Impact statement
Fibromyalgia syndrome (FMS) has a substantial impact on daily functioning, emotional well-being, social participation and quality of life. Beyond pain intensity, the way in which people interpret and respond to pain may influence the persistence and severity of symptoms. Pain catastrophising is particularly relevant because it is associated with greater distress, disability, avoidance behaviour and perceived loss of control.
This review examines the potential clinical value of multicomponent pain therapy (MPT) for FMS. By integrating several therapeutic strategies, MPT may address physical symptoms together with cognitive, emotional and behavioural factors that contribute to the burden of the condition. The findings suggest that pain catastrophising may be a relevant therapeutic target within interdisciplinary care, although the certainty of evidence remains limited.
The results may inform clinical decision-making and the design of structured MPT programmes. Medium-duration interventions yielded the largest subgroup estimate; however, this exploratory finding should not be interpreted as establishing an optimal treatment duration. Coordinated, person-centred models of care may help address coping, self-management and functioning alongside symptom reduction.
Overall, this study contributes to understanding how psychological and behavioural processes may be addressed within FMS care. Further high-quality trials are needed before firm conclusions can be drawn about the magnitude, durability and clinical significance of the observed effects.
Introduction
Fibromyalgia syndrome (FMS) is a chronic disorder characterised by widespread, persistent musculoskeletal pain and is associated with central sensitisation, structural and functional brain alterations, physical and mental fatigue, sleep disturbance, emotional symptoms (e.g., anxiety and depression) and cognitive difficulties (e.g., attention, concentration and processing speed) (Galvez-Sánchez et al., Reference Galvez-Sánchez, Reyes Del Paso and Duschek2018). FMS is diagnosed predominantly in women; in studies using criteria-based classifications, women account for ~61–90% of cases, while prevalence in the general population has been estimated at 2–4%. However, these epidemiological estimates warrant cautious interpretation. Gender-related bias in diagnostic criteria and clinical recognition may contribute to the overrepresentation of women and the underdiagnosis or underrecognition of men with FMS (Wolfe et al., Reference Wolfe, Walitt, Perrot, Rasker and Häuser2018; Srinivasan et al., Reference Srinivasan, Maloney, Wright, Kennedy, Kallail, Rasker, Häuser and Wolfe2019).
FMS frequently co-occurs with conditions such as irritable bowel syndrome, migraine, chronic fatigue syndrome and gastrointestinal disorders, and is commonly accompanied by fatigue, insomnia, morning stiffness, depression and anxiety (Iannuccelli et al., Reference Iannuccelli, Favretti, Dolcini, Di Carlo, Pellegrino, Bazzichi, Atzeni, Lucini, Varrassi, Leoni, Fornasari, Conti, Salaffi, Sarzi-Puttini and Di Franco2025). FMS adversely affects physical, psychological and social functioning, including social relationships, the ability to fulfil family and work responsibilities and activities of daily living, owing not only to pain but also to fatigue (González-Álvarez et al., Reference González-Álvarez, Rossettini and Villafañe2025). From a biopsychosocial perspective, the condition may also adversely affect self-esteem and self-efficacy. Psychological factors are therefore relevant to the perception, processing and interpretation of chronic pain and to how individuals with FMS cope with it (Galvez-Sánchez et al., Reference Galvez-Sánchez, Duschek and Del Paso2019).
Regarding coping-related processes, pain catastrophising has been defined as an exaggerated negative cognitive-affective response to actual or anticipated pain, characterised by rumination, magnification and feelings of helplessness (Quartana et al., Reference Quartana, Campbell and Edwards2009). In chronic pain, this construct is particularly relevant within the fear-avoidance model, which proposes that catastrophic interpretations of pain may give rise to pain-related fear, avoidance behaviour, physical deconditioning and long-term disability (Vlaeyen and Linton, Reference Vlaeyen and Linton2000; Vlaeyen et al., Reference Vlaeyen, Crombez and Linton2016). In FMS, pain catastrophising, together with negative affect and stress, has been associated with greater symptom severity, fear and discomfort, which may worsen physical and cognitive symptoms (Alcon et al., Reference Alcon, Bergman, Humphrey, Patel and Wang-Price2023). Catastrophising may also exacerbate pain by increasing attentional focus on pain, strengthening negative emotional responses and disrupting adaptive attentional processing (López-Gómez et al., Reference López-Gómez, Velasco, Gutiérrez, Écija, Catalá and Peñacoba2023). These processes may reinforce functional limitations and reduce engagement in movement-related activities. In clinical practice, fear of exacerbating symptoms is frequently identified as a barrier to movement and exercise among individuals with FMS (Nijs et al., Reference Nijs, Roussel, Van Oosterwijck, De Kooning, Ickmans, Struyf, Meeus and Lundberg2013), consistent with the fear-avoidance framework and the importance of addressing catastrophising and pain-related fear in treatment.
For these reasons, pain catastrophising warrants specific attention in FMS as a clinically relevant cognitive-affective process that may shape adaptation to symptoms, rather than as a substitute for other important outcomes such as pain intensity, fatigue, or sleep disturbance (Petrini and Arendt-Nielsen, Reference Petrini and Arendt-Nielsen2020). Higher catastrophising has been associated with poorer emotional adjustment, reduced functional capacity and lower quality of life in this population (Lami et al., Reference Lami, Martínez, Miró, Sánchez and Guzmán2018a; López-Gómez et al., Reference López-Gómez, Velasco, Gutiérrez, Écija, Catalá and Peñacoba2023). Because catastrophising is potentially modifiable through psychological and multicomponent approaches, it may represent a meaningful treatment-related outcome alongside more traditional symptom indicators (Schütze et al., Reference Schütze, Rees, Smith, Slater, Campbell and O’Sullivan2018; Schumann et al., Reference Schumann, Coombes, Gascho, Geske, McDermott, Morrison, Reynolds, Bernau and Gilliam2022). However, compared with outcomes such as pain intensity or global functional impact, the effects of MPT on pain catastrophising in FMS have received less focused quantitative attention (Saracoglu et al., Reference Saracoglu, Akin and Aydin Dincer2022; Araya-Quintanilla et al., Reference Araya-Quintanilla, Cuyul-Vásquez, Lizama-Lefno, Jeria-Diaz, Fuentes, Serrat and Gutiérrez-Espinoza2025).
