Cognitive–behavioral mechanisms in a pain-avoidance and a pain-persistence treatment for high-risk fibromyalgia patients

Authors


Abstract

Objective

The heterogeneity of cognitive–behavioral patterns in patients with fibromyalgia (FM) has been proposed to underlie the variability in treatment outcomes. It has previously been shown that pain-avoidance and pain-persistence treatments tailored to the patient's pattern are effective in improving physical and psychological functioning and overall impact in high-risk patients with heigthened psychological distress. In the present study, the cognitive–behavioral effects of these treatments were evaluated to provide insight into the main proposed mechanisms, specifically pain-avoidance behaviors and activity pacing in the pain-avoidance and pain-persistence treatments, respectively.

Methods

High-risk FM patients were classified into 2 groups, pain avoidance and pain persistence, and randomized in groups to the relevant treatment or waiting-list control condition. The pain-avoidance and pain-persistence treatments both comprised 16 twice-weekly sessions of cognitive–behavioral therapy and exercise training. Cognitive–behavioral factors assessed at pre- and posttreatment and 6 months of followup were evaluated using linear mixed models.

Results

A significant treatment effect was found for pain-avoidance behavior in the pain-avoidance treatment and for activity pacing in the pain-persistence treatment, showing improvements in the treatment condition relative to the controls. Furthermore, the effect on functioning was mediated by changes in pain-avoidance behavior in the pain-avoidance treatment and by changes in activity pacing in the pain-persistence treatment. Both treatments also showed significant improvements in other relevant cognitive–behavioral factors.

Conclusion

Both the pain-avoidance and pain-persistence treatments are effective in improving cognitive–behavioral factors in high-risk FM patients. Pain-avoidance behavior and activity pacing might be important mediating mechanisms for beneficial outcomes in pain-avoidance and pain-persistence treatments, respectively.

INTRODUCTION

Multidisciplinary chronic pain treatments are aimed at modifying maladaptive cognitive and behavioral responses to pain in order to improve the patient's physical and psychological functioning. Meta-analyses showed that combinations of cognitive–behavioral therapy (CBT) and exercise training yield the most favorable results in fibromyalgia (FM), a chronic musculoskeletal pain disorder characterized by widespread pain, fatigue, functional disability, and heightened psychological distress (1, 2). However, effects were limited and showed large individual variability (3–6). Supporting evidence has since been found for several chronic pain conditions, showing that the efficacy of interventions improves when the heterogeneity of the patient group is taken into consideration and treatments are matched to the patient's cognitive–behavioral mechanisms (7–13).

Specific cognitive–behavioral factors have been proposed to account for the maintenance and exacerbation of symptoms in patients with chronic pain. Fear-avoidance models postulate that factors such as pain-avoidance behavior, fear of pain, catastrophizing, hypervigilance, and social reinforcement contribute to chronic pain and long-term disability (14–17). However, these pain-avoidance patterns seem less applicable for patients with pain-persistence patterns, who tend to continuously persist in their activities despite the pain and ignore pain-related signals (18–22). Various studies have demonstrated the prognostic value and independent contribution of both pain-avoidance and pain-persistence factors for the development and maintenance of physical and psychological impairments in various chronic pain disorders, including FM (10, 17, 18, 21–25). Preliminary evidence also suggests that adapting treatment to these specific cognitive–behavioral factors may be promising (10, 11, 26). Because distress has been shown to be an indicator of maladaptive cognitive–behavioral patterns targeted in multidisciplinary CBTs, these patterns specifically pose a problem in high-risk patients with heightened levels of distress (7, 27). Several studies indeed found relatively highly distressed FM patients to obtain the better treatment outcomes (27–29). As to the timing of treatment, several studies demonstrated that chronic pain patients with a shorter duration of symptoms benefit most from CBT (30, 31). In sum, identifying high-risk patients in an early stage and offering them timely CBT and exercise training tailored to their specific pain-avoidance or pain-persistence pattern is likely to promote treatment outcomes in FM. In two previous reports, we indeed showed this approach to be effective in improving self-reported physical and psychological functioning, the impact of FM, as well as the level of physical fitness (13, Van Koulil S, et al: unpublished observations). These studies, however, did not address the cognitive–behavioral mechanisms responsible for the obtained effects. The aim of the present study accordingly was to examine the cognitive–behavioral mechanisms of a pain-avoidance treatment and a pain-persistence treatment. Since pain-avoidance behaviors and activity pacing reflect the main cognitive–behavioral mechanisms in the pain-avoidance and pain-persistence treatments, respectively, we expected the pain-avoidance treatment to be effective in diminishing pain-avoidance behaviors and the pain-persistence treatment to be effective in improving activity pacing. We also explored whether the previously reported improvements in physical and psychological functioning (13) were mediated by pain-avoidance behaviors in the pain-avoidance group and by activity pacing in the pain-persistence group.

