We conducted a six-month, multicentre, randomised, parallel group, controlled trial in which patients were randomly assigned to one of three study arms (ratio 1:1:1): CBT, RPT with pregabalin and an antidepressant (duloxetine) if there was comorbid depression and (3) TAU at the primary care level. Evaluators were blinded to participants' treatment group assignments. The protocol of this study has been previously published 30 . This trial followed the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials, or IMMPACT, recommendations for chronic pain clinical trials 31 and the Consolidated Standards of Reporting Trials, or CONSORT, recommendations for randomised, controlled trials 32 . No changes to the methods were made after the trial began.

Patients were recruited from any of the 41 primary healthcare centres in the city of Zaragoza, Spain. Zaragoza is the fifth-largest city in Spain, with a population of 713, 000. This study was carried out from January 2009 to June 2010. Participants were recruited from January to December 2009, and the six-month follow-up examinations were completed from January to June 2010. Patients were consecutively recruited by doctors working in primary care centres until the required sample size was attained, without a quota of patients assigned from each centre. Patients considered for inclusion were 18 to 65 years of age, able to understand and read Spanish, fulfilled the criteria for FM according to the American College of Rheumatology 18 , had undergone no psychological treatment during the preceding two years, were receiving no pharmacological treatment at that time or were willing to discontinue it for two weeks before the start of the study, and had signed an informed consent statement. Those excluded were patients with severe axis I psychiatric disorders (dementia, schizophrenia, paranoid disorder and alcohol and/or drug abuse); patients with severe axis II psychiatric disorders or other medical disorders that, from the clinician's point of view, prevented the patient from following the treatment protocol; women who were pregnant or nursing; and those who declined to participate.

The study personnel who carried out the measurements were kept blinded to which treatment each patient received. The assessments took place at baseline, posttreatment and after one, three and six months. Posttreatment assessment took place nine weeks after the baseline assessment for all groups, because this was the amount of time required for the CBT group to complete group therapy.

The primary objective of this study was to assess the efficacy of CBT, RPT and TAU for the treatment of PC in patients with FM in primary care settings. The major outcome of this study was PC in patients with FM. This construct was assessed using the Spanish version 39 of the PCS 6 . The PCS is a 13-item self-report questionnaire that comprises three dimensions: rumination, magnification and helplessness. There is no established 'cutoff' point, because PC is considered to be distributed in a continuous way in the general population. All items are rated on a Likert scale from 0 (not at all) to 4 (all the time). The total possible score ranges from 0 to 52, with a higher score indicating higher PC.

The secondary objectives of this study were to evaluate, in patients with FM, the efficacy of CBT, RPT and TAU in primary care for depression (measured using the Hamilton Rating Scale for Depression (HAM-D)); anxiety (assessed using the Hamilton Anxiety Rating Scale (HARS)); pain (measured using the Pain Visual Analogue Scale (PVAS)); global function (assessed using the Fibromyalgia Impact Questionnaire (FIQ)); and quality of life (assessed using the European Quality of Life Scale 5-D (EuroQol-5D) Questionnaire).

This study followed Helsinki Convention norms and later modifications and the Declaration of Madrid of the World Psychiatric Association. The study protocol was approved by the Ethical Review Board of Aragon. All patients provided their written informed consent before the commencement of any study activities or procedures.

A total of 218 patients were screened, and 49 were excluded (Figure 1). Of these 49 patients, 30 were ineligible because they did not meet the entry criteria, 16 decided not to participate, 3 were impossible to contact and 169 were enrolled. Of the patients enrolled, 57 were randomly assigned to the CBT group, 56 to the RPT group and 56 to the TAU group. All of them received the allocated intervention (N = 168), except one patient in the TAU group who moved to another city during the study period. A total of 141 patients (83.9%) completed the study, comprising 49 (85.9%) in the CBT group, 46 (82.1%) in the RPT group and 46 (83.6%) in the TAU group (Figure 1). The patients' reasons for discontinuation were as follows. In the CBT group, one patient (1.7%) did not complete the study due to lack of efficacy of the treatment, four patients (7%) due to a personal decision and three patients (5.2%) were lost to follow-up. In the RPT group, three patients (5.3%) withdrew due to adverse effects of the treatment (two due to digestive problems and the other due to dizziness), three patients (5.3%) due to a personal decision and four patients (7.1%) were lost to follow-up. In the TAU group, two patients (3.6%) withdrew due to adverse events, two patients (3.6%) due to a personal decision, three patients due to lack of efficacy (5.4%) and two patients (3.6%) were lost to follow-up. The three groups did not differ significantly with regard to the percentage of patients who completed the study or to the reasons for discontinuation.