Management strategies for FMS include pharmacological and non-pharmacological approaches, and current clinical guidelines emphasise individualised care combining education, exercise, psychological therapies and symptom-oriented management (Macfarlane et al., Reference Macfarlane, Kronisch, Dean, Atzeni, Häuser, Fluß, Choy, Kosek, Amris, Branco, Dincer, Leino-Arjas, Longley, McCarthy, Makri, Perrot, Sarzi-Puttini, Taylor and Jones2017; Jones et al., Reference Jones, Asaad, Patel, Jain and Abd-Elsayed2024; Li et al., Reference Li, Li, Wu, Long, Chen, Wang, Zhao, Bingheng, Chang and Jiao2025). Exercise and psychological approaches, including cognitive-behavioural therapy, are relevant non-pharmacological components of FMS management. Nevertheless, no single intervention adequately addresses the broad and heterogeneous symptom profile of FMS. Pain neuroscience education (Suso-Martí et al., Reference Suso-Martí, Cuenca-Martínez, Alba-Quesada, Muñoz-Alarcos, Herranz-Gómez, Varangot-Reille, Domínguez-Navarro and Casaña2022), repetitive transcranial magnetic stimulation (Sun et al., Reference Sun, Fang, Zhang, Liu, Wang and Qi2022), virtual reality (Brea-Gómez et al., Reference Brea-Gómez, Pérez-Gisbert, Fernández-Castro, Valenza and Torres-Sánchez2025) and exercise-based approaches (Rodríguez-Domínguez et al., Reference Rodríguez-Domínguez, Rebollo-Salas, Chillón-Martínez, Rosales-Tristancho, Villa-del-Pino and Jiménez-Rejano2025) have shown promising results for pain catastrophising.
Taken together, these considerations underscore the need for integrated approaches capable of addressing the multifaceted symptom burden of FMS (Galvez-Sánchez and Del Paso, Reference Galvez-Sánchez and Del Paso2020). Although several emerging interventions have shown promising effects on specific dimensions of chronic pain, their relevance to the heterogeneous clinical presentation of FMS may be limited when they are applied in isolation. Accordingly, there is a rationale for comprehensive, interdisciplinary frameworks that combine complementary modalities to address biological, psychological and social determinants of chronic pain (Llàdser et al., Reference Llàdser, Montesó-Curto, López, Rosselló, Lear, Toussaint and Casadó-Martín2022).
Multicomponent pain therapy (MPT) is an interdisciplinary treatment approach grounded in the biopsychosocial model of chronic pain, in which pain arises from interacting biological, psychological and social factors (Deckert et al., Reference Deckert, Kaiser, Kopkow, Trautmann, Sabatowski and Schmitt2016). MPT typically integrates several evidence-based components delivered in a coordinated manner, such as exercise therapy, pain education, psychological interventions (e.g., cognitive-behavioural strategies) and, when appropriate, physical or manual therapies. These components aim to address pain-related symptoms and associated functional and emotional impairment. Evidence suggests that MPT may reduce pain intensity and FMS severity relative to comparator interventions and may also reduce pain catastrophising, depression and anxiety (Saracoglu et al., Reference Saracoglu, Akin and Aydin Dincer2022). Accordingly, this study aimed to evaluate whether MPT reduced pain catastrophising in individuals with FMS from a transdisciplinary perspective.
Despite growing interest in MPT, the available evidence remains difficult to interpret because protocols differ substantially in therapeutic composition, intensity, delivery format and follow-up duration. Their specific contribution to pain-related cognitive-affective outcomes therefore remains unclear. This issue is particularly relevant for pain catastrophising, which is associated with greater pain severity, emotional distress, fear of movement, functional limitations and poorer adaptation in individuals with FMS (Nijs et al., Reference Nijs, Roussel, Van Oosterwijck, De Kooning, Ickmans, Struyf, Meeus and Lundberg2013; Lami et al., Reference Lami, Martínez, Miró, Sánchez and Guzmán2018a; Petrini and Arendt-Nielsen, Reference Petrini and Arendt-Nielsen2020). Previous reviews of FMS interventions have focused primarily on pain intensity, global functional impact and quality of life rather than on pain catastrophising specifically (Saracoglu et al., Reference Saracoglu, Akin and Aydin Dincer2022; Araya-Quintanilla et al., Reference Araya-Quintanilla, Cuyul-Vásquez, Lizama-Lefno, Jeria-Diaz, Fuentes, Serrat and Gutiérrez-Espinoza2025). A focused systematic review was therefore needed to clarify whether MPT is associated with reductions in pain catastrophising among individuals with FMS.
The present study provided a focused quantitative synthesis of the association between MPT and pain catastrophising in FMS. It also explored, where data permitted, whether intervention duration accounted for variation in effect estimates. By synthesising a heterogeneous evidence base, this review sought to quantify post-intervention and follow-up effects, identify remaining evidence gaps and inform the design of future trials in which pain catastrophising is treated as a potentially modifiable outcome relevant to symptom persistence and disability.
Methods
This systematic review and meta-analysis were conducted in accordance with Cochrane Collaboration guidance and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al., Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson, McDonald, McGuinness, Stewart, Thomas, Tricco, Welch, Whiting and Moher2021). The completed PRISMA checklist is provided as Supplementary Material. The eligibility criteria and analyses were specified in advance, and the protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration ID: CRD42024596489).
The PubMed/MEDLINE, Scopus, Web of Science and PEDro databases were searched independently by two reviewers from inception to May 2026, without restrictions on publication date or language. The reference lists of relevant published studies were also searched. The PICOS framework was used to formulate the review question (Amir-Behghadami and Janati, Reference Amir-Behghadami and Janati2020): population (individuals with FMS), intervention (MPT), comparison (no intervention or therapies that did not include MPT), outcome (pain catastrophising) and study design (RCTs or pilot RCTs). The following MeSH and entry terms were used: “Fibromyalgia Syndrome,” “Fibromyalgia Fibromyositis Syndrome,” “Diffuse Myofascial Pain Syndrome,” “Catastrophization,” “Pain Catastrophizing,” “Catastrophic Thinking,” “Pain Related Cognitive Rumination” and “Pain Related Catastrophic Thinking.” The Boolean operators “AND” and “OR” were used. The final literature search was conducted on 7 May 2026.
Eligibility criteria
Studies were eligible if they (1) were RCTs or pilot RCTs; (2) were peer-reviewed original studies; (3) included adults (aged ≥18 years) diagnosed with FMS according to established criteria; and (4) compared MPT with a pharmacological intervention, another active intervention, a wait-list control or usual care. Studies were excluded if they (1) did not report pain catastrophising; (2) included samples that were not composed exclusively of participants with FMS; (3) did not provide quantitative data on the variables required for inclusion in the meta-analysis; or (4) were duplicate reports.