SUBJECTS AND METHODS

The study reported here is part of a larger trial of which the methods have been extensively described elsewhere (13). The section below is therefore restricted to those aspects that are relevant for the current study. The study was approved by the medical ethics committee and the trial was registered in a clinical trials registration (http://clinicaltrials.gov/).

Participants and procedure.

For the inclusion, 457 eligible patients were first screened for risk profiles of heightened psychological distress (7, 32, 33) and subsequently for the pain-avoidance and pain-persistence patterns (34, 35). The questionnaire was returned by 379 patients (83%) and of this sample, 242 patients (64%) had a risk profile of heightened psychological distress. Results indicated that high-risk patients were characterized by significantly higher levels of pain, fatigue, functional disability, and impact of FM on daily life, and scored higher on cognitive–behavioral factors of pain-avoidance behavior, pain-related retreating, worrying about pain, fear of pain, hypervigilance, and helplessness, and scored lower on acceptance and perceived social support than the low-risk patients (all P < 0.001). No significant differences were found on social reinforcement and activity pacing. When comparing the pain-avoidance and pain-persistence treatments, both groups had similar sociodemographic backgrounds; however, consistent with our previous study (35), the patients with a pain-avoidance pattern had significantly higher scores on functional disability, pain, negative mood, and impact of FM, and the cognitive–behavioral factors of pain-avoidance behavior, pain-related retreating, worrying about pain, fear of pain, hypervigilance, social reinforcement, helplessness, and activity pacing (all P < 0.05). No significant differences were found for acceptance and perceived social support. The sociodemographic background of the patients is shown in Table 1.

Table 1. Baseline sociodemographic characteristics of the participants for each of the study conditions*
 Pain persistencePain avoidance
TC (n = 39)WLC (n = 45)TC (n = 29)WLC (n = 45)
  • *

    TC = treatment condition; WLC = waiting-list control condition.

  • Primary, secondary, and tertiary education represents an average of 7, 12, and 17 years of formal education, respectively.

Female sex, %97899396
Married/cohabiting, %82777276
Age, mean ± SD years41.1 ± 9.440.9 ± 10.442.3 ± 12.439.4 ± 10.4
Educational level, %    
 Primary9542
 Secondary77718193
 Tertiary1424155

Physical and psychological functioning and cognitive–behavioral factors were assessed at baseline, posttreatment, and 6 months of followup in the treatment condition (TC) and at corresponding intervals in the waiting-list control condition (WLC) using validated self-report questionnaires (see Table 2 for measures). No significant differences were found with regard to the treatment credibility of the pain-avoidance and pain-persistence treatments after the first session and at the end of the treatment, suggesting that both treatments were perceived to be equally credible (36).

Table 2. Mean ± SD (number of patients) for the cognitive–behavioral factors in the treatment condition (TC) and the waiting-list control condition (WLC) for the pain-avoidance and pain-persistence groups at baseline, postassessment, and 6-month followup
 BaselinePostassessment6-month followup
  1. * Response categories were slightly modified to 4-point Likert scales ranging from “rarely or never” to “very frequently.”