The sample was mainly made up of middle-aged, married European females who lived with a spouse and/or offspring, had a primary or secondary school education, were unemployed or on sick leave and had a low or medium income level. From a clinical point of view, the patients, on average, had had FM for more than ten years. Half of them had been diagnosed with major depressive disorder, more than 10% had experienced sexual abuse and about 25% were currently engaged in litigation (Table 1). There were no statistically significant differences in baseline demographics or clinical characteristics between the three treatment groups, except for income, which was lower in the TAU group (Table 1). The questionnaire scores at baseline revealed that the participants had high PC, low pain acceptance, moderate depression, severe pain, limited global function and low quality of life (see Table 2). Moreover, the patients' anxiety levels were not clinically significant. There were significant differences (P < 0.05) between the groups with regard to acceptance and anxiety at baseline. Post hoc comparisons indicated that the CBT group had higher acceptance scores than the RPT and TAU groups (Cohen's d = 0.48 and 0.43, respectively). In addition, the RPT group had significantly higher anxiety than the TAU group (Cohen's d = 0.50).

Table 2 displays the means ± SD for all outcome variables at baseline, posttreatment and the six-month follow-up. As shown in the right column, the ANCOVA yielded significant effects for global PC and the three PCS dimensions (rumination, magnification and helplessness). The pairwise comparisons yielded the same pattern of results for all PCS dimensions. After treatment, we found that CBT had been more effective than the other two treatments in reducing PC, rumination, magnification and helplessness. These improvements were still observed at the six-month follow-up. In addition, we found that RPT was more effective than TAU at reducing all of the PCS dimensions.

ANCOVA revealed significant effects (see Table 2) of pain acceptance (CPAQ), depression (HAM-D), anxiety (HARS), pain (PVAS), functional impairment (FIQ) and health-related quality of life (EuroQol VAS). Pairwise comparisons within the pain acceptance findings indicated that the patients who received CBT had improved more at the posttreatment and six-month follow-up examinations than those assigned to other treatments. There were no significant differences between the groups with regard to depression posttreatment; however, we found that depression had reduced more in the CBT group than in the TAU group at the six-month follow-up. CBT and RPT were equally effective at reducing anxiety: Both treatment options were significantly better than TAU posttreatment and at the six-month follow-up. There were no significant differences in pain level between the groups posttreatment. Pain levels had reduced more for the RPT group than for the TAU group by the six-month follow-up. An analysis of functional impairment revealed a group effect that favoured the CBT group posttreatment and at the six-month follow-up. We observed a group effect of the EuroQol VAS, such that patients who received CBT showed more improvement than those treated with RPT or TAU posttreatment and at the six-month follow-up.

To the best of our knowledge, this is the first study to assess the efficacy of CBT, RPT and TAU in reducing PC in patients diagnosed with FM. Only five previous randomised, controlled trials assessed the efficacy of treatments on PC in FM patients 26 27 28 29 30 , but none of them considered PC as the main outcome. Other strengths of this study are the assessment of the influences of variables such as depression, anxiety or pain on the variations in PC. These results can be generalised because of the high external validity of the sample study (recruited in primary care settings). One of the main limitations of the study is the number of secondary variables analysed, which raises the concern of obtaining a significant result by chance (type I error). However, because of the lack of similar studies and the range of possible outcomes in FM 30 , we considered it appropriate to use several different outcome measures.

CBT significantly decreased global PC at the six-month follow-up, with effect sizes of Cohen's d = 0.73 and 1.01 compared with RPT and TAU, respectively. Previous studies in which the effectiveness of CBT in reducing PC in FM patients was assessed found effect sizes ranging from 0.17 29 and 0.26 25 up to a maximum of 0.56 29 . The effect sizes in our study are similar to those found in previous studies on the efficacy of CBT in FM patients and in a meta-analysis of psychological and pharmacological treatments for FM administered at primary and secondary levels of healthcare 30 . However, our current study is the first to assess the effectiveness of CBT compared with RPT and TAU on reducing the three subscales of PC. It confirms that CBT was significantly more effective than the other two treatments.

In some previous studies, comparison of CBT with placebo therapies for the treatment of PC in FM patients produced inconsistent results. For instance, in Vlaeyen et al.'s study 29 , CBT was less effective (effect size 0.17) than education (effect size 0.27). In our study, we found that CBT was more effective than both RPT and TAU and that RPT was more effective than TAU in the reduction of all PCS dimensions.