Two reviewers (C.G.-L. and J.A.C.-R.) independently screened all records and selected those that met the eligibility criteria for full-text assessment. Titles and abstracts were screened to remove irrelevant records, after which the full texts of potentially eligible reports were assessed against the inclusion and exclusion criteria. Any disagreement was reviewed by C.M.G.-S., who made the final decision regarding study inclusion. The screening and selection process is presented in the PRISMA flow diagram (Figure 1). Before data extraction and quality assessment, all authors reviewed the selected articles to confirm their eligibility.
PRISMA 2020 flow diagram of study selection. Note: This work is licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0. Source: Page et al. (Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson, McDonald, McGuinness, Stewart, Thomas, Tricco, Welch, Whiting and Moher2021; 372:n71).
Figure 1. Long description
The flowchart is organized into three vertical phases: Identification, Screening, and Included.
1. Identification Phase:
* Top-left box: Records identified from Databases (n = 1552), including PubMed (n = 317), Scopus (n = 464), W O S (n = 744), and P E D r o (n = 27).
* Top-right box: An arrow points from the identification box to Records removed before screening: Duplicate records removed (n = 651).
2. Screening Phase:
* Second box down (center-left): Records screened (n = 901).
* Second box down (right): An arrow points to Records excluded (n = 881).
* Third box down (center-left): Reports sought for retrieval (n = 20).
* Third box down (right): An arrow points to Reports not retrieved (n = 0).
* Fourth box down (center-left): Reports assessed for eligibility (n = 20).
* Fourth box down (right): An arrow points to Reports excluded: Not multimodal pain therapy (n = 2), Not only S F M (n = 2), Not pain catastrophizing (n = 1), and Control group M P T intervention (n = 2).
3. Included Phase:
* Bottom box: Studies included in review (n = 13).
Data extraction
Study characteristics, methods and results were extracted independently by C.G.-L. and I.C.-P.; any discrepancies were reviewed by C.M.G.-S. and E.O.-G. The following data were extracted: general study characteristics (authors, year of publication, country, blinding, setting and funding), sample characteristics (number of groups, number of participants per group and age), characteristics of MPT (type of intervention, number of sessions and number of sessions per week) and characteristics of the comparator (type of intervention, number of sessions and number of sessions per week). For the outcome data, the variable name, measurement instrument, post-intervention mean and standard deviation and assessment time point were recorded. Table 1 summarises the characteristics of the included studies.
Characteristics of the included studies
Table 1. Long description
The table consists of six columns and 13 study rows.
* Column 1: Study (author, year, country, design, blinding, funding).
* Column 2: Participants (N, age).
* Column 3: M P T group (N, intervention characteristics, gender ratio).
* Column 4: Control group (N, intervention characteristics, gender ratio).
* Column 5: Outcomes (Variable/test and time points).
* Column 6: Qualitative findings.
Key data points include:
* Arroyo-Fernández et al. (2022, Spain): 120 participants. M P T groups used active or sham t D C S plus T E. Both showed marked reduction in P C S scores (p < .001) compared to no-intervention.
* Araya-Quintanilla et al. (2024, Chile): 65 women. M P T (G M I plus P N E) showed P C S total scores decrease from 30.42 to 10.58 (p < .001).
* Bourgault et al. (2015, Canada): 56 patients. M P T (C B T plus T E plus P E) showed significant mean differences in P C S (p = .0007) at 3-month follow-up.
* Castel et al. (2012, 2013, 2015, Spain): Multiple studies showing C B T plus hypnosis or C B T plus T E significantly reduced catastrophizing (p < .001) compared to usual care.
* Ceballos-Laita et al. (2020, Spain): 32 women. P N E plus T E showed larger P C S decreases than T E alone.
* Gur-Kabul et al. (2026, Turkey): 25 women. T E plus T N E provided extra benefit over T E alone (p < 0.05).
* Pérez-Aranda et al. (2019, Spain): 225 patients. M B S R produced significantly greater P C S decreases than usual care (p < .001).
* Sanabria-Mazo et al. (2020, Spain): 41 women. M A I R group maintained significant improvement at follow-up (p = 0.002).
* Kircali et al. (2024, Turkey): 50 patients. P N E group had the lowest post-treatment P C S scores.
* Serrat et al. (2020, Spain): 169 patients. N A T-F M group showed significant P C S reductions (p < .001).
* Rasmussen et al. (2017, Denmark): 187 women. M P T showed reduction, but between-group difference was not statistically significant (p = 0.095).
Abbreviations: .FMS, fibromyalgia syndrome; MPT, multimodal pain therapy; RCT, randomized controlled trial; tDCS, transcranial direct current stimulation; TE, therapeutic exercise; GMI, graded motor imagery; PNE, pain neuroscience education; TNE, therapeutic neuroscience education; CBT, cognitive behavioral therapy; PE, psychoeducation; PCS, Pain Catastrophizing Scale; CSQ, Coping Strategies Questionnaire; MBSR, mindfulness-based stress reduction; TAU, treatment as usual; RT, relaxation therapy; MAIR, Mindfulness + Amygdala and Insula Retraining; MIEP, motor imagery-based exercise protocol; T0, baseline; T1, post-intervention; T2, 3-month follow-up.
The primary outcome was pain catastrophising, assessed using either the Pain Catastrophizing Scale (PCS; Sullivan et al., Reference Sullivan, Bishop and Pivik1995) or the pain catastrophising subscale of the Coping Strategies Questionnaire (CSQ; Rosenstiel and Keefe, Reference Rosenstiel and Keefe1983).
Assessment of methodological quality, risk of bias and certainty of evidence
Two reviewers (C.G.-L. and J.A.C.-R.) independently assessed the risk of bias in each study using the Cochrane Risk of Bias (RoB) tool. The tool evaluates seven domains: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other bias. Each domain was rated as having a low, high or unclear risk of bias. Disagreements were resolved through discussion with a third reviewer (I.C.-P.). The results are presented in Table 2 and Figures 2 and 3.
ROB tool
Table 2. Long description
The table evaluates thirteen studies across eight categories: Item 1 (Random sequence generation), Item 2 (Allocation concealment), Item 3 (Blinding of participants and personnel), Item 4 (Blinding of outcome assessment), Item 5 (Incomplete outcome data), Item 6 (Selective reporting), Item 7 (Other sources of bias), and an Overall rating.