Pain-avoidance behavior (34)   
 Pain avoidance   
  TC2.5 ± 0.5 (n = 28)2.3 ± 0.4 (n = 24)2.3 ± 0.4 (n = 22)
  WLC2.5 ± 0.5 (n = 39)2.5 ± 0.4 (n = 38)2.5 ± 0.4 (n = 37)
 Pain persistence   
  TC2.1 ± 0.5 (n = 37)2.2 ± 0.4 (n = 36)2.1 ± 0.4 (n = 34)
  WLC2.2 ± 0.5 (n = 42)2.2 ± 0.5 (n = 42)2.1 ± 0.5 (n = 41)
Activity pacing (39)*   
 Pain avoidance   
  TC2.4 ± 0.5 (n = 28)2.8 ± 0.5 (n = 24)2.7 ± 0.3 (n = 23)
  WLC2.4 ± 0.5 (n = 39)2.3 ± 0.5 (n = 39)2.3 ± 0.6 (n = 37)
 Pain persistence   
  TC2.1 ± 0.5 (n = 37)2.6 ± 0.6 (n = 36)2.5 ± 0.6 (n = 33)
  WLC2.2 ± 0.5 (n = 42)2.2 ± 0.5 (n = 42)2.1 ± 0.6 (n = 42)
Pain-related retreating (34)   
 Pain avoidance   
  TC2.3 ± 0.4 (n = 28)2.2 ± 0.5 (n = 24)2.2 ± 0.5 (n = 22)
  WLC2.1 ± 0.5 (n = 39)2.2 ± 0.5 (n = 38)2.1 ± 0.5 (n = 37)
 Pain persistence   
  TC1.9 ± 0.4 (n = 37)1.8 ± 0.3 (n = 36)1.8 ± 0.4 (n = 34)
  WLC1.8 ± 0.5 (n = 42)1.8 ± 0.5 (n = 42)1.8 ± 0.6 (n = 41)
Worrying about pain (34)   
 Pain avoidance   
  TC2.4 ± 0.5 (n = 28)1.8 ± 0.6 (n = 24)1.9 ± 0.6 (n = 22)
  WLC2.3 ± 0.6 (n = 39)2.2 ± 0.6 (n = 38)2.2 ± 0.6 (n = 37)
 Pain persistence   
  TC2.1 ± 0.5 (n = 37)1.7 ± 0.3 (n = 36)1.7 ± 0.3 (n = 34)
  WLC2.1 ± 0.5 (n = 42)2.0 ± 0.4 (n = 42)1.9 ± 0.5 (n = 41)
Fear of pain (40)   
 Pain avoidance   
  TC28.5 ± 6.8 (n = 28)22.3 ± 5.3 (n = 24)24.3 ± 6.5 (n = 23)
  WLC28.0 ± 7.0 (n = 39)26.1 ± 5.7 (n = 38)26.1 ± 7.4 (n = 36)
 Pain persistence   
  TC23.3 ± 5.9 (n = 37)20.7 ± 5.0 (n = 36)20.5 ± 5.7 (n = 34)
  WLC24.0 ± 5.8 (n = 43)23.5 ± 6.0 (n = 41)24.0 ± 5.2 (n = 41)
Hypervigilance (41)   
 Pain avoidance   
  TC39.8 ± 9.2 (n = 28)30.7 ± 9.0 (n = 24)31.5 ± 11.3 (n = 23)
  WLC38.1 ± 11.6 (n = 39)36.2 ± 10.9 (n = 38)33.9 ± 9.4 (n = 36)
 Pain persistence   
  TC31.1 ± 9.1 (n = 37)27.3 ± 8.3 (n = 36)26.1 ± 8.8 (n = 34)
  WLC33.6 ± 10.5 (n = 43)32.6 ± 11.4 (n = 41)32.9 ± 12.1 (n = 42)
Social reinforcement (42)   
 Pain avoidance   
  TC3.0 ± 1.2 (n = 28)2.5 ± 1.1 (n = 25)2.4 ± 1.4 (n = 22)
  WLC3.2 ± 1.3 (n = 38)3.2 ± 1.3 (n = 40)3.1 ± 1.2 (n = 35)
 Pain persistence   
  TC2.7 ± 1.3 (n = 37)2.7 ± 1.3 (n = 35)2.6 ± 1.2 (n = 33)
  WLC2.5 ± 1.3 (n = 43)2.4 ± 1.4 (n = 40)2.6 ± 1.1 (n = 41)
Acceptance (43)   
 Pain avoidance   
  TC12.4 ± 3.8 (n = 36)12.4 ± 3.8 (n = 36)12.4 ± 3.8 (n = 36)
  WLC12.4 ± 3.8 (n = 36)12.4 ± 3.8 (n = 36)12.4 ± 3.8 (n = 36)
 Pain persistence   
  TC13.1 ± 3.8 (n = 36)16.3 ± 3.5 (n = 36)16.5 ± 3.5 (n = 34)
  WLC12.8 ± 3.5 (n = 43)13.4 ± 3.3 (n = 41)13.3 ± 3.8 (n = 42)
  TC16.9 ± 3.5 (n = 28)13.0 ± 4.0 (n = 25)12.2 ± 3.8 (n = 23)
  WLC16.6 ± 3.9 (n = 39)15.8 ± 3.7 (n = 40)15.7 ± 4.4 (n = 36)
 Pain persistence   
  TC14.4 ± 3.8 (n = 36)11.3 ± 2.9 (n = 36)11.1 ± 3.1 (n = 34)
  WLC13.4 ± 2.9 (n = 43)12.6 ± 3.3 (n = 41)13.0 ± 3.4 (n = 42)
Perceived social support (32, 44)   
 Pain avoidance   
  TC13.1 ± 4.1 (n = 28)14.0 ± 3.7 (n = 25)14.0 ± 3.6 (n = 23)
  WLC12.8 ± 3.4 (n = 38)13.7 ± 3.8 (n = 40)13.3 ± 4.0 (n = 37)
 Pain persistence   
  TC14.4 ± 3.5 (n = 37)15.2 ± 3.8 (n = 35)15.2 ± 3.7 (n = 33)
  WLC13.4 ± 4.1 (n = 43)13.1 ± 3.4 (n = 41)13.2 ± 3.3 (n = 41)