CBT is more effective than RPT or TAU, both at posttreatment and at the six-month follow-up, for the following secondary outcomes measured in this study: pain acceptance, anxiety, global function and quality of life. However, CBT is not more effective than the other two interventions at reducing depression and pain.

CBT was more effective than RPT and TAU at increasing pain acceptance at the six-month follow-up (effect sizes of Cohen's d = 0.77 and 0.80, respectively). This report is one of the first to study the effectiveness of CBT in treating pain acceptance, a key concept of third-wave therapies, especially Acceptation and Commitment Therapy (ACT) 54 . ACT proposes that attempting to control internal events such as pain sensations and negative emotional reactions is problematic. 'Pain acceptance' is a psychological construct that refers to the process of learning to live with pain. Although we have described the components of the psychotherapy, we did not include ACT elements. We hypothesise that CBT increases pain acceptance because it promotes the acquisition of diverse skills needed to manage pain. For instance, cognitive restructuring enhances coping (by reframing) and might affect pain acceptance. Indeed, a previous study found great similarities between a behavioural coping strategy (that is, task persistence) and pain acceptance 55 .

Regarding depression, there are no differences among the three treatments in decreasing depression either posttreatment or at the six-month follow-up. This is not surprising, because patients in the pharmacological and TAU groups who were diagnosed with depression on the basis of the MINI psychiatric interview used antidepressants according to treatment guidelines 37 38 . Many studies have demonstrated that pharmacological and psychological treatments are similarly effective in treating depression 30 . Researchers in most previous studies of CBT in FM patients have found effect sizes of about 0.3 to 0.4, with some outliers reaching 1.22 25 . Some investigators have found CBT to be ineffective for treating depression in FM patients 29 . Another meta-analysis concluded that the efficacy of CBT in treating depression could not be distinguished without some risk of bias 56 . In our study, CBT was effective for decreasing depression, but RPT and TAU were also quite effective, so we found no significant differences among the three.

In our study, CBT and RPT improved anxiety symptoms significantly more than did TAU. All patients in the pharmacological group were treated with pregabalin, a drug that is effective for both pain and anxiety. In addition, about 50% of patients in this group who were diagnosed with depression took antidepressants that are also effective for anxiety. We do not think this is due to low effectiveness of CBT, because previous studies of anxiety have confirmed that it is effective 57 . Instead, we believe that this result was obtained because TAU is effective in reducing anxiety. It is difficult to compare these results with those of previous studies, because anxiety is not usually an outcome of interest for psychological interventions in FM patients 25 30 .

In our study, CBT did not improve pain (as assessed by PVAS) more than the other two treatments either posttreatment or at the six-month follow-up. Only RPT is more effective than TAU at six months. Previous studies of CBT in FM patients 24 have found effect sizes of 0.35 to 0.5 for pain. In our study, all of the patients in the RPT arm took pregabalin, a potent analgesic, and over half of them also took duloxetine, which, in addition to being an antidepressant, has an analgesic effect. It seems that for pain, CBT cannot overcome the effect of RPT. The lesser effectiveness of TAU can be attributed to many family doctors' not systematically using pregabalin to treat FM patients or duloxetine to treat patients with associated depression.

For patients with FM, CBT, compared with RPT and TAU, was more effective at improving global function as assessed by the FIQ (six-month effect sizes Cohen's d = 0.44 and 0.53, respectively) and quality of life as assessed by the EuroQol VAS (six-month effect sizes: Cohen's d = 0.11 and 0.40, respectively). Previous studies of CBT in FM patients 24 have found effect sizes of 0.2 to 0.5 based on the FIQ 24 . However, a recent meta-analysis in which quality of life was assessed 56 suggested that CBT is not effective for the achievement of this outcome. CBT may be more effective than other treatments at changing these two variables because they are global assessments of the patient, and CBT is more able than pharmacological approaches to improve many aspects of FM.

How pharmacotherapy might improve PC has not been discussed herein, thus only tentative suggestions can be proposed to explain our results. The current conceptualisations of pain incorporate a biopsychosocial approach that involves behavioural reactions (for example, avoidance behaviour), cognitive reactions (for example, PC) and physiological reactions (for example, elevated autonomic arousal and muscle tension). These approaches are highly related and establish a vicious cycle 58 . RPT improves behavioural and physiological reactions 59 , thus their effects might indirectly affect cognitive reactions (that is, PC).