* Araya-Quintanilla et al., 2024: Low for Items 1, 2, 5, 6, 7; High for Items 3, 4; Overall High.
* Arroyo-Fernández et al., 2022: Low for Items 1, 2, 4, 5, 6; High for Item 3; Unclear for Item 7; Overall High.
* Bourgault et al., 2015: Low for Items 1, 6; High for Item 3; Unclear for Items 2, 4, 5, 7; Overall High.
* Castel et al., 2012: Low for Items 1, 2, 4; High for Items 3, 5; Unclear for Items 6, 7; Overall High.
* Castel et al., 2013: Low for Items 1, 4, 6, 7; High for Items 3, 5; Unclear for Item 2; Overall High.
* Castel et al., 2015: Low for Items 1, 5, 6, 7; High for Items 3, 4; Unclear for Item 2; Overall High.
* Ceballos-Laita et al., 2020: Low for Items 1, 2, 6; High for Item 3; Unclear for Items 4, 5, 7; Overall High.
* Gur-Kabul et al., 2026: Low for Item 1; High for Item 3; Unclear for Items 2, 4, 5, 6, 7; Overall High.
* Kircali et al., 2024: Low for Item 1; High for Items 3, 4; Unclear for Items 2, 5, 6, 7; Overall High.
* Pérez-Aranda et al., 2019: Low for Items 1, 6; High for Items 3, 4; Unclear for Items 2, 5, 7; Overall High.
* Sanabria-Mazo et al., 2020: Low for Items 1, 2, 4, 5, 6, 7; High for Item 3; Overall High.
* Serrat et al., 2020: Low for Items 1, 5; High for Items 3, 4; Unclear for Items 2, 6, 7; Overall High.
* Rasmussen et al., 2017: Low for Items 1, 6; High for Items 3, 4; Unclear for Items 2, 5, 7; Overall High.
Abbreviations: Item 1, Random sequence generation; Item 2, Allocation concealment; Item 3, Blinding of participants and personnel; Item 4, Blinding of outcome assessment; Item 5, Incomplete outcome data; Item 6, Selective reporting; Item 7; Other sources of bias.
Risk-of-bias summary for the included studies.
Figure 2. Long description
A grid-based risk-of-bias summary table. The top header row lists domains D 1 through D 7 and an Overall column. The left-most column lists thirteen studies by author and year.
Legend for symbols:
* Green circle with a plus sign: Low risk.
* Yellow circle with a minus sign: Unclear risk.
* Red circle with an X: High risk.
Domain Definitions:
* D 1: Random sequence generation.
* D 2: Allocation concealment.
* D 3: Blinding of participants and personnel.
* D 4: Blinding of outcome assessment.
* D 5: Incomplete outcome data.
* D 6: Selective reporting.
* D 7: Other sources of bias.
Data breakdown by study (D 1 to D 7, then Overall):
* Araya-Quintanilla, F et al 2024: Low, Low, High, High, Low, Low, Low, Overall High.
* Arroyo-Fernández, R et al 2022: Low, Low, High, Low, Low, Low, Unclear, Overall High.
* Bourgault, P et al 2015: Low, Unclear, High, Unclear, Unclear, Low, Unclear, Overall High.
* Castel, A et al 2012: Low, Low, High, Low, High, Unclear, Unclear, Overall High.
* Castel, A et al 2013: Low, Unclear, High, Low, High, Low, Low, Overall High.
* Castel, A et al 2015: Low, Unclear, High, High, Low, Low, Low, Overall High.
* Ceballos-Laita, L et al 2020: Low, Low, High, Unclear, Unclear, Low, Unclear, Overall High.
* Gur-Kabul, E et al 2026: Low, Unclear, High, Unclear, Unclear, Unclear, Unclear, Overall High.
* Kircali, S et al 2024: Low, Unclear, High, High, Unclear, Unclear, Unclear, Overall High.
* Pérez-Aranda, A et al 2019: Low, Unclear, High, High, Unclear, Low, Unclear, Overall High.
* Sanabria-Mazo, J P et al 2020: Low, Low, High, Low, Low, Low, Low, Overall High.
* Serrat, M et al 2020: Low, Unclear, High, High, Low, Unclear, Unclear, Overall High.
* Rasmussen, M et al 2017: Low, Unclear, High, High, Unclear, Low, Unclear, Overall High.
Risk-of-bias graph showing review authors’ judgements for each domain across the included studies.
Figure 3. Long description
The chart features a horizontal X-axis representing percentages from 0 percent to 100 percent in increments of 25. The Y-axis lists eight categories from top to bottom.
1. Random sequence generation: 100 percent green (Low risk).
2. Allocation concealment: approximately 40 percent green (Low risk) and 60 percent yellow (Unclear risk).
3. Blinding of participants and personnel: 100 percent red (High risk).
4. Blinding of outcome assessment: approximately 30 percent green (Low risk), 25 percent yellow (Unclear risk), and 45 percent red (High risk).
5. Incomplete outcome data: approximately 40 percent green (Low risk), 45 percent yellow (Unclear risk), and 15 percent red (High risk).
6. Selective reporting: approximately 70 percent green (Low risk) and 30 percent yellow (Unclear risk).
7. Other sources of bias: approximately 30 percent green (Low risk) and 70 percent yellow (Unclear risk).
8. Overall: 100 percent red (High risk).
A legend at the bottom defines the colors: Blue for No information, Dark Red for Critical, Red for High, Yellow for Unclear, and Green for Low.
The certainty of evidence for each meta-analytic outcome was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework and the checklist developed by Meader et al. (Reference Meader, King, Llewellyn, Norman, Brown, Rodgers, Moe-Byrne, Higgins, Sowden and Stewart2014). Five domains were considered: risk of bias, inconsistency, indirectness, imprecision and publication bias. The certainty of evidence was downgraded by one level for each domain with serious concerns. The GRADE assessment is presented in Table 3.
The certainty of evidence (GRADE statement)
Table 3. Long description
The table contains 18 columns and 2 data rows.
Header columns include: Time, K, N, N s, S M D, C I 95 percent, p, Q (df), I super 2 (p), Egger p, Trim and Fill (sub-divided into S M D adjust and percent Var), Risk of Bias, Inc, Ind, Imp, Pub, and Level of evidence.
Row 1: Post intervention.
- K: 18
- N: 1,348
- N s: 74.89
- S M D: minus 0.67
- C I 95 percent: minus 0.95 to minus 0.38
- p: less than .001
- Q (df): 30.093 (17)
- I super 2 (p): 43.508 percent (p less than .05)
- Egger p: 0.304
- Trim and Fill: S M D adjust minus 0.95, 41 percent Var
- G R A D E factors: Risk of Bias is High, Inc is Yes, Ind is No, Imp is Possible, Pub is Yes
- Level of evidence: Low
Row 2: Three-month follow-up.
- K: 7
- N: 472
- N s: 67.43
- S M D: minus 0.957
- C I 95 percent: minus 1.543 to minus 0.370
- p: 0.001
- Q (df): 23.371 (6)
- I super 2 (p): 49.436 percent (0.062)
- Egger p: 0.131
- Trim and Fill: S M D adjust minus 1.24, 29 percent Var
- G R A D E factors: Risk of Bias is High, Inc is Yes, Ind is No, Imp is Yes, Pub is Yes
- Level of evidence: Very low
Abbreviations: K, number of studies; N, number of participants; Ns, number of participants per comparison; SMD, standardised mean difference; 95% CI, confidence interval; p, p-value; Q (df), Cochran’s Q statistic (degrees of freedom); I 2, I 2 heterogeneity statistic; SMD adjust, standardised mean difference adjusted; %Var, variation percentage; ROB, risk of bias; Inc, inconsistency; Ind, indirectness; Imp, imprecision; Pub, publication bias.
Statistical analysis
The meta-analyses were conducted by two authors (C.G.-L. and E.O.-G.) using Comprehensive Meta-Analysis, version 4 (Biostat, Englewood, NJ, USA) (Borenstein et al., Reference Borenstein, Hedges, Higgins and Rothstein2023). A meta-analysis was performed only when at least two comparisons were available (Higgins and Green, Reference Higgins and Green2011). For continuous outcomes, pooled effects were expressed as Cohen’s standardised mean differences (SMDs) with 95% confidence intervals (CIs) and estimated using a random-effects model (Cohen, Reference Cohen1977; DerSimonian and Laird, Reference DerSimonian and Laird1986). The absolute magnitude of the SMD was interpreted using Cohen’s criteria: small (~0.20), medium (~0.50) and large (≥0.80) (Cohen, Reference Cohen1988). Potential publication bias and small-study effects were examined using funnel plots, Egger’s test and the trim-and-fill method (Egger et al., Reference Egger, Smith, Schneider and Minder1997; Duval and Tweedie, Reference Duval and Tweedie2000; Sterne and Egger, Reference Sterne and Egger2001). These analyses were interpreted cautiously when a meta-analysis included 10 or fewer studies. The trim-and-fill method was used to estimate an adjusted pooled effect. If the absolute relative difference between the unadjusted and adjusted pooled estimates exceeded 10%, the certainty of evidence was downgraded by one level (Rothman et al., Reference Rothman, Greenland and Lash2008). Statistical heterogeneity was evaluated using Cochran’s Q statistic and I 2 (Higgins et al., Reference Higgins, Thompson and Altman2002). I 2 values were interpreted as follows: 0–40%, heterogeneity might not be important; >40–60%, moderate heterogeneity; >60–90%, substantial heterogeneity; and > 90–100%, considerable heterogeneity. These categories were used as mutually exclusive thresholds to facilitate the interpretation of heterogeneity. The magnitude and direction of the pooled effects and the p-value of Cochran’s Q test were also considered when assessing heterogeneity (Higgins et al., Reference Higgins, Thompson, Deeks and Altman2003). For multi-arm studies with more than one eligible intervention group and a shared comparator, the comparator sample was divided proportionally across the relevant comparisons to avoid double-counting participants. The comparator means and standard deviations remained unchanged; accordingly, no within-study correlation coefficient was assumed.
Results
The results are organised into four sections. First, the study selection process is described. Second, the characteristics of the included RCTs and their participants are summarised. Third, methodological quality and risk of bias are appraised. Finally, the quantitative synthesis presents the post-intervention and 3-month follow-up effects of MPT on pain catastrophising, together with subgroup analyses by intervention duration.
Study selection
A total of 1,552 records were identified through PubMed/MEDLINE (n = 317), Scopus (n = 464), Web of Science (n = 744) and PEDro (n = 27). After duplicates were removed, 901 records were screened by title and abstract, of which 881 were excluded as irrelevant. Seven full-text reports were excluded because they did not meet the eligibility criteria (Jensen et al., Reference Jensen, Turner and Romano2007; van Wilgen et al., Reference van Wilgen, Bloten and Oeseburg2007; García-Campayo et al., Reference García-Campayo, Serrano-Blanco, Rodero, Magallón, Alda, Andrés, Luciano and del Hoyo2009; Dowd et al., Reference Dowd, Hogan, McGuire, Davis, Sarma, Fish and Zautra2015; Lami et al., Reference Lami, Martínez, Miró, Sánchez, Prados, Cáliz and Vlaeyen2018b; Ughreja et al., Reference Ughreja, Venkatesan, Gopalakrishna, Singh and Lakshmi2024; Norouzi et al., Reference Norouzi, Pournazari, Joybari, Sufivand, Asar, Bratty and Khazaie2025); the reasons are provided in Figure 1. Thirteen studies were included (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020); Arroyo-Fernández et al., Reference Arroyo-Fernández, Avendaño-Coy, Velasco-Velasco, Palomo-Carrión, Bravo-Esteban and Ferri-Morales2022; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026). The selection process is summarised in the PRISMA flow diagram (Figure 1).
Characteristics of the included studies
The included RCTs were conducted between 2012 and 2026 in Spain (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Arroyo-Fernández et al., Reference Arroyo-Fernández, Avendaño-Coy, Velasco-Velasco, Palomo-Carrión, Bravo-Esteban and Ferri-Morales2022), Chile (Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024), Canada (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015), Turkey (Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026) and Denmark (Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017). Together, they enrolled 1,314 individuals with FMS (mean age = 49.97 ± 8.18 years). The MPT and comparator groups included 738 and 576 participants, respectively. Comparator conditions included therapeutic exercise, pain neuroscience education, cognitive-behavioural therapy, relaxation therapy, wait-list control and no intervention. Interventions were grouped as short duration (1–10 sessions) (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020), medium duration (11–20 sessions) (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Arroyo-Fernández et al., Reference Arroyo-Fernández, Avendaño-Coy, Velasco-Velasco, Palomo-Carrión, Bravo-Esteban and Ferri-Morales2022) or long duration (≥21 sessions) (Castel et al., Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026). All outcomes were assessed immediately after the intervention. Six studies provided 3-month follow-up data (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020). Ten studies reported external funding (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Castel et al., Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026). Table 1 summarises the study characteristics.
Methodological quality and risk of bias of the included studies
Overall, the included RCTs had several methodological limitations and risks of bias. Although random sequence generation was adequately described across trials, allocation concealment was insufficiently reported or unclear in most studies, resulting in concerns about selection bias (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Castel et al., Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026). Given the nature of the interventions, blinding participants and personnel was not feasible in any study, resulting in performance bias across the included RCTs. Blinding of outcome assessors was not implemented, was insufficiently reported or was ineffective for self-reported outcomes, contributing to detection bias in nearly all studies (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024). Incomplete outcome data were also a concern in several trials because attrition was high or the handling of missing data was inadequately described. Common author-reported limitations included short follow-up periods, which prevented assessment of long-term effects (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026), and small samples or pilot designs, which limited generalisability and statistical power. Table 2 and Figures 2 and 3 summarise the methodological quality and risk-of-bias assessments.
Quantitative findings
Post-intervention
Thirteen studies contributing 18 comparisons (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Arroyo-Fernández et al., Reference Arroyo-Fernández, Avendaño-Coy, Velasco-Velasco, Palomo-Carrión, Bravo-Esteban and Ferri-Morales2022; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026), comprising 1,314 participants, were included in the post-intervention meta-analysis. Pain catastrophising was assessed with the PCS in nine studies (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Arroyo-Fernández et al., Reference Arroyo-Fernández, Avendaño-Coy, Velasco-Velasco, Palomo-Carrión, Bravo-Esteban and Ferri-Morales2022; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026) and with the pain catastrophising subscale of the CSQ in four studies (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017). Low-certainty evidence indicated a medium pooled effect in favour of MPT (SMD = −0.668; 95% CI: −0.953 to −0.383; p < .001). Heterogeneity was moderate (I 2 = 43.508%; Q = 30.093; df = 17; p = .026). Adjustment for potential publication bias and small-study effects using the trim-and-fill method yielded an SMD of −0.95 (95% CI: −1.24 to −0.65), representing a 42.2% increase in the magnitude of the pooled effect compared with the unadjusted estimate (Figure 4).
Forest plot of pain catastrophising at post-intervention.
Note: Short-duration interventions comprised 1–10 sessions, medium-duration interventions comprised 11–20 sessions and long-duration interventions comprised ≥21 sessions. Negative standardised mean differences favour multicomponent pain therapy. CI, confidence interval; MPT, multicomponent pain therapy; SMD, standardised mean difference.
Figure 4. Long description
A forest plot table with 18 rows of study data and a summary diamond.
Columns from left to right include.
1. Study name. Lists authors from Araya-Quintanilla (2024) to Rasmussen (2017).
2. Statistics for each study. Includes Std diff in means, Standard error, Variance, Lower limit, Upper limit, Z-Value, and p-Value.
3. Std diff in means and 95% C I. A graphical plot where the x-axis ranges from -1.00 to 1.00. A vertical line of null effect is at 0.00. Points to the left favor M P T and points to the right favor Control.
4. Relative weight. Percentages ranging from 4.25 to 6.53.
Key Data Points.
* Individual study S M D values range from -3.889 (Castel et al 2012 b) to -0.052 (Ceballos-Laita et al 2020).
* The graphical section shows black squares representing individual study effect sizes with horizontal bars for 95% C I. Several studies have C I bars that cross the zero line, while others (like Castel et al 2012 a and b) are entirely to the left.
* At the bottom, a red summary diamond is centered at an S M D of -0.668 with a 95% C I from -0.953 to -0.383. The p-Value for the overall effect is 0.000, indicating a statistically significant result favoring M P T.
Three-month follow-up
Six studies contributing seven comparisons (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012, Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020), comprising 472 participants, were included in the 3-month follow-up meta-analysis. Pain catastrophising was assessed using the PCS or the pain catastrophising subscale of the CSQ. Very-low-certainty evidence indicated a large pooled effect in favour of MPT (SMD = −0.957; 95% CI: −1.543 to −0.370; p = .001). The manuscript reported moderate heterogeneity (I 2 = 49.436%; Q = 23.371; df = 6; p = .062). Adjustment for potential publication bias and small-study effects using the trim-and-fill method yielded an SMD of −1.24 (95% CI: −1.83 to −0.64), which was more negative than the unadjusted estimate (Figure 5).
Forest plot of pain catastrophising at the 3-month follow-up.
Figure 5. Long description
The forest plot is organized into three main sections from left to right.
1. Data Table: Lists seven studies: Bourgault, P et al 2015; Castel, A et al 2012 a; Castel, A et al 2012 b; Castel, A et al 2013; Castel, A et al 2015; Ceballos-Laita, L et al 2020; and Sanabria-Mazo, J P et al 2020. Columns include Std diff in means, Standard error, Variance, Lower limit, Upper limit, Z-Value, and p-Value. A final row provides the pooled effect size of -0.957 with a p-Value of 0.001.
2. Graphical Plot: The x-axis represents the Std diff in means and 95% C I, ranging from -1.00 to 1.00. A vertical line of no effect is at 0.00. Values to the left are labeled Favours M P T and values to the right are labeled Favours Control. Each study is represented by a black square and horizontal error bars. Most studies favor M P T, with Castel 2012 b showing the strongest effect beyond the left axis limit. A red diamond at the bottom represents the overall pooled effect, centered at -0.957, entirely to the left of the zero line.
3. Relative Weight: The far-right column lists the percentage weight of each study, ranging from 12.51 to 15.79.
Subgroup analysis
A subgroup analysis was conducted to explore whether the effect of MPT on pain catastrophising varied by intervention duration. Three predefined subgroups were analysed: short duration (1–10 sessions) (Bourgault et al., Reference Bourgault, Lacasse, Marchand, Courtemanche-Harel, Charest, Gaumond, De Souza and Choinière2015; Pérez-Aranda et al., Reference Pérez-Aranda, Feliu-Soler, Montero-Marin, García-Campayo, Andrés-Rodríguez, Borràs, Rozadilla-Sacanell, Peñarrubia-María, Angarita-Osorio, McCracken and Luciano2019; Ceballos-Laita et al., Reference Ceballos-Laita, Mingo-Gómez, Navas-Cámara, Estébanez-de-Miguel, Caudevilla-Polo, Verde-Rello, Fernández-Araque and Jiménez-del-Barrio2020), medium duration (11–20 sessions) (Castel et al., Reference Castel, Cascón, Padrol, Sala and Rull2012; Rasmussen et al., Reference Rasmussen, Amris, Rydahl-Hansen, Danneskiold-Samsoe, Mortensen, Christensen and Sjölund2017; Sanabria-Mazo et al., Reference Sanabria-Mazo, Montero-Marin, Feliu-Soler, Gasión, Navarro-Gil, Morillo-Sarto, Colomer-Carbonell, Borràs, Tops, Luciano and García-Campayo2020; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Arroyo-Fernández et al., Reference Arroyo-Fernández, Avendaño-Coy, Velasco-Velasco, Palomo-Carrión, Bravo-Esteban and Ferri-Morales2022) and long duration (≥21 sessions) (Castel et al., Reference Castel, Fontova, Montull, Periñán, Poveda, Miralles, Cascón-Pereira, Hernández, Aragonés, Salvat, Castro, Monterde, Padrol, Sala, Añez and Rull2013, Reference Castel, Castro, Fontova, Poveda, Cascón-Pereira, Montull, Padrol, Qanneta and Rull2015; Araya-Quintanilla et al., Reference Araya-Quintanilla, Álvarez-Bueno, Cavero-Redondo, Ramírez-Vélez, Sepúlveda-Loyola and Gutiérrez-Espinoza2024; Kircali et al., Reference Kircali, Özcan and Karahan2024; Gür Kabul et al., Reference Gür Kabul, Başakcı Çalık, Baş Aslan and Karasu2026).
The pooled estimate for the short-duration subgroup was not statistically significant (SMD = −0.226; 95% CI: −0.491 to 0.039; p = .095; I 2 = 0%; Q = 2.551; df = 3; p = .466). The medium-duration subgroup yielded a large, statistically significant estimate (SMD = −1.023; 95% CI: −1.661 to −0.385; p = .002), with substantial heterogeneity (I 2 = 54.841%; Q = 13.287; df = 6; p = .039). The long-duration subgroup yielded a medium estimate (SMD = −0.754; 95% CI: −0.945 to −0.562; p < .001), with reported heterogeneity values of I 2 = 0%, Q = 5.579, df = 3 and p = .472. The overall heterogeneity statistic was Q = 33.217 (df = 17; p = .011; I 2 = 48.821%). Although the point estimate was largest in the medium-duration subgroup, the test for subgroup differences was not statistically significant, indicating that the effect of MPT did not differ significantly by intervention duration. The GRADE assessment is presented in Table 3.
Discussion
Pain catastrophising is a clinically meaningful cognitive-affective process in FMS because it is associated with poorer adjustment and quality of life (Lami et al., Reference Lami, Martínez, Miró, Sánchez and Guzmán2018a). This review synthesised 13 RCTs comprising 1,314 participants. At post-intervention, low-certainty evidence yielded a medium pooled estimate in favour of MPT; at the 3-month follow-up, very-low-certainty evidence yielded a large pooled estimate. To our knowledge, this is the first FMS-specific meta-analysis focused on the effect of MPT on pain catastrophising. The findings should be interpreted in light of the risk of bias, heterogeneity among interventions and potential publication bias and small-study effects.
Medium-duration interventions (11–20 sessions) yielded the largest subgroup point estimate. However, this finding is exploratory: the test for differences between subgroups was not reported, heterogeneity within the medium-duration subgroup was moderate and the number and characteristics of studies differed across duration categories. Consequently, the results do not establish an optimal therapeutic dose. Direct comparisons and prospectively designed dose–response studies are needed to determine whether intervention duration modifies the effect of MPT (Loftus et al., Reference Loftus, Dobbin and Crampton2022; Niu et al., Reference Niu, Zheng, Deng, Yang and Du2024).
The mechanisms underlying the association between MPT and reduced pain catastrophising are likely multifactorial. Because MPT combines education, exercise, psychological interventions and sometimes physical or manual therapies, the pooled effect cannot be attributed to a single component. Plausible treatment targets include threat appraisal, pain-related hypervigilance, fear avoidance, maladaptive emotional reactivity, low perceived control and pain self-efficacy (Quartana et al., Reference Quartana, Campbell and Edwards2009; Petrini and Arendt-Nielsen, Reference Petrini and Arendt-Nielsen2020; Schumann et al., Reference Schumann, Coombes, Gascho, Geske, McDermott, Morrison, Reynolds, Bernau and Gilliam2022).
At the 3-month follow-up, the pooled estimate remained in favour of MPT. Nevertheless, only six studies contributed follow-up data, the confidence interval was wide, the certainty of evidence was very low and the reported heterogeneity statistics require verification. These data therefore do not establish that the effect was maintained or increased over time, nor do they establish clinical significance.
A possible explanation for a persisting effect is that cognitive and behavioural skills promoted by MPT, such as cognitive reframing, emotion regulation, activity pacing and adaptive coping, may consolidate through continued practice after treatment (Carson et al., Reference Carson, Keefe, Lynch, Carson, Goli, Fras and Thorp2006; Kashikar-Zuck et al., Reference Kashikar-Zuck, Sil, Lynch-Jordan, Ting, Peugh, Schikler, Hashkes, Arnold, Passo, Richards and Lovell2013; Nicholas et al., Reference Nicholas, Asghari and Corbett2013; Ehde et al., Reference Ehde, Dillworth and Turner2014; Martinez-Calderon et al., Reference Martinez-Calderon, Zamora-Campos, Navarro-Ledesma and Luque-Suarez2018; Aaron et al., Reference Aaron, Mun, McGill, Finan and Campbell2022). However, this hypothesis was not tested in the included trials and should be examined using longitudinal mediation analyses.
The fear-avoidance model and cognitive models of catastrophising provide one explanatory framework (Vlaeyen and Linton, Reference Vlaeyen and Linton2000; Sullivan et al., Reference Sullivan, Thorn, Haythornthwaite, Keefe, Martin, Bradley and Lefebvre2001; Vlaeyen et al., Reference Vlaeyen, Crombez and Linton2016). Psychoeducation, graded activity, behavioural exposure and adaptive coping may reduce threat appraisals, fear-based monitoring of bodily sensations and avoidance. In parallel, MPT may weaken attentional bias towards pain-related threat and promote more flexible interpretations of bodily sensations.
Self-efficacy and emotion regulation provide complementary pathways. Repeated mastery experiences and corrective learning may strengthen perceived control over symptoms, whereas mindfulness, acceptance-based strategies, relaxation and regulated body awareness may reduce emotional reactivity and perseverative thinking (Bandura, Reference Bandura1977; Jackson et al., Reference Jackson, Wang, Wang and Fan2014; de Jong et al., Reference de Jong, Lazar, Hug, Mehling, Hölzel, Sack, Peeters, Ashih, Mischoulon and Gard2016; Veehof et al., Reference Veehof, Trompetter, Bohlmeijer and Schreurs2016; Cheng et al., Reference Cheng, Leung, Chan, Chen, Chow, Chung, Law, Lee, Leung and Tam2018; Bravo et al., Reference Bravo, Skjaerven, Guitard Sein-Echaluce and Catalan-Matamoros2019; Haugmark et al., Reference Haugmark, Hagen, Smedslund and Zangi2019; Aaron et al., Reference Aaron, Sturgeon and Finan2020, Reference Aaron, Mun, McGill, Finan and Campbell2022; Serrat et al., Reference Serrat, Almirall, Musté, Sanabria-Mazo, Feliu-Soler, Méndez-Ulrich, Luciano and Sanz2020; Vambheim et al., Reference Vambheim, Kyllo, Hegland and Bystad2021; Voss et al., Reference Voss, Boachie, Nieves and Gothe2023). These pathways are theoretically plausible but were not directly compared in the present meta-analysis.
Process-oriented research provides indirect support for these interpretations. Reductions in pain catastrophising and increases in pain self-efficacy have mediated outcomes after interdisciplinary pain rehabilitation (Jensen et al., Reference Jensen, Turner, Romano and Karoly2001; Spinhoven et al., Reference Spinhoven, ter Kuile, Kole-Snijders, Hutten Mansfeld, den Ouden and Vlaeyen2004; Schumann et al., Reference Schumann, Coombes, Gascho, Geske, McDermott, Morrison, Reynolds, Bernau and Gilliam2022) and changes in catastrophising, fear-avoidance beliefs and self-efficacy have mediated associations between pain intensity and disability in chronic low back pain treatment (Ryum et al., Reference Ryum, Stiles, Svendsen, Wigemyr and Abrahamsen202 Reference Amir-Behghadami and Janati4). Because these findings were not derived exclusively from FMS samples or from the trials included here, they do not establish the mechanisms of MPT in FMS.
Future mechanistic studies should use longitudinal and mediation-based designs to identify which MPT components are associated with changes in catastrophising, for whom these pathways are most relevant and when during treatment they become active. Such work could clarify the relative contribution of attentional, cognitive, emotional, interpersonal and self-regulatory processes and inform more targeted interventions.
Clinically, the findings support considering pain catastrophising as one treatment-related outcome within integrated FMS care. Assessment with validated instruments such as the PCS or CSQ may inform treatment planning and progress monitoring, and elevated catastrophising may indicate a need for psychologically informed or interdisciplinary support (Martinez-Calderon et al., Reference Martinez-Calderon, Flores-Cortes, Morales-Asencio, Fernandez-Sanchez and Luque-Suarez2020a, Reference Martinez-Calderon, Flores-Cortes, Morales-Asencio and Luque-Suarez2020b; Campos et al., Reference Campos, Vázquez and Vilhena2024). However, the present evidence does not establish that MPT should replace effective pharmacological or single-component options.
Intervention duration should likewise not be prescribed on the basis of subgroup point estimates alone. When MPT is used, its components and intensity should be individualised according to clinical presentation, patient preferences, feasibility and response, consistent with recommendations for integrated and person-centred FMS management (Macfarlane et al., Reference Macfarlane, Kronisch, Dean, Atzeni, Häuser, Fluß, Choy, Kosek, Amris, Branco, Dincer, Leino-Arjas, Longley, McCarthy, Makri, Perrot, Sarzi-Puttini, Taylor and Jones2017; Sarzi-Puttini et al., Reference Sarzi-Puttini, Giorgi, Marotto and Atzeni2020; Cohen et al., Reference Cohen, Vase and Hooten2021; Nijs et al., Reference Nijs, Malfliet, Roose, Lahousse, Van Bogaert, Johansson, Runge, Goossens, Labie, Bilterys, Van Campenhout, Polli, Wyns, Hendrix, Xiong, Ahmed, De Baets and Huysmans2024).
Several limitations must be acknowledged. The included studies varied substantially in MPT components, intervention dose, comparator conditions and methodological quality, which may have contributed to variation in effect estimates. Pain catastrophising was not the primary outcome in many trials, and follow-up data were scarce. The certainty of evidence was low at post-intervention and very low at the three-month follow-up. Funnel plots and trim-and-fill analyses, particularly when based on few studies, are imprecise and do not by themselves demonstrate publication bias or identify the true effect. The adjusted estimates should therefore be interpreted cautiously rather than as evidence that the unadjusted effects necessarily underestimated the benefit of MPT.
Future research should use standardised, well-described MPT protocols, prespecify pain catastrophising as an outcome, report clinically interpretable change and include longer follow-up periods. Adequately powered RCTs should also improve allocation concealment, use blinded outcome assessment where feasible and investigate active components and mechanisms. Further high-quality trials may substantially change the magnitude and stability of the pooled estimates. The duration subgroup findings should remain exploratory until supported by a valid test of subgroup differences and direct dose–response evidence.
Conclusion
This meta-analysis suggests that MPT may reduce pain catastrophising in individuals with FMS. The pooled estimate was medium at post-intervention and large at the 3-month follow-up; however, the certainty of evidence was low and very low, respectively. Medium-duration interventions yielded the largest subgroup point estimate, but this exploratory finding does not establish an optimal treatment duration. MPT may be considered within individualised, biopsychosocial care, while more rigorous and adequately powered RCTs with standardised protocols, longer follow-up and transparent reporting are needed to strengthen the evidence.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2026.10308.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/gmh.2026.10308.
Author contributions
Conceptualisation: C.G.-L. and C.M.G.-S. Data curation: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Formal analysis: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Investigation: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Methodology: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Project administration: C.M.G.-S. Resources: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Supervision: C.M.G.-S. Validation: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Visualisation: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Writing – original draft: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. Writing – review and editing: C.G.-L., J.A.C.-R., I.C.-P., E.O.-G. and C.M.G.-S. All authors read and approved the final version of the manuscript.
Competing interests
The authors declare no conflicts of interest.
Data availability statement
The data supporting the findings of this study are available from the corresponding author upon reasonable request.
Ethics and approvals
Ethical approval was not required for this systematic review and meta-analysis. Patient consent statement is not applicable or needed for this study.
Funding statement
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Open access