Tailored treatment.

The patients received a highly structured outpatient treatment in a group setting of 8 participants. Patients received a pain-avoidance or a pain-persistence treatment, depending on their specific cognitive–behavioral pattern of pain avoidance or pain persistence. Both the pain-avoidance and pain-persistence treatments consisted of 16 twice-weekly sessions and one booster session 3 months after treatment completion, with every regular session starting with 2 hours of CBT followed by 2 hours of exercise training. The goal of both treatments was to diminish the cognitive, emotional, behavioral, and social consequences of pain and accompanying symptoms in daily life. At the third, ninth, and fifteenth sessions, the partner or another significant relation of the patient attended the CBT. The exercise training included aerobic exercises (e.g., cycling, gymnastic exercises), anaerobic exercises (e.g., strength and flexibility exercises, functional walking training), hydrotherapy, and relaxation therapy. The pain-avoidance treatment was specifically aimed at increasing the patient's level of daily activities and diminishing their pain-avoidance behaviors by stimulating them to gradually and systematically increase their daily activities with individual goals and exposure to fear-related situations as the guiding principle. The pain-persistence treatment first focused on regulating and diminishing their pain-persistence behaviors by teaching the participants to pace their activities and to alternate between activity and inactivity, followed by gradually increasing their daily activities. In both the pain-avoidance and pain-persistence treatments, the patients received consolidating homework assignments supporting the CBT and exercise sessions. The booster session focused on relapse prevention and a further improvement of the attained goals (for extended treatment descriptions of the pain-avoidance and pain-persistence treatments, see refs.12 and13).

Results with regard to the level of functioning.

In a previous report, we evaluated the effects of the tailored treatment approach in high-risk FM patients with regard to the level of physical functioning (pain, fatigue, functional disability), psychological functioning (negative mood, anxiety), and impact of FM (13). Treatment effects were significant for all primary outcomes, showing significant improvements for all outcomes for the TC in comparison to the controls. Effect sizes in the TC overall were large, and reliable change indices indicated a clinically relevant improvement among the TC. In addition, the treatment was also effective in improving the level of physical fitness relative to the controls, with large effect sizes (Van Koulil S, et al: unpublished observations).

Statistical analyses.

Differences between subgroups were tested with chi-square analyses for categorical data and Student's t-test for continuous variables with a significance level of P values less than 0.05 (2-tailed). Treatment effects were evaluated for the pain-avoidance and pain-persistence treatments separately using a linear mixed model taking into account the specific design features of this trial. For the cognitive–behavioral factors of pain-avoidance behavior and activity pacing, the posttreatment and followup assessments were used as dependent variables and condition and baseline assessment of the cognitive–behavioral factor and time as independent variables in the primary analyses. The same procedure was followed for the other 8 cognitive–behavioral factors that were exploratively tested. Random effects were added for randomization groups and an unstructured covariance matrix was used to model the dependence of the two posttreatment measurements. Secondary analyses contained time by treatment interactions to test for a stable treatment effect over the posttreatment and followup measurements. All analyses were performed using the intent-to-treat principle. Data sets for the cognitive–behavioral variables were complete for 89% of the total sample of 158 patients and for 97% of the 137 completers. Since we used a mixed model and missing values only occurred on the dependent variables, our present analyses can only be guaranteed to be valid when missing values occurred as missing at random. Therefore, we also performed an analysis using last observation carried forward (LOCF) as a sensitivity analysis.

Mediation was examined by the joint significance test according to the procedures proposed by MacKinnon et al, and was shown to outperform other mediation tests with regard to Type I error and statistical power (37, 38). This test requires the path from condition to mediator and the path from mediator to outcome both to be statistically significant in order to provide evidence for a mediation effect. Therefore, the effect of treatment on the mediators between pre- and postassessment was first examined with mixed regression by using the posttreatment scores of the mediators as dependent variables and condition and baseline assessment of the mediators as independent variables. Random effects were added for randomization groups. Second, we tested the effect of the mediator on the primary outcomes of physical (standardized composite score of pain, fatigue, functional disability) and psychological functioning (standardized composite score of anxiety, negative mood), adjusted for treatment by using the posttreatment scores of the outcome as the dependent variable and condition, baseline assessments of the outcome, and mediator as independent variables, and by adding the mediator as a within-subject covariate. The joint significance of the association between treatment and mediator, and the association between the mediator and outcome adjusted for treatment, provides evidence for mediation (37, 38).

RESULTS

Outcomes for the pain-avoidance group.

As expected, a significant condition effect was found for pain-avoidance behavior (F[1,57.62] = 4.09, P < 0.05), and the TC showed a 0.17-point lower posttreatment and followup assessment score compared to the WLC. Furthermore, in additional explorative analyses for the cognitive–behavioral factors of fear of pain (F[1,55.89] = 10.59, P < 0.01), activity pacing (F[1,10.20] = 20.17, P < 0.01), acceptance (F[1,58.41] = 22.35, P < 0.001), and social reinforcement (F[1,57.35] = 4.66, P < 0.05), a significant condition effect was also found, with the TC showing a posttreatment and followup assessment score that was, compared to the WLC, 3.34 points lower for fear of pain, 0.47 points higher for activity pacing, 3.49 points higher for acceptance, and 0.43 points lower for social reinforcement. Effect sizes were overall moderate to large for the TC (range 0.40–1.75 postassessment, range 0.40–1.59 for the followup assessment). In addition, trends were found for worrying about pain (F[1,9.72] = 4.76, P = 0.06) and hypervigilance (F[1,8.69] = 4.66, P = 0.06), with the TC showing a 0.30- and 4.13-point lower posttreatment and followup assessment score, respectively, compared to the WLC. No significant condition effects were found with regard to pain-related retreating, helplessness, and perceived social support. Furthermore, no significant time × condition interaction effects were found, indicating a stable effect between posttreatment and followup assessments. The sensitivity analysis showed that the results of the analyses on the complete data set obtained after using LOCF were comparable to results shown above (Table 2).

Outcomes for the pain-persistence group.

As expected, a significant condition effect was found for activity pacing (F[1,76.00] = 17.87, P < 0.001), and the TC showed a 0.42-point higher posttreatment and followup assessment score compared to the WLC. Furthermore, in additional explorative analyses for the cognitive–behavioral factors of worrying about pain (F[1,8.37] = 17.37, P < 0.01), fear of pain (F[1,75.53] = 14.31, P < 0.001), hypervigilance (F[1,9.01] = 6.90, P < 0.05), acceptance (F[1,74.84] = 23.66, P < 0.001), and helplessness (F[1,76.89] = 12.81, P = 0.001), a significant condition effect was also found, with the TC showing a posttreatment and followup assessment score that was, compared to the WLC, 0.27 points lower for worrying about pain, 2.57 points lower for fear of pain, 4.03 points lower for hypervigilance, 2.71 points higher for acceptance, and 2.04 points lower for helplessness. Effect sizes were overall moderate to large for the TC (range 0.44–1.00 postassessment, range 0.48–0.99 for the followup assessment). In addition, a trend was found for perceived social support (F[1,9.75] = 4.50, P = 0.06), and the TC showed a 1.27-point higher posttreatment and followup assessment score compared to the WLC. No significant effects of condition were found with regard to pain-avoidance behavior, pain-related retreating, and social reinforcement. Furthermore, no significant time × condition interaction effects were found, indicating a stable effect between posttreatment and followup assessments. The sensitivity analysis showed that the results of the analyses on the complete data set obtained after using LOCF were comparable to results shown above (Table 2).

Mediating cognitive–behavioral factors.

Mediation was examined with regard to the primary outcomes of physical and psychological functioning by the joint significance test (37). In an earlier report of this study it was shown that both the pain-avoidance and pain-persistence treatments were effective in improving the primary outcomes of physical and psychological functioning (12, 13). For the pain-avoidance group, we analyzed the mediating effect of pain-avoidance behavior on functioning. First, a significant effect was found of condition on the mediator pain-avoidance behavior (F[1,59] = 4.91, P < 0.05). Second, results on the relationship between the mediator and physical and psychological functioning when controlling for treatment showed that pain-avoidance behavior was significantly related to physical functioning (F[1,56] = 6.20, P < 0.05), but not to psychological functioning. These results show that the effect of the TC relative to the WLC on physical functioning but not on psychological functioning was mediated by pain-avoidance behavior in the pain-avoidance group. For the pain-persistence group, we analyzed the mediating effect of activity pacing on physical and psychological functioning. First, the effect of condition on the mediator was examined. A significant effect was found of condition on the mediator activity pacing (F[1,77] = 17.44, P < 0.001). Second, results on the relationship between the mediator and physical and psychological functioning when controlling for treatment showed that activity pacing was significantly related to physical functioning (F[1,77] = 20.20, P < 0.001) and psychological functioning (F[1,76] = 5.06, P < 0.05), showing that the effect of the TC relative to the WLC on physical and psychological functioning was mediated by activity pacing in the pain-persistence group.

DISCUSSION

In a previous study, we showed the effectiveness of a pain-avoidance treatment and a pain-persistence treatment for high-risk FM patients with regard to physical and psychological functioning (13). The aim of the present study was to examine the cognitive–behavioral mechanisms that might be responsible for the obtained effects of these treatments. In line with our hypothesis, relative to the results of the WLC, the pain-avoidance treatment was effective in changing pain-avoidance behavior and the pain-persistence treatment was effective in changing activity pacing. In addition, our mediation analyses showed that changes in physical functioning were mediated by pain-avoidance behavior in the pain-avoidance treatment and changes in physical and psychological functioning were mediated by activity pacing in the pain-persistence treatment, providing some preliminary support that these mechanisms partially account for the treatment effects obtained.

The pain-avoidance and pain-persistence treatments evaluated in this study were based on specific pain-related cognitive–behavioral mechanisms that have been shown to be important in the maintenance and exacerbation of chronic pain (17, 18, 23–25). Preliminary evidence suggests that tailoring treatment to these maladaptive cognitive–behavioral factors may be promising. Exposure in vivo, for example, a pain-avoidance approach aimed at improving functioning by diminishing fear of pain, was found to be particularly effective in highly fearful patients with chronic low back pain (26). Thieme and colleagues (10) found that the patients with FM that benefited most from a pain-avoidance intervention had higher baseline levels of pain behavior, social reinforcement, and catastrophizing, while the patients who benefited most from a pain-persistence treatment had lower baseline levels of pain behavior and social reinforcement. Accordingly, patients exhibiting pain-avoidance patterns appear to benefit most from interventions aimed at diminishing their pain-avoidance behavior and increasing their daily activities, whereas for patients showing pain-persistence patterns, regulating their daily activities and restructuring of their pain-persistence cognitions seem more favorable (9, 10, 28). In extending these findings, we offered patients a pain-avoidance or a pain-persistence treatment tailored to their cognitive–behavioral pattern, and our results provide preliminary evidence that pain-avoidance behavior and activity pacing are relevant cognitive–behavioral mechanisms of these treatments and might justify separate treatment approaches. However, both the pain-avoidance and pain-persistence interventions also showed common significant or almost significant effects with regard to several cognitive–behavioral factors, such as activity pacing, fear of pain, and acceptance, indicating that both treatments also consist of generic treatment components. In addition, post hoc analyses showed that activity pacing also mediated changes in physical functioning in the pain-avoidance treatment. These findings overall support the role of both generic and specific cognitive–behavioral mechanisms that are responsible for the beneficial effects.

To our knowledge, our study is the first to specifically address pain-avoidance and pain-persistence patterns in FM. In a previous study we reported preliminary evidence of distinctive differences between these patterns, i.e., self-reported pain-avoidance and pain-persistence factors, physical fitness levels, and therapists' judgments after a semistructured interview (35). Other relevant questions about the two cognitive–behavioral patterns remain. For example, in line with previous findings (18, 20), we found the patients with a pain-persistence pattern to report fewer and less severe symptoms (including functional disability) than the patients with a pain-avoidance pattern, suggesting that the former pattern might be less maladaptive, at least in early stages of the condition. In this context, it is important to underline that we only enrolled patients reporting heightened levels of distress in whom the cognitive–behavioral patterns are presumably maladaptive. The adverse effects of the cognitive–behavioral patterns were supported by the fact that the pain-avoidance as well as the pain-persistence group reported more (serious) symptoms than the patients we had assessed as not being at risk (13). Nevertheless, additional research is needed to further clarify the short- and long-term as well as the adaptive and maladaptive consequences of these two cognitive–behavioral patterns in high- and low-risk patients. Also, the stability of these patterns merits further investigation because it is plausible that some patients might change strategies at some stages. It has, for instance, been proposed that pain-persistence patterns might be particularly relevant in the earlier acute phase of the condition, which would then eventually lead to exhaustion and subsequent pain-avoidance behavior in the long-term, when chronicity has set in (22). Finally, we need to learn more about the motivational mechanisms underlying pain-persistence patterns, such as pain-related fear of failure, low self-esteem, personality characteristics such as perfectionism, and specific stop rules (20, 22).

Some drawbacks of this study need mentioning. Our trial is, of course, limited in that we did not directly test whether our tailored interventions were more effective in improving cognitive–behavioral mechanisms than a standard, nontailored treatment. Second, we used cross-sectional data for our mediator analyses, and therefore causality may be bidirectional. To establish whether changes in cognitive–behavioral factors occur before or after changes in functioning, future studies might add several midtreatment assessments or use cross-lagged analyses. Finally, next to self-reported instruments, there is a need for behavioral measures such as activity levels to assess more objectively specific pain-avoidance and pain-persistence characteristics.

In conclusion, the present study shows that a pain-avoidance and a pain-persistence treatment tailored to the patient's main cognitive–behavioral pattern is effective in improving pain-avoidance behavior and activity pacing, respectively, and these factors in turn mediate the beneficial health outcomes of these treatments in FM. Future research should aim at further investigating specific cognitive–behavioral mechanisms in pain-avoidance and pain-persistence treatments and their independent and common contribution to long-term treatment effects.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. van Koulil had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Kraaimaat, van Lankveld, Thieme, Evers.

Acquisition of data. Van Koulil, Kraaimaat, van Lankveld, van Helmond, Vedder, van Hoorn, Cats, van Riel, Evers.

Analysis and interpretation of data. Van Koulil, Kraaimaat, van Lankveld, Donders, van Riel, Evers.

Acknowledgements

The authors would like to thank M. Effting, M. Limborgh, M. Font Freide, C. Barends, R. Speulman, and A. J. L. Verborg for their help in collecting the research data. We thank all of the rheumatologists and rheumatology nurse consultants of the St. Maartenskliniek, University Medical Centre St. Radboud Nijmegen, Rijnstate Hospital, Jeroen Bosch Ziekenhuis, and Rivierenland Ziekenhuis for referring patients. Finally, we would like to thank all the patients that have participated in this trial.

Ancillary