaking for Laboratoires Pierre Fabre, Servier and Roche. JB has been paid by Pierre Fabre and Pfizer for running educational programmes and for speaking at international conferences and reimbursed by Eli Lilly for attending international conferences. DB has been reimbursed by Pierre Fabre Company, the manufacturer of Milnacipran, for attending several symposiums and by Pfizer for consulting. EC has served on advisory panels of Pierre Fabre Medicament, Jazz Pharmaceutical, Allergan and Pfizer. EC has also lectured in meetings organised by Pierre Fabre Medicament, Eli Lilly and Pfizer. The Rheumatology Department received a research grant from Pierre Fabre Medicament. CH has participated in symposia organised by Laboratoires Pierre Fabre and received reimbursement for participation. She has also received fees for written material in proceedings from these symposia. KH has participated in symposia organised by Laboratoires Pierre Fabre and received reimbursement for participation. He has also received fees for written material in proceedings from these symposia. He has held a lecture on pain mechanisms and received a fee from Pfizer. EK has participated as a consultant in advisory board meetings (total of four) for the following pharmaceutical companies: Pfizer, Wyeth and Pierre Fabre. She gave a speech on a satellite symposium organised by Pfizer. She has currently research collaboration with Pierre Fabre. SP has been paid by Pfizer, Eli Lilly, Grunenthal, Pierre Fabre and Sanofi Aventis for running educational programmes and participating in advisory boards. MS has served on advisory panels of Pierre Fabre Medicament, Jazz Pharmaceuticals and Allergan. MS has also lectured in meetings organised by Novartis, Pierre Fabre Medicament and Lilly.


Responses to this article

Free sample This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of ARD.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